| cloudnull | from setup-infra? | 00:00 |
|---|---|---|
| hughmFLEXin | yea | 00:00 |
| cloudnull | I know that bug, but i thought that was fixed. | 00:03 |
| cloudnull | -cc stevelle | 00:03 |
| *** thorst_ has joined #openstack-ansible | 00:04 | |
| cloudnull | hughmFLEXin: you mind doing a git branch ? | 00:04 |
| cloudnull | are you on (detached from 14.0.0.0b3) ? | 00:05 |
| hughmFLEXin | hmmm apparently im on no branch | 00:05 |
| cloudnull | the file to modify is playbooks/inventory/group_vars/utility_all.yml:44: | 00:06 |
| cloudnull | and then remove the gnocchiclient entry | 00:06 |
| cloudnull | on line 44 that is | 00:06 |
| cloudnull | but in "master" that was fixed | 00:06 |
| cloudnull | which is only slightly ahead from that tag | 00:06 |
| hughmFLEXin | hmm ok removed it | 00:07 |
| cloudnull | you can just run ``openstack-ansible utility-install.yml`` and you'll be done with the setup infra part | 00:07 |
| *** woodard has quit IRC | 00:07 | |
| * hughmFLEXin dances | 00:08 | |
| cloudnull | stevelle: Matias: if you get a chance would you mind reviewing https://review.openstack.org/#/c/352054/ | 00:10 |
| cloudnull | once that's in we can update the ceilo role to allow for list overrides in nested yaml files | 00:10 |
| cloudnull | which would finally give us the ability to change things in those nested lists of lists.. | 00:11 |
| *** thorst_ has quit IRC | 00:11 | |
| * cloudnull shakes fist at ceilometer. | 00:11 | |
| hughmFLEXin | aw yisss, on to the final playbook | 00:12 |
| stevelle | hughmFLEXin: can you taste it? does it taste like bitter? | 00:12 |
| hughmFLEXin | no just taste like the usual openstack | 00:12 |
| cloudnull | the bitter bite of OpenStack | 00:12 |
| hughmFLEXin | exactly ^ lmao | 00:12 |
| hughmFLEXin | need to install mosh so i can run it and go home | 00:13 |
| cloudnull | tmux | 00:14 |
| cloudnull | for the win | 00:14 |
| *** jcannava has quit IRC | 00:14 | |
| hughmFLEXin | same thing :-P | 00:14 |
| cloudnull | I like mosh though | 00:16 |
| cloudnull | its kinda amazing | 00:16 |
| hughmFLEXin | MIT fellas are pretty smart eh | 00:16 |
| stevelle | cloudnull: the os_glance thing may just be an ansible problem | 00:16 |
| stevelle | trying out a fix | 00:16 |
| cloudnull | :( | 00:16 |
| cloudnull | :) | 00:16 |
| *** sdake_ has quit IRC | 00:17 | |
| *** sdake has joined #openstack-ansible | 00:17 | |
| *** psilvad has joined #openstack-ansible | 00:20 | |
| hughmFLEXin | so if physically, my eth3 is the external (outbound/public) , but i've shoved all the OSA network vlan stuff on eth2 | 00:20 |
| hughmFLEXin | is there any special ip forwarding / natting i need to do afterwards? | 00:20 |
| hughmFLEXin | or will the external_vip thing handle it (i assigned it a public ip) | 00:20 |
| cloudnull | that should handle it | 00:21 |
| hughmFLEXin | sweet | 00:21 |
| cloudnull | the handoff is in haproxy and we use multiple backends to get it done. | 00:21 |
| cloudnull | so "it should just work"-tm | 00:21 |
| hughmFLEXin | *waits for the ymmv | 00:22 |
| * hughmFLEXin waits for the ymmv | 00:22 | |
| hughmFLEXin | :-( | 00:22 |
| *** schwicht has joined #openstack-ansible | 00:25 | |
| cloudnull | its less ymmv and more https://www.youtube.com/watch?v=StTqXEQ2l-Y | 00:27 |
| *** adrian_otto has quit IRC | 00:27 | |
| hughmFLEXin | lmao | 00:29 |
| hughmFLEXin | alright ill catch you tmrw and let you know | 00:30 |
| openstackgerrit | Steve Lewis (stevelle) proposed openstack/openstack-ansible-os_glance: Fix os_glance role functional testing https://review.openstack.org/369726 | 00:30 |
| cloudnull | hughmFLEXin: cheers ! | 00:30 |
| *** hughmFLEXin has quit IRC | 00:31 | |
| stevelle | digging into list extensions now... | 00:31 |
| *** woodard has joined #openstack-ansible | 00:33 | |
| *** woodard has quit IRC | 00:34 | |
| *** woodard has joined #openstack-ansible | 00:34 | |
| *** nwonknu has quit IRC | 00:38 | |
| *** itsuugo has quit IRC | 00:40 | |
| *** itsuugo has joined #openstack-ansible | 00:40 | |
| *** smatzek has joined #openstack-ansible | 00:44 | |
| cloudnull | finally relocating home but like for real this time . | 00:44 |
| cloudnull | :) bbl | 00:44 |
| *** oben has joined #openstack-ansible | 00:45 | |
| *** itsuugo has quit IRC | 00:45 | |
| *** itsuugo has joined #openstack-ansible | 00:46 | |
| *** schwicht has quit IRC | 00:47 | |
| *** itsuugo has quit IRC | 00:51 | |
| *** schwicht has joined #openstack-ansible | 00:51 | |
| *** nwonknu has joined #openstack-ansible | 00:52 | |
| *** itsuugo has joined #openstack-ansible | 00:53 | |
| *** asettle has joined #openstack-ansible | 00:59 | |
| *** asettle has quit IRC | 01:04 | |
| *** schwicht has quit IRC | 01:04 | |
| *** smatzek has quit IRC | 01:07 | |
| *** thorst_ has joined #openstack-ansible | 01:09 | |
| *** itsuugo has quit IRC | 01:10 | |
| *** itsuugo has joined #openstack-ansible | 01:11 | |
| *** thorst_ has quit IRC | 01:12 | |
| *** thorst_ has joined #openstack-ansible | 01:14 | |
| *** gouthamr has quit IRC | 01:17 | |
| *** albertcard1 has joined #openstack-ansible | 01:23 | |
| *** poopcat has quit IRC | 01:23 | |
| *** albertcard1 has quit IRC | 01:23 | |
| *** poopcat has joined #openstack-ansible | 01:25 | |
| Matias | stevelle: new issue: http://paste.openstack.org/show/575585/ | 01:25 |
| Matias | 2 pipeline queues are getting stuck with this | 01:26 |
| *** itsuugo has quit IRC | 01:26 | |
| *** jamielennox is now known as jamielennox|away | 01:26 | |
| *** itsuugo has joined #openstack-ansible | 01:27 | |
| Matias | stevelle: i'm surprised by SSLV3_ALERT_BAD_RECORD_MAC | 01:27 |
| Matias | sslv3??? | 01:27 |
| stevelle | Matias: all of the stuff on master *except* ceilometermiddleware should be using some kind of SSL for rabbit, and I think the default protocol is SSLv3 :(( | 01:30 |
| Matias | sslv3 O_o | 01:31 |
| stevelle | I remember seeing something ugly when I was digging... | 01:33 |
| stevelle | https://github.com/openstack/oslo.messaging/blob/master/oslo_messaging/_drivers/impl_rabbit.py#L640 | 01:33 |
| stevelle | Matias: ^^ | 01:33 |
| *** poopcat has quit IRC | 01:33 | |
| *** poopcat has joined #openstack-ansible | 01:34 | |
| stevelle | not updated since 2014 | 01:34 |
| *** schwicht has joined #openstack-ansible | 01:34 | |
| Matias | well, what surprises me is that sslv3 is chosen by default by.. kombu maybe? | 01:35 |
| stevelle | maybe | 01:36 |
| Matias | oslo.messaging itself does not choose a default.. | 01:36 |
| stevelle | looks like we can override by specifying kombu_ssl_version = tlsv1_2 | 01:36 |
| Matias | https://github.com/celery/kombu/search?utf8=%E2%9C%93&q=ssl_version | 01:37 |
| Matias | maybe it just passes as-is to ampq | 01:37 |
| *** schwicht has quit IRC | 01:37 | |
| stevelle | under the [oslo_messaging_rabbit] section | 01:38 |
| stevelle | Matias: wondering if there is an underlying network issue causing that issue | 01:39 |
| *** itsuugo has quit IRC | 01:39 | |
| Matias | no, it is passed to https://docs.python.org/2/library/ssl.html#ssl.wrap_socket | 01:39 |
| Matias | looks like python itself is choosing sslv3 as default O_o | 01:39 |
| Matias | maybe we should change from ubuntu to openbsd | 01:40 |
| stevelle | or even py3 | 01:40 |
| stevelle | as radical as that idea is :P | 01:41 |
| *** itsuugo has joined #openstack-ansible | 01:41 | |
| stevelle | odyssey4me: ^ see above about specifying [oslo_messaging_rabbit] kombu_ssl_version = tlsv1_2 to get around use of SSLv3 | 01:42 |
| stevelle | will let him pipe in on that tomorrow | 01:42 |
| *** sdake_ has joined #openstack-ansible | 01:45 | |
| *** sdake_ has quit IRC | 01:45 | |
| *** sdake_ has joined #openstack-ansible | 01:45 | |
| cloudnull | https://review.openstack.org/#/c/369726/ -- is that all that's needed to make glance happy ? | 01:47 |
| *** woodard has quit IRC | 01:47 | |
| stevelle | cloudnull: yeah seems like | 01:47 |
| *** sdake has quit IRC | 01:47 | |
| stevelle | just needed to rub some ansible on it | 01:47 |
| cloudnull | ha | 01:48 |
| *** itsuugo has quit IRC | 01:52 | |
| *** thorst_ has quit IRC | 01:52 | |
| *** thorst_ has joined #openstack-ansible | 01:53 | |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible: Ensure that gnocchi uses keystone authentication https://review.openstack.org/369610 | 01:54 |
| *** itsuugo has joined #openstack-ansible | 01:54 | |
| *** thorst_ has quit IRC | 02:02 | |
| openstackgerrit | Merged openstack/openstack-ansible-security: Add network conf auditing on CentOS https://review.openstack.org/368991 | 02:02 |
| Matias | cloudnull: already moved? :D | 02:03 |
| cloudnull | yup, i'm quick like that :P | 02:03 |
| Matias | cloudnull: so do you think it is better to created systemd files only if the system really uses systemd? | 02:04 |
| *** dragonmaster has joined #openstack-ansible | 02:04 | |
| cloudnull | i do | 02:05 |
| cloudnull | if thats possible | 02:05 |
| stevelle | +1, but I'm rollin out for the night | 02:05 |
| Matias | stevelle: good evening | 02:06 |
| cloudnull | have a good one stevelle | 02:06 |
| Matias | good night | 02:06 |
| Matias | good sleep? | 02:06 |
| stevelle | only 7pm, going out for a couple hours before sleep | 02:06 |
| Matias | repo-install now takes an eternity (unless you are on a fresh install) | 02:06 |
| Matias | 23h here | 02:07 |
| *** weezS has joined #openstack-ansible | 02:07 | |
| cloudnull | repo-install now takes an eternity (unless you are on a fresh install)? | 02:12 |
| cloudnull | Matias: ^ | 02:12 |
| cloudnull | what is it taking too long on? | 02:12 |
| Matias | just testing your PR once more | 02:14 |
| Matias | recusing into the fully populated git repos is really slow | 02:14 |
| Matias | recursing* | 02:14 |
| Matias | but there is nothing we can do | 02:15 |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible-repo_server: Convert role testing to use Ansible 2.1.1 https://review.openstack.org/369782 | 02:19 |
| cloudnull | I wonder if its the ansible task being slow with recurse | 02:19 |
| cloudnull | or if its the bash task ? | 02:19 |
| cloudnull | find should be super quick | 02:19 |
| cloudnull | but I guess either could be terrible | 02:19 |
| *** thorst_ has joined #openstack-ansible | 02:20 | |
| *** thorst_ has quit IRC | 02:21 | |
| Matias | argh, ran repo-install accidentally instead of repo-server | 02:21 |
| Matias | let me run it again to get the stats | 02:21 |
| Matias | cloudnull: if owner is already ok, the file module with recurse is neatly instantaneous, which is good | 02:22 |
| openstackgerrit | Merged openstack/openstack-ansible-os_tempest: Ensure tempest runs independently in testing https://review.openstack.org/366335 | 02:22 |
| Matias | it only takes long if it needs to write to the disk | 02:22 |
| Matias | at least in my system | 02:22 |
| Matias | probably because the entire repo is cached in RAM | 02:22 |
| Matias | s/neatly/nearly/ | 02:23 |
| cloudnull | interesting. | 02:23 |
| Matias | TASK: repo_server : Ensure git directory permissions ------------------- 81.00s | 02:23 |
| Matias | that's how long it takes | 02:23 |
| Matias | now let me measure again the ansible task when it has to change the owner | 02:24 |
| cloudnull | everytime ? | 02:24 |
| Matias | probably yes, but just a sec I will measure when it needs to change something | 02:24 |
| Matias | cloudnull: unrelated, the memcached containers seem to be logging the entire memcached traffic to /var/log/syslog | 02:28 |
| Matias | this is probably a huge hog | 02:28 |
| Matias | -vv very verbose (also print client commands/reponses) | 02:30 |
| Matias | probably because of this | 02:30 |
| Matias | cloudnull: https://github.com/openstack/openstack-ansible-memcached_server/blob/master/templates/memcached.conf.debian.j2#L7 | 02:31 |
| Matias | I think here it should be -v | 02:31 |
| Matias | https://github.com/openstack/openstack-ansible-memcached_server/commit/c99d1debe905c06a71c457f60045428533988d5e | 02:32 |
| Matias | I don't get why it was increased here | 02:32 |
| Matias | /dev/lxc/controller-02_memcached_container-18575240 4.8G 4.6G 0 100% / | 02:33 |
| cloudnull | hum. | 02:34 |
| cloudnull | so we should turn that log facility off ? | 02:34 |
| Matias | well, it is easily filling up the container disk until the rotation kicks in and compresses the log | 02:35 |
| cloudnull | ok | 02:35 |
| Matias | I think memcached traffic is overly intense to log every message | 02:36 |
| *** thorst_ has joined #openstack-ansible | 02:36 | |
| Matias | logging every sql statement in galera would be less of a hog than logging memcached, I think | 02:37 |
| Matias | TASK: repo_server : Git service data folder setup --------------------- 371.70s | 02:37 |
| Matias | TASK: repo_server : Ensure git directory permissions ------------------- 82.54s | 02:37 |
| Matias | cloudnull: ^ | 02:37 |
| Matias | that's when it needs to change both owner/group and permissions | 02:38 |
| Matias | so changing the mode is like a constant 80 seconds | 02:38 |
| Matias | and changing the owner is slow like hell, but ansible is smart enough to only write to disk if needed | 02:38 |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible-memcached_server: Lower verbosity on memcached server https://review.openstack.org/369784 | 02:39 |
| cloudnull | Matias: ^ memcached | 02:39 |
| Matias | so that if the metadata is in RAM cache it is instantaneous | 02:39 |
| cloudnull | so I think we need to remove recurse from 'Git service data folder setup' | 02:40 |
| Matias | nice, thanks for the memcached patch | 02:40 |
| cloudnull | and then just add the chmod into 'Ensure git directory permissions' | 02:40 |
| Matias | well, it is a compromise you make | 02:40 |
| cloudnull | **chown | 02:41 |
| Matias | chown is faster when you actually need to change stuff | 02:41 |
| Matias | however it is slower when you are only checking if it is OK | 02:41 |
| Matias | another thing I just noticed is that to compute state we are calling ls -l | 02:42 |
| cloudnull | but it looks like the additional recurse function is SUPER slow | 02:42 |
| Matias | which would list files twice | 02:42 |
| Matias | one for the dir which contains the file, another for the file itself | 02:42 |
| Matias | we could do find -type d, or find -exec ls -ld | 02:42 |
| Matias | cloudnull: "Git service data folder setup" runs in a few milliseconds if the owner/group is already correct | 02:43 |
| Matias | it only takes 370s if it needs to change the owner/group of every file | 02:43 |
| Matias | so it's a single-time wait | 02:44 |
| Matias | e.g. when upgrading from mitaka to newton | 02:44 |
| cloudnull | fair point | 02:44 |
| cloudnull | and going from here those perms should not be broken | 02:44 |
| Matias | but again, if caches are clean it would take longer, didn't measure how much | 02:44 |
| cloudnull | and if they are they'll be corrected | 02:44 |
| Matias | but find would need to go across the entire tree anyway | 02:45 |
| cloudnull | I'm good with the single hit when upgrading | 02:46 |
| Matias | however, the 80s hit of "Ensure git directory permissions" will always occur | 02:46 |
| *** hblah has quit IRC | 02:46 | |
| Matias | but I can't think of anything we could do about that | 02:46 |
| cloudnull | we could only run that if the previous task changed | 02:48 |
| Matias | cloudnull: wow | 02:49 |
| Matias | TASK: repo_server : Ensure git directory permissions ------------------- 28.62s | 02:49 |
| Matias | cloudnull: that was changing from find "${GIT_DIR}" -exec ls -l {} \; | 02:50 |
| Matias | to find "${GIT_DIR}" -type d -exec ls -l {} \; | 02:50 |
| Matias | so that files are listed only once | 02:50 |
| Matias | sent as comments in the PR | 02:54 |
| Matias | stevelle: it would be strange that the 3 nodes would have the same issue if it was a networking issue | 02:57 |
| Matias | stevelle: also no other sympton arises in the other containers, other rabbit queues, nor other protocols running over the management network | 02:58 |
| Matias | stevelle: it's just ceilometer-agent-notification, and if I restart it, it consumes the messages stuck in the queue | 02:58 |
| Matias | only to get stuck again after a few hours | 02:58 |
| *** dragonmaster has quit IRC | 03:01 | |
| Matias | well, need to get some sleep | 03:01 |
| Matias | it's already midnight here | 03:01 |
| Matias | cloudnull: thanks, see you tomorrow | 03:02 |
| cloudnull | Matias: cool so lets change that in the task | 03:18 |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible-repo_server: Set permissions on the openstackgit directory https://review.openstack.org/368997 | 03:20 |
| cloudnull | ^ done | 03:20 |
| *** itsuugo has quit IRC | 03:21 | |
| *** itsuugo has joined #openstack-ansible | 03:23 | |
| openstackgerrit | melissaml proposed openstack/openstack-ansible: Modify use of assertTrue(A in B) https://review.openstack.org/369803 | 03:24 |
| *** schwicht has joined #openstack-ansible | 03:27 | |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible-repo_server: Convert role testing to use Ansible 2.1.1 https://review.openstack.org/369782 | 03:28 |
| *** adrian_otto has joined #openstack-ansible | 03:31 | |
| *** itsuugo has quit IRC | 03:32 | |
| *** itsuugo has joined #openstack-ansible | 03:34 | |
| *** hughmFLEXin has joined #openstack-ansible | 03:35 | |
| hughmFLEXin | du -a /var | sort -n -r | head -n 10 | 03:42 |
| hughmFLEXin | oops | 03:42 |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible-plugins: Implement distro package lookups and index https://review.openstack.org/363022 | 03:45 |
| cloudnull | not enough space ? | 03:45 |
| cloudnull | i like du -ah /var | sort -n -r | head -n 10 -- because i'm a dumb human | 03:47 |
| cloudnull | :) | 03:47 |
| *** itsuugo has quit IRC | 03:48 | |
| *** itsuugo has joined #openstack-ansible | 03:49 | |
| *** thorst_ has quit IRC | 03:50 | |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible: Remove assumption that the neutron_lbaas var is set https://review.openstack.org/368905 | 03:51 |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible: [TESTING] Ansible Devel https://review.openstack.org/351165 | 03:53 |
| *** adrian_otto has quit IRC | 03:56 | |
| *** adrian_otto has joined #openstack-ansible | 03:56 | |
| *** itsuugo has quit IRC | 03:59 | |
| *** itsuugo has joined #openstack-ansible | 04:00 | |
| *** thorst_ has joined #openstack-ansible | 04:02 | |
| *** hblah has joined #openstack-ansible | 04:06 | |
| *** dragonmaster has joined #openstack-ansible | 04:07 | |
| *** hj-hpe has quit IRC | 04:08 | |
| *** hblah has quit IRC | 04:08 | |
| *** hj-hpe has joined #openstack-ansible | 04:08 | |
| *** hj-hpe has quit IRC | 04:09 | |
| *** hblah has joined #openstack-ansible | 04:10 | |
| *** itsuugo has quit IRC | 04:10 | |
| *** Fenuks has joined #openstack-ansible | 04:11 | |
| *** itsuugo has joined #openstack-ansible | 04:12 | |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible-lxc_container_create: Change the container prep into a single task https://review.openstack.org/354120 | 04:14 |
| *** adrian_otto has quit IRC | 04:18 | |
| *** weezS has quit IRC | 04:18 | |
| *** itsuugo has quit IRC | 04:19 | |
| *** weezS has joined #openstack-ansible | 04:20 | |
| *** thorst_ has quit IRC | 04:20 | |
| *** itsuugo has joined #openstack-ansible | 04:22 | |
| *** thorst_ has joined #openstack-ansible | 04:28 | |
| *** Fenuks has quit IRC | 04:32 | |
| *** Fenuks has joined #openstack-ansible | 04:32 | |
| *** thorst_ has quit IRC | 04:33 | |
| *** itsuugo has quit IRC | 04:42 | |
| *** itsuugo has joined #openstack-ansible | 04:44 | |
| *** askb has quit IRC | 04:48 | |
| *** askb has joined #openstack-ansible | 04:49 | |
| *** itsuugo has quit IRC | 04:51 | |
| *** thorst_ has joined #openstack-ansible | 04:52 | |
| *** itsuugo has joined #openstack-ansible | 04:53 | |
| *** thorst_ has quit IRC | 04:56 | |
| *** Jeffrey4l_ has joined #openstack-ansible | 05:00 | |
| *** thorst_ has joined #openstack-ansible | 05:03 | |
| *** Jeffrey4l has quit IRC | 05:04 | |
| *** thorst_ has quit IRC | 05:08 | |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible-lxc_container_create: Change the container prep into a single task https://review.openstack.org/354120 | 05:11 |
| *** weezS has quit IRC | 05:11 | |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible-lxc_container_create: Change the container prep into a single task https://review.openstack.org/354120 | 05:14 |
| *** thorst_ has joined #openstack-ansible | 05:15 | |
| *** phschwartz has quit IRC | 05:15 | |
| *** phschwartz has joined #openstack-ansible | 05:18 | |
| *** thorst_ has quit IRC | 05:20 | |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible-lxc_container_create: Change the container prep using more intelligent commands https://review.openstack.org/354120 | 05:23 |
| *** dragonmaster has quit IRC | 05:25 | |
| *** hughmFLEXin has quit IRC | 05:26 | |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible-lxc_container_create: Change the container prep using more intelligent commands https://review.openstack.org/354120 | 05:26 |
| *** dragonmaster has joined #openstack-ansible | 05:27 | |
| *** itsuugo has quit IRC | 05:28 | |
| *** itsuugo has joined #openstack-ansible | 05:29 | |
| SamYaple | cloudnull: go to bed | 05:32 |
| cloudnull | ha! | 05:32 |
| cloudnull | you go to bed. | 05:32 |
| cloudnull | :) | 05:32 |
| SamYaple | i am in bed | 05:32 |
| SamYaple | checkmate | 05:32 |
| cloudnull | im on the couch | 05:32 |
| cloudnull | thats close | 05:32 |
| cloudnull | :P | 05:32 |
| SamYaple | so ive been working on the snappy packaging thing. there is a small little group that has formed | 05:33 |
| SamYaple | including the guys that do the openstack juju stuff | 05:34 |
| SamYaple | been trying to think of a way to add that in as an lxc alternative. i think i see a few entry points | 05:34 |
| SamYaple | gotta refresh myself on the project more | 05:34 |
| cloudnull | hows snappy been shaping up? | 05:35 |
| SamYaple | ive been hammering on the python plugin and its in good shape (as of today) | 05:35 |
| SamYaple | good enough for openstack anyway | 05:35 |
| cloudnull | sweet! | 05:35 |
| cloudnull | shipit! | 05:35 |
| cloudnull | :P | 05:35 |
| SamYaple | there are still some daemon stuff things that im worried about | 05:35 |
| SamYaple | its merged upstream, yea | 05:35 |
| SamYaple | actually funny story about that, they have coverage tests for merging and I failed the coverage test, overall coverage dropped | 05:36 |
| SamYaple | HOWEVER, i raised the coverage of all the files I was working on to 100% | 05:36 |
| SamYaple | :) | 05:36 |
| SamYaple | if snapcraft as a whole had 100% coverage, my test would not have failed | 05:37 |
| *** thorst_ has joined #openstack-ansible | 05:37 | |
| cloudnull | its the wheel of progress. | 05:38 |
| cloudnull | :) | 05:38 |
| cloudnull | ima go sleep now. i needed to get a few more things done so i could put my mind at rest. and its gfn . | 05:39 |
| cloudnull | have a good night SamYaple ttys | 05:39 |
| *** jamielennox|away is now known as jamielennox | 05:40 | |
| *** itsuugo has quit IRC | 05:43 | |
| *** itsuugo has joined #openstack-ansible | 05:45 | |
| SamYaple | you too cloudnull nigh | 05:46 |
| *** itsuugo has quit IRC | 05:52 | |
| *** itsuugo has joined #openstack-ansible | 05:53 | |
| *** hughmFLEXin has joined #openstack-ansible | 06:01 | |
| *** galstrom_zzz is now known as galstrom | 06:06 | |
| *** itsuugo has quit IRC | 06:07 | |
| *** itsuugo has joined #openstack-ansible | 06:09 | |
| odyssey4me | o/ | 06:10 |
| openstackgerrit | Merged openstack/openstack-ansible-os_glance: Fix os_glance role functional testing https://review.openstack.org/369726 | 06:17 |
| *** itsuugo has quit IRC | 06:27 | |
| *** itsuugo has joined #openstack-ansible | 06:29 | |
| *** pcaruana has joined #openstack-ansible | 06:32 | |
| *** itsuugo has quit IRC | 06:34 | |
| *** itsuugo has joined #openstack-ansible | 06:35 | |
| *** dragonmaster has quit IRC | 06:37 | |
| *** itsuugo has quit IRC | 06:40 | |
| odyssey4me | relocating | 06:40 |
| *** itsuugo has joined #openstack-ansible | 06:41 | |
| *** dragonmaster has joined #openstack-ansible | 06:46 | |
| *** itsuugo has quit IRC | 06:49 | |
| *** itsuugo has joined #openstack-ansible | 06:50 | |
| *** hblah026 has joined #openstack-ansible | 07:04 | |
| *** hblah has quit IRC | 07:06 | |
| *** hblah has joined #openstack-ansible | 07:07 | |
| *** hblah026 has quit IRC | 07:11 | |
| *** thorst_ has quit IRC | 07:12 | |
| *** pmannidi has quit IRC | 07:13 | |
| *** drifterza has joined #openstack-ansible | 07:14 | |
| *** hblah469 has joined #openstack-ansible | 07:16 | |
| *** galstrom is now known as galstrom_zzz | 07:16 | |
| *** hblah has quit IRC | 07:17 | |
| drifterza | Hello peeps | 07:20 |
| *** karimb has joined #openstack-ansible | 07:22 | |
| *** hughmFLEXin has quit IRC | 07:25 | |
| *** fxpester has joined #openstack-ansible | 07:35 | |
| *** itsuugo has quit IRC | 07:39 | |
| odyssey4me | back | 07:40 |
| *** itsuugo has joined #openstack-ansible | 07:41 | |
| *** bgmccollum has quit IRC | 07:41 | |
| *** homerp_ has quit IRC | 07:42 | |
| *** bgmccollum has joined #openstack-ansible | 07:42 | |
| *** homerp has joined #openstack-ansible | 07:42 | |
| *** karimb has quit IRC | 07:46 | |
| *** hughmFLEXin has joined #openstack-ansible | 07:47 | |
| Attilla | odyssey4me: I've checked and haproxy is running, listening on all interfaces for the keystond admin port | 07:50 |
| Attilla | So that should be forwarded to the correct container | 07:51 |
| Attilla | Goodmorning by the way :-) | 07:51 |
| Attilla | So it is still unclear to me why that auth error is occuring | 07:53 |
| *** schwicht has quit IRC | 07:53 | |
| odyssey4me | Attilla did you try enabling debug logging on keystone, then monitoring the logs while the heat playbook executes? | 07:54 |
| *** jascott1 has quit IRC | 07:55 | |
| *** thorst_ has joined #openstack-ansible | 07:56 | |
| Attilla | I'm kinda an openstack newbie, because I only want to play around with the networking side of it. I'm just running the "run-playbook" script that takes about 90minutes to get where I am now | 07:57 |
| Attilla | I can't really turn on anything, because it simply never completed a build | 07:58 |
| *** thorst_ has quit IRC | 07:59 | |
| *** thorst_ has joined #openstack-ansible | 08:01 | |
| *** jwitko has quit IRC | 08:01 | |
| *** schwicht has joined #openstack-ansible | 08:02 | |
| *** permalac has joined #openstack-ansible | 08:03 | |
| Attilla | So it is hard to debug any further than that. | 08:06 |
| *** admin0 has joined #openstack-ansible | 08:07 | |
| *** qiliang27 has joined #openstack-ansible | 08:15 | |
| *** thorst_ has quit IRC | 08:16 | |
| *** thorst_ has joined #openstack-ansible | 08:21 | |
| odyssey4me | Attilla OK, let me help you get it done | 08:24 |
| odyssey4me | Attilla are you using an AIO for this build? | 08:24 |
| *** hblah469 has quit IRC | 08:25 | |
| *** ikalnitsky is now known as ikalnytskyi | 08:27 | |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-repo_build: Add binary package lookup output for repo build tests https://review.openstack.org/365458 | 08:28 |
| *** asettle has joined #openstack-ansible | 08:29 | |
| pjm6 | Hi there :) | 08:35 |
| *** thorst_ has quit IRC | 08:38 | |
| *** thorst_ has joined #openstack-ansible | 08:39 | |
| Attilla | odyssey4me: I based my config on the AIO example | 08:39 |
| odyssey4me | Attilla ok, to enable debug logging set 'debug: yes' or 'debug: true' in /etc/openstack_deploy/user_variables.yml | 08:40 |
| odyssey4me | when that's done, execute this: cd /opt/openstack-ansible/playbooks; openstack-ansible os-keystone-install.yml | 08:40 |
| odyssey4me | that will reconfigure keystone to use debug logging | 08:41 |
| Attilla | ok, have to run to a customer, back in a bit | 08:41 |
| odyssey4me | then open a session to each keystone container, and in each container tail the keystone logs | 08:41 |
| odyssey4me | hughsaunders could you review https://review.openstack.org/363022 please if you have a moment? | 08:44 |
| *** thorst_ has quit IRC | 08:44 | |
| hughsaunders | odyssey4me: yep | 08:44 |
| odyssey4me | hughsaunders note my review comment to see the sample output | 08:44 |
| *** thorst_ has joined #openstack-ansible | 08:46 | |
| *** thorst_ has quit IRC | 08:48 | |
| *** itsuugo has quit IRC | 08:51 | |
| *** itsuugo has joined #openstack-ansible | 08:52 | |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible: Reduce minimum data disk size for the AIO to 55GB https://review.openstack.org/369944 | 08:53 |
| odyssey4me | andymccr evrardjp ^ we're getting bluebox failures because their disk size is 58GB - that patch should bypass that issue | 08:53 |
| odyssey4me | in master it's already reduced to 50GB thanks to overlayfs usage | 08:54 |
| *** thorst_ has joined #openstack-ansible | 08:54 | |
| openstackgerrit | Merged openstack/openstack-ansible: Ensure that gnocchi uses keystone authentication https://review.openstack.org/369610 | 08:56 |
| *** oben has quit IRC | 08:58 | |
| *** oneswig has joined #openstack-ansible | 08:59 | |
| *** thorst_ has quit IRC | 08:59 | |
| *** kstev has quit IRC | 09:00 | |
| *** kstev has joined #openstack-ansible | 09:00 | |
| *** itsuugo has quit IRC | 09:02 | |
| openstackgerrit | Merged openstack/openstack-ansible-plugins: Add option to toggle list extensions https://review.openstack.org/352054 | 09:02 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible: [TESTING] Ansible Devel https://review.openstack.org/351165 | 09:02 |
| *** itsuugo has joined #openstack-ansible | 09:03 | |
| odyssey4me | asettle a docs patch for review: https://review.openstack.org/364627 | 09:04 |
| hughsaunders | odyssey4me: so is the idea of that lookup patch to list all the required distro packages for a version so they can be artefacts or precached etc? | 09:04 |
| odyssey4me | hughsaunders yep | 09:04 |
| asettle | odyssey4me: gracias | 09:04 |
| odyssey4me | it'll allow us to use the lookup data to prep container base images without actually building a running container | 09:04 |
| odyssey4me | we can install all the apt and python packages in a chroot, and ignore all config implementation | 09:05 |
| *** thorst_ has joined #openstack-ansible | 09:09 | |
| *** thorst__ has joined #openstack-ansible | 09:12 | |
| *** thorst_ has quit IRC | 09:13 | |
| openstackgerrit | Merged openstack/openstack-ansible-plugins: Implement distro package lookups and index https://review.openstack.org/363022 | 09:16 |
| *** karimb has joined #openstack-ansible | 09:17 | |
| odyssey4me | evrardjp https://review.openstack.org/#/c/367521/ | 09:18 |
| odyssey4me | or look at any of the number of patches currently open for the role | 09:18 |
| *** thorst__ has quit IRC | 09:20 | |
| pjm6 | I have a doubt, normally in OS deployments exists redundancy of instances? | 09:20 |
| *** sdake_ has quit IRC | 09:28 | |
| *** thorst_ has joined #openstack-ansible | 09:28 | |
| *** itsuugo has quit IRC | 09:33 | |
| *** itsuugo has joined #openstack-ansible | 09:35 | |
| odyssey4me | pjm6 I don't think I understand the question? | 09:35 |
| *** thorst_ has quit IRC | 09:35 | |
| *** permalac has quit IRC | 09:38 | |
| *** b3rnard0 has joined #openstack-ansible | 09:41 | |
| *** haojing has joined #openstack-ansible | 09:42 | |
| haojing | #openstack-meeting | 09:43 |
| pjm6 | odyssey4me, lets imagine that my compute-node01 (which has 5 instances) | 09:44 |
| pjm6 | die | 09:44 |
| pjm6 | i loose that 5 instances, if are in a local storage | 09:44 |
| odyssey4me | haojing ? | 09:44 |
| pjm6 | my question was if its normal using shared storage, in that way, it's possible to recover those instances in the new compute noed | 09:45 |
| haojing | sorry, ignore that | 09:45 |
| odyssey4me | pjm6 yes, unless you have configured shared storage for nova's ephemeral storage (for example, using nfs or ceph) | 09:45 |
| pjm6 | odyssey4me, nfs performance with KVM is good? | 09:45 |
| odyssey4me | pjm6 that would depend on your networking | 09:46 |
| odyssey4me | pjm6 the options are to configure shared storage if you care about all your instance ephemeral disks, or if you don't care about them all - to set the instance to boot from volume (and your volume back-end must be networked storage and safe from disaster), or you configure the application to be able to run in a case of a disaster (ie if you lose the instance it does not matter) | 09:47 |
| vnogin | pjm6, also as another option to use disaster recovery tool :) | 09:47 |
| *** hughmFLEXin has quit IRC | 09:48 | |
| vnogin | morning osa :) o/ | 09:48 |
| pjm6 | thanks odyssey4me vnogin | 09:48 |
| pjm6 | vnogin, o/ | 09:48 |
| pjm6 | odyssey4me, one thing, in KVM its possible to configure a shared storage connection to a Storage | 09:48 |
| pjm6 | with icqn? | 09:48 |
| odyssey4me | pjm6 yes, as vnogin said - another option is to backup instances and restore them | 09:48 |
| vnogin | pjm6, we are trying to use freezer now | 09:48 |
| pjm6 | vnogin, freezer? | 09:49 |
| odyssey4me | pjm6 generally it is done by simply mounting the /var/lib/nova/instances folder onto shared storage - or using the ceph back-end configuration | 09:49 |
| vnogin | pjm6, https://wiki.openstack.org/wiki/Freezer | 09:49 |
| odyssey4me | pjm6 it has nothing to do with KVM - it's a file system mount | 09:49 |
| pjm6 | vnogin, thanks, i will take a look to it :) | 09:50 |
| *** hughmFLEXin has joined #openstack-ansible | 09:50 | |
| pjm6 | odyssey4me, yes I see, but correct me if I'm wrong | 09:50 |
| *** haojing has quit IRC | 09:50 | |
| pjm6 | if I use a ICQN connection individually for each compute node, that would cause problems right? | 09:51 |
| pjm6 | if exists concurrent access's | 09:51 |
| *** hughmFLEXin has quit IRC | 09:53 | |
| odyssey4me | pjm6 well - not really - each host is exclusively writing to their own files... no two compute hosts can be trying to write to the same files... but yes, that is a risk | 09:55 |
| odyssey4me | that is why using ceph is a better model | 09:55 |
| pjm6 | yes, i could just create a icqn to that specific host | 09:55 |
| pjm6 | but if I understood well | 09:55 |
| pjm6 | the ceph is good if we have many disks | 09:56 |
| pjm6 | right? | 09:56 |
| odyssey4me | pjm6 one method is to set each host to have their own exclusive storage space on a storage unit - this is essentially just offloading the storage from local disk to somewhere else which perhaps has better DR and is easier to re-mount to another server if need be | 09:56 |
| odyssey4me | using ceph it's nice because you get the liver-migration benefits that a shared storage unit gives you, but you don't have to deal with the issue of multiple compute units trying to access the same LUN | 09:57 |
| Attilla | odyssey4me: Had to redeploy it, so have to wait for it to finish to start some more debugging. | 09:58 |
| Attilla | Thanks already :-) | 09:58 |
| odyssey4me | Attilla it may be best to actually use a built and working AIO for reference | 09:59 |
| pjm6 | I see odyssey4me, but with shared storage live migration works too, no? | 10:01 |
| *** thorst_ has joined #openstack-ansible | 10:02 | |
| odyssey4me | pjm6 yes it does, but only if two compute hosts have access to the same LUN | 10:03 |
| pjm6 | hmm I see :) thanks odyssey4me | 10:03 |
| pjm6 | one last question for now xD how do you deal with OS ugprades ? | 10:03 |
| pjm6 | for instance, updating the compute host | 10:04 |
| pjm6 | sometimes needs to reboot (because kernel) | 10:04 |
| pjm6 | a possible scenario I see is migrating all compute nodes from that host to another one, reboot, and then just put it again | 10:04 |
| odyssey4me | pjm6 ideally, live migrate the instances to another host, disable the compute agent on the host you want to upgrade, upgrade the compute host, then verify that it's fine, then enable the compute agent again | 10:05 |
| odyssey4me | then migrate the instances back | 10:05 |
| pjm6 | seems simple and straightforward :D | 10:05 |
| vnogin | odyssey4me, pjm6 as option bifrost and ironic ansible driver can be used for provisioning :) | 10:05 |
| odyssey4me | vnogin well, yes - whatever your provisioning system is is rather immaterial | 10:06 |
| *** oneswig has quit IRC | 10:06 | |
| odyssey4me | for example, logan- has a system where the compute OS is PXE booted so it is always rebuilt from the latest software on reboot | 10:06 |
| vnogin | odyssey4me, agree :) as I said - as option )) | 10:06 |
| pjm6 | vnogin, i was seeing, the bifrost is a clean install of a base image pre configured ? | 10:07 |
| odyssey4me | personally I would rather be using cobbler - trying to use pre-built images would require fairly homogenous hardware, whereas a scripted build may take longer but more easily handles disparate hardware | 10:07 |
| odyssey4me | but each to their own | 10:07 |
| odyssey4me | instead of bifrost you could implement a standalone ironic install using out ironic role too... bifrost is just a standalone ironic implementation | 10:08 |
| odyssey4me | our role caters for both a standalone and an integrated build | 10:08 |
| vnogin | pjm6, yep | 10:08 |
| pjm6 | I was thinking in adding pxe on boot, but I think that could be manual tasks to do, like networking bonding and other stuff | 10:10 |
| pjm6 | (like Mirantis do) | 10:10 |
| *** thorst_ has quit IRC | 10:11 | |
| vnogin | odyssey4me, I've sent you how I use bifrost in private message :) | 10:11 |
| vnogin | pjm6, do you mean in fuel? | 10:12 |
| pjm6 | Yeah | 10:12 |
| *** thorst_ has joined #openstack-ansible | 10:12 | |
| vnogin | pjm6, yep, it can be done :) | 10:16 |
| Attilla | odyssey4me: I can't even get AIO to work or any setup for that matter :P | 10:16 |
| pjm6 | But you use PXE on boot to have a clean OS installed with openstack services running, or just with network setup | 10:16 |
| pjm6 | and after it | 10:16 |
| pjm6 | run openstack-ansible? | 10:17 |
| odyssey4me | Attilla in that case, I would recommend starting with a standard AIO - get that working before trying to customise anything | 10:17 |
| odyssey4me | Attilla the AIO is an opportunity for you to figure out what bits may need to be configured to cater for firewalling, proxies, etc - it's a known good configuration | 10:17 |
| *** thorst_ has quit IRC | 10:20 | |
| Attilla | odyssey4me: Already tried that, but even a plain AIO setup didn't work. But I'm trying this on VM's, would that be the issue? | 10:22 |
| openstackgerrit | Andy McCrae proposed openstack/openstack-ansible-openstack_openrc: Add alias for OS_ENDPOINT_TYPE https://review.openstack.org/370009 | 10:23 |
| Attilla | And my config that I have is almost the same as AIO, just with two different compute nodes | 10:24 |
| odyssey4me | Attilla if a plain AIO is not working, then you need to resolve that before trying to build out a multi-node deployment... whatever issue you hit for an AIO is going to hit you for multi-node too... but multi-node will also be plagued by other possible config issues | 10:25 |
| vnogin | pjm6, only basic network configuration. Then ansible configures what I need on the node and then openstack-ansible :) | 10:25 |
| odyssey4me | Attilla from what I saw your config is an AIO anyway, but with your own config attempt... so you may as well get it right on an AIO first | 10:25 |
| odyssey4me | (using our known AIO config) | 10:26 |
| pjm6 | vnogin, so you just use PXE to acelerate the preparation of host and then run your ansible script and then openstack-ansible | 10:26 |
| vnogin | odyssey4me, installation guide for AIO should be reviewed :)))) First time I also faced a lot of issues :) | 10:26 |
| Attilla | vnogin: There are some parts of the guide that aren't really clear if you start with all of this. | 10:27 |
| vnogin | Attilla, totally agree with you :) | 10:27 |
| pjm6 | I just updated to last openstack-ansible and i'm getting error in "Get remove venv checksum" | 10:27 |
| pjm6 | it gives me a 404 Not found page | 10:27 |
| odyssey4me | Attilla this is why the AIO is a great tool with a known configuration to explore things | 10:28 |
| pjm6 | forget, I just forgot to update the other playbooks | 10:28 |
| odyssey4me | and yes, we've done a ton of stuff to the guide and are looking for feedback for improvements - please register bugs for anything needed... points of clarity, example configs, etc | 10:28 |
| vnogin | pjm6, correct. I don't want to provision each node using IPMI manually :) | 10:28 |
| pjm6 | I just need to excecute boostrap-ansible after updating the openstack-ansible, right ? | 10:28 |
| odyssey4me | ideally then once you have the right understanding, help us to improve the install guide to make it better for newcomers | 10:29 |
| Attilla | odyssey4me: Will do,, in general there is a lot of documentation | 10:29 |
| pjm6 | vnogin, yes, it make sense :D | 10:29 |
| Attilla | But I'm just trying to get it running in our reference topology to test with vxlan overlay etc. | 10:29 |
| odyssey4me | pjm6 http://docs.openstack.org/developer/openstack-ansible/mitaka/install-guide/app-minorupgrade.html | 10:29 |
| Attilla | https://raw.githubusercontent.com/CumulusNetworks/cldemo-vagrant/master/cldemo_topology_l3.png | 10:29 |
| pjm6 | thanks odyssey4me, after a while i didn't remember | 10:30 |
| *** qiliang27 has quit IRC | 10:33 | |
| vnogin | odyssey4me, evrardjp guys are you going to summit? :) | 10:35 |
| openstackgerrit | Merged openstack/openstack-ansible: Fix log path option https://review.openstack.org/367649 | 10:35 |
| openstackgerrit | Merged openstack/openstack-ansible: Reduce minimum data disk size for the AIO to 55GB https://review.openstack.org/369944 | 10:35 |
| Fenuks | Hello. Anyone know if it's possible to enable Swift3 and EC2 credentials with OSA? | 10:35 |
| *** itsuugo has quit IRC | 10:36 | |
| evrardjp | vnogin: we are | 10:37 |
| *** itsuugo has joined #openstack-ansible | 10:37 | |
| openstackgerrit | Merged openstack/openstack-ansible: Use correct version when creating backup dir https://review.openstack.org/369650 | 10:37 |
| evrardjp | vnogin: will you? | 10:37 |
| evrardjp | I'd be happy to meet you! | 10:37 |
| vnogin | evrardjp, yep :) confirmed :) so I'd be happy to meet you too :) | 10:37 |
| odyssey4me | Fenuks as far as I recall, those options are out of tree now for the respective projects - and no-one's done the integration work to bring the new out of tree options back to work in OSA | 10:38 |
| odyssey4me | until someone does, support will diminish | 10:38 |
| Fenuks | odyssey4me: Out of tree in OSA or in Openstack in general? | 10:38 |
| pjm6 | nice :D i will go to | 10:38 |
| *** thorst_ has joined #openstack-ansible | 10:39 | |
| evrardjp | we'll have to coordinate this | 10:39 |
| evrardjp | if that's the proper term | 10:39 |
| odyssey4me | Fenuks openstack in general - the options are there, but they are no longer part of the swift, keystone, nova code trees... they are now independent projects | 10:39 |
| *** karimb has quit IRC | 10:40 | |
| *** karimb has joined #openstack-ansible | 10:41 | |
| Fenuks | odyssey4me: Is there anything to replace them? I need a way to store objects in swift from within virtual instances, but any user in openstack has access to entire project and is a huge security risk should it be compromised | 10:41 |
| *** rk4n has joined #openstack-ansible | 10:41 | |
| odyssey4me | Fenuks ok, so digging into the current state - the thing I was thinking about was the ec2 api which is now in https://github.com/openstack/ec2-api and has been pulled out of nova's code tree | 10:41 |
| odyssey4me | trying to find whther keystone/swift support any amazon bits these days | 10:42 |
| pjm6 | odyssey4me, that link: ec2-api its for using the ec2 API in the openstack, right? | 10:44 |
| odyssey4me | pjm6 yes | 10:45 |
| odyssey4me | ah, so I see that swift has its own repo for s3 support too: http://docs.openstack.org/mitaka/config-reference/object-storage/configure-s3.html | 10:47 |
| odyssey4me | andymccr ^ FYI | 10:47 |
| odyssey4me | so yes, at this point we have no s3 support - but it'd likely be fairly trivial to instrument it | 10:47 |
| *** rk4n has quit IRC | 10:48 | |
| *** rk4n has joined #openstack-ansible | 10:48 | |
| *** itsuugo has quit IRC | 10:48 | |
| *** itsuugo has joined #openstack-ansible | 10:49 | |
| Fenuks | odyssey4me: Trivial? Any guess how soon can that be? Or, perhaps, you can kick me in the right direction so I can try to implement it myself? | 10:49 |
| *** thorst_ has quit IRC | 10:51 | |
| *** rk4n has quit IRC | 10:53 | |
| *** itsuugo has quit IRC | 10:56 | |
| *** rk4n has joined #openstack-ansible | 10:58 | |
| *** itsuugo has joined #openstack-ansible | 10:58 | |
| *** thorst_ has joined #openstack-ansible | 10:59 | |
| *** psilvad has quit IRC | 11:01 | |
| *** electrofelix has joined #openstack-ansible | 11:04 | |
| odyssey4me | Fenuks you could learn from the integration of the lxd and powervm drivers for nova, but apply it to swift | 11:05 |
| odyssey4me | you'd need to optionally include the swift3 repo, optionally include the configs for it | 11:06 |
| Fenuks | odyssey4me: Thanks, will try | 11:07 |
| Fenuks | odyssey4me: But I see nothing about lxd or powervm in ansible roles or OSA playbooks… | 11:10 |
| *** itsuugo has quit IRC | 11:10 | |
| *** thorst_ has quit IRC | 11:11 | |
| odyssey4me | Fenuks the facilitation is in https://github.com/openstack/openstack-ansible-os_nova | 11:11 |
| *** itsuugo has joined #openstack-ansible | 11:11 | |
| odyssey4me | Fenuks you can see some of the patches in the whiteboard here: https://blueprints.launchpad.net/openstack-ansible/+spec/nova-lxd-support | 11:12 |
| odyssey4me | some more here: https://blueprints.launchpad.net/openstack-ansible/+spec/powervm-virt-driver | 11:13 |
| Fenuks | Ah, that's an upcomming feature, not in mitaka yet? | 11:13 |
| odyssey4me | Fenuks yes, it'll be part of the NEwton release | 11:13 |
| odyssey4me | all development starts at the master branch, so if you want a feature to be included in the next release then that's where you build it | 11:14 |
| Fenuks | Oh, alright, looks like it is going to be feature-rich | 11:14 |
| *** thorst_ has joined #openstack-ansible | 11:16 | |
| Matias | morning | 11:18 |
| Matias | odyssey4me: sorry for the crazy "recheck" | 11:19 |
| * Matias poluting the gate | 11:19 | |
| odyssey4me | Matias ? | 11:20 |
| openstackgerrit | Jean-Philippe Evrard proposed openstack/openstack-ansible-os_neutron: Ensure the nets, subnets and ports are present https://review.openstack.org/370046 | 11:20 |
| Matias | odyssey4me: that one where I tought Depends-On pulled the repo with the referenced PR | 11:21 |
| *** thorst_ has quit IRC | 11:21 | |
| odyssey4me | Matias ah ok, no that's a common thing - no worries... I do hope to make that work at some point | 11:21 |
| *** rk4n has quit IRC | 11:22 | |
| *** itsuugo has quit IRC | 11:24 | |
| *** arbrandes has joined #openstack-ansible | 11:26 | |
| *** itsuugo has joined #openstack-ansible | 11:26 | |
| Fenuks | One more silly question — after I've installed Openstack with OSA I did not change anything by hand. Since then, OSA has 50 new commits. Is it safe to pull them, bootstrap again and run OSA? | 11:27 |
| Fenuks | It should update everything that has changed? | 11:28 |
| *** arbrandes1 has quit IRC | 11:28 | |
| Matias | Fenuks: usually yes, but to have >99.9% guarantee in a production env, you would need to checkout only stable tags, and update only across them | 11:29 |
| Fenuks | Matias: I'm using stable/mitaka | 11:29 |
| Matias | Fenuks: are you checking out the branch directly, or the version tags? | 11:30 |
| Fenuks | Matias: Branch | 11:30 |
| Matias | it is recommended to checkout version tags to guarantee you do not get something somewhat WIP in the middle | 11:31 |
| Matias | however usually you shouldn't have problems | 11:31 |
| Fenuks | Alright, I'll switch to tags, there should be some ahead | 11:32 |
| Matias | nice | 11:32 |
| *** retreved has joined #openstack-ansible | 11:34 | |
| *** openstackgerrit has quit IRC | 11:34 | |
| *** openstackgerrit has joined #openstack-ansible | 11:35 | |
| Matias | gnocchi is generating enough swift traffic to fill up my swift-proxy-container disk | 11:35 |
| Matias | and that's because I only have 2 instances running | 11:35 |
| drifterza | I had to do filtering | 11:35 |
| Matias | drifterza: share please :D | 11:36 |
| Matias | dude are you only online during late night in new zealand? :P | 11:37 |
| drifterza | lol its 13:37 here | 11:37 |
| Matias | ah, then you are not in new zealand | 11:38 |
| Matias | as I supposed from your nickname | 11:38 |
| drifterza | za=south africa | 11:38 |
| Matias | ah lol | 11:38 |
| Matias | new zealand = nz | 11:38 |
| Matias | mental bug | 11:39 |
| drifterza | hehe | 11:39 |
| drifterza | jeeze trying to find where i did it | 11:39 |
| drifterza | one sec. | 11:39 |
| Matias | drifterza: I partially found my problem.. OSA was not being able to set the fd limit in systemd, so the rabbit servers were getting full of connections | 11:41 |
| drifterza | ah, bug in the systemd scripts ? | 11:41 |
| drifterza | [dispatcher_gnocchi] | 11:41 |
| drifterza | filter_project = gnocchi_swift | 11:41 |
| drifterza | filter_service_activity = True | 11:41 |
| drifterza | archive_policy = low | 11:41 |
| drifterza | let me check if thats in master. | 11:42 |
| Matias | yes, i'm working on it now | 11:42 |
| Matias | drifterza: I included this using an override | 11:42 |
| Matias | saw it in gnocchi docs | 11:42 |
| *** asettle has quit IRC | 11:42 | |
| Matias | problem is, i'm using archive_policy medium | 11:42 |
| Matias | so probably my gnocchi is handling 5x more data than yours | 11:43 |
| drifterza | aah riight | 11:43 |
| *** asettle has joined #openstack-ansible | 11:43 | |
| drifterza | yes that is true | 11:43 |
| drifterza | Note that it configures a filter to filter out all samples generated by Gnocchi. Otherwise each time we write to Swift that will generate samples to write again to Swift and this will create a storm of samples that grows indefinitely. The filter breaks this infinite loop. | 11:43 |
| Matias | yes | 11:43 |
| Matias | this is being integrated into the playbook by stevelle | 11:43 |
| Matias | there are some pending PRs for taht | 11:43 |
| drifterza | Ah right sweet. | 11:44 |
| drifterza | i haven't looked at the gnocchi role in a while | 11:44 |
| drifterza | :P | 11:44 |
| Matias | he did it in another way, on which he filters directly into the swift ceilometermiddleware | 11:44 |
| Matias | so that generates less rabbit traffic | 11:44 |
| drifterza | nice, I should of checked before spreading my lies then :) | 11:44 |
| Matias | but in the meantime I'm using the same method as yours, using an ceilometer_ceilometer_conf_override | 11:44 |
| *** karimb has quit IRC | 11:44 | |
| Matias | drifterza: the stevelle sweetness is not in master yet | 11:45 |
| Matias | still a PR | 11:45 |
| Matias | unmerged | 11:45 |
| drifterza | cool at least its been tagged then | 11:45 |
| drifterza | so thats good. | 11:45 |
| Matias | now my issue is lots of swift traffic, even if the positive feedback loop is already cutted | 11:45 |
| drifterza | which metric ? | 11:46 |
| Matias | I think I need to logrotate once per hour, maybe | 11:46 |
| drifterza | odyssey4me, you around ? | 11:47 |
| Matias | just default metrics | 11:47 |
| Matias | from the ceilometer's default pipeline | 11:47 |
| *** asettle has quit IRC | 11:47 | |
| Matias | but I reduced the interval from 600 to 15 seconds | 11:47 |
| drifterza | so with the gnocchiclient you do metric list, can you see which one is the noisy one ? | 11:48 |
| drifterza | or is it all in general | 11:48 |
| Matias | it's all in general | 11:48 |
| drifterza | ah, then its your sample rate then | 11:48 |
| drifterza | most likely | 11:48 |
| Matias | 22M /var/log/swift/proxy.log | 11:48 |
| Matias | that with 7min running | 11:48 |
| drifterza | yeah you could turn the verbose off on the swift-prox | 11:49 |
| drifterza | yeah you could turn the verbose off on the swift-proxy | 11:49 |
| drifterza | perhaps that would save some space | 11:49 |
| Matias | however I would not be able to log e.g. IP addr / time for external access to public swift containers | 11:50 |
| drifterza | hmm true | 11:50 |
| Matias | personally I'd prefer to use ceph as storage for gnocchi | 11:50 |
| Matias | but that you take some time to get to work | 11:50 |
| Matias | that would* | 11:51 |
| *** askb has quit IRC | 11:51 | |
| *** schwicht has quit IRC | 11:52 | |
| *** itsuugo has quit IRC | 11:54 | |
| Matias | drifterza: now my real issue is that: http://paste.openstack.org/show/576118/ | 11:54 |
| *** hughmFLEXin has joined #openstack-ansible | 11:54 | |
| *** dragonmaster has quit IRC | 11:54 | |
| Matias | drifterza: after some minutes or a few hours working, my ceilometer-agent-notification disconnects from some rabbit queues | 11:55 |
| Matias | and data gets stuck there | 11:55 |
| Matias | have you ever seen something like this? | 11:55 |
| *** drifterza has quit IRC | 11:55 | |
| *** itsuugo has joined #openstack-ansible | 11:56 | |
| *** hughmFLEXin has quit IRC | 11:59 | |
| *** psilvad has joined #openstack-ansible | 12:03 | |
| odyssey4me | Fenuks yep, the minor upgrade process is documented here: http://docs.openstack.org/developer/openstack-ansible/mitaka/install-guide/app-minorupgrade.html | 12:03 |
| odyssey4me | drifterza yep, around now - was just out for a lunch break | 12:05 |
| *** schwicht has joined #openstack-ansible | 12:06 | |
| *** drifterza has joined #openstack-ansible | 12:07 | |
| Matias | drifterza: now my real issue is that: http://paste.openstack.org/show/576118/ | 12:07 |
| Matias | drifterza: after some minutes or a few hours working, my ceilometer-agent-notification disconnects from some rabbit queues, and data gets stuck there | 12:07 |
| Matias | drifterza: have you ever seen something like this? | 12:07 |
| *** asettle has joined #openstack-ansible | 12:07 | |
| drifterza | I haven't no, but that sounds like fd limit on rabbitmq | 12:09 |
| drifterza | or it could be the middleware | 12:09 |
| *** thorst_ has joined #openstack-ansible | 12:10 | |
| odyssey4me | drifterza Matias yeah, that's very possibly an issue with rabbit - if it runs out of memory, or fd's, or disk space, then it'll block all queue processing until they clear | 12:10 |
| *** gouthamr has joined #openstack-ansible | 12:10 | |
| Matias | drifterza: odyssey4me: http://i.imgur.com/DYNW5Hk.png | 12:10 |
| Matias | fds were an issue before we detected it, but it is now fixed by my preliminary PR | 12:10 |
| odyssey4me | hmm, ok - that's odd | 12:11 |
| drifterza | does that solve the hangup of ceilometer-agent-notification ? | 12:11 |
| Matias | not completely.. see, before the fd fix, I barely got it to start working | 12:11 |
| Matias | now it works for some minutes or even a few hours before getting stuck | 12:11 |
| drifterza | and a HUP of the service fixes it ? | 12:12 |
| Matias | restarting it fixes it until it gets stuck again | 12:12 |
| drifterza | strace it when it happens | 12:13 |
| drifterza | maybe we can get clues from that | 12:13 |
| Matias | other services in the same container, e.g. the central ceilometer-polling, don't have any issues with rabbit | 12:13 |
| Matias | hmm let's try, it is stuck right now | 12:14 |
| drifterza | I think you can do strace inside lxc | 12:14 |
| drifterza | I know docker blocks that | 12:14 |
| drifterza | odyssey4me, you still there | 12:15 |
| *** psilvad has quit IRC | 12:15 | |
| odyssey4me | drifterza yep | 12:16 |
| *** schwicht has quit IRC | 12:16 | |
| drifterza | So still trying to get hold of that upstream dev, no luck so far. I've updated my repo to include the requirements for gateing. How can I commit a PR so that the CI builds ? | 12:17 |
| drifterza | if I don't have a repo that is | 12:17 |
| drifterza | im talking about prometheus | 12:18 |
| evrardjp | odyssey4me: could you help me merge this? https://review.openstack.org/#/c/370046/ | 12:18 |
| evrardjp | andymccr: ^ | 12:19 |
| Matias | drifterza: select(0, NULL, NULL, NULL, {999993005, 746148} | 12:19 |
| Matias | just endless select on a null fd set | 12:19 |
| Matias | maybe that is the problem | 12:19 |
| Matias | look at this timeout | 12:19 |
| drifterza | hmm so its getting no feeds | 12:19 |
| *** psilvad has joined #openstack-ansible | 12:20 | |
| Matias | how is it supposed to return from the syscall in a reasonable time? | 12:20 |
| Matias | 999993005 is in seconds ,right? | 12:20 |
| Matias | with nfds=0, it is effectively a sleep | 12:20 |
| mgariepy | good morning everyone | 12:21 |
| Matias | good morning | 12:21 |
| drifterza | strace -c >/dev/null ls | 12:22 |
| Matias | hm? | 12:22 |
| drifterza | can you do that, it should give you a list of syscalls | 12:22 |
| drifterza | and there time | 12:22 |
| drifterza | their* | 12:22 |
| Matias | should I do that in the ceilometer-agent-notification pid? | 12:23 |
| drifterza | not on the pid, in the container | 12:23 |
| Matias | ah, you want to strace ls, not ceilometer | 12:23 |
| Matias | ok | 12:23 |
| drifterza | yes | 12:23 |
| drifterza | want to see which syscall is hanging | 12:23 |
| Matias | nothing weird: http://paste.openstack.org/show/576125/ | 12:24 |
| *** schwicht has joined #openstack-ansible | 12:24 | |
| odyssey4me | hughsaunders if you have a moment, can you peek at https://review.openstack.org/370046 ? | 12:24 |
| Matias | my hypothesis is that something in ampq or kombu lib is setting a huge timeout for reconnecting to rabbit in case of error, which is never reached | 12:25 |
| Matias | this does not explain why the error occurred in the first place, but such a timeout is asking to make the service unreliable | 12:25 |
| *** psilvad has quit IRC | 12:25 | |
| *** psilvad has joined #openstack-ansible | 12:25 | |
| drifterza | hmm | 12:25 |
| *** karimb has joined #openstack-ansible | 12:29 | |
| *** psilvad has quit IRC | 12:29 | |
| andymccr | evrardjp: https://review.openstack.org/#/c/369547/ | 12:31 |
| andymccr | similar for swift https://review.openstack.org/#/c/369353/ (although there is no cool way to get that var from a running cluster) | 12:31 |
| *** itsuugo has quit IRC | 12:32 | |
| *** itsuugo has joined #openstack-ansible | 12:33 | |
| *** d3n14l_ has joined #openstack-ansible | 12:34 | |
| *** d3n14l_ has quit IRC | 12:34 | |
| *** psilvad has joined #openstack-ansible | 12:35 | |
| *** itsuugo has quit IRC | 12:38 | |
| * hughsaunders is too slow at peeking | 12:39 | |
| *** itsuugo has joined #openstack-ansible | 12:39 | |
| odyssey4me | andymccr comment in the swift patch | 12:39 |
| andymccr | odyssey4me: cool sounds good - will fix | 12:40 |
| *** schwicht has quit IRC | 12:40 | |
| *** d3n14l_ has joined #openstack-ansible | 12:42 | |
| *** d3n14l_ has left #openstack-ansible | 12:42 | |
| *** d3n14l has joined #openstack-ansible | 12:42 | |
| *** spotz_zzz is now known as spotz | 12:42 | |
| odyssey4me | andymccr the only downside to putting that check into the role is that if that condition is ever hit, many things have already happened to the environment to change it - so it'll fail an upgrade or deployment after all sorts of changes have happened | 12:43 |
| odyssey4me | I do think, at some point, we need to figure out a way to do pre-requisite checks. | 12:44 |
| andymccr | odyssey4me: yeah thats true - i think it also opens it up to some other checks we could add in | 12:44 |
| andymccr | the galera role does a few more, so i think other roles could do similar things for values that should be immutable (effectively) | 12:44 |
| *** Fenuks has quit IRC | 12:44 | |
| odyssey4me | perhaps a way to do it would be to have a pre-requisite playbook which executes a the 'pre-requisite checks' for all the roles without changing anything. | 12:44 |
| *** schwicht has joined #openstack-ansible | 12:44 | |
| *** jamesdenton has joined #openstack-ansible | 12:44 | |
| odyssey4me | so basically each role has a pre-requisite check lifecycle phase which is activated by a lifecycle var (similar to the swift setup vs swift sync var) | 12:45 |
| odyssey4me | I guess that could be added as part of https://review.openstack.org/346038 | 12:45 |
| *** d3n14l has left #openstack-ansible | 12:46 | |
| *** d3n14l1 has joined #openstack-ansible | 12:46 | |
| openstackgerrit | Andy McCrae proposed openstack/openstack-ansible-os_swift: Add check for swift_hash_path_ variables https://review.openstack.org/369353 | 12:47 |
| *** schwicht has quit IRC | 12:49 | |
| *** d3n14l1 has quit IRC | 12:49 | |
| *** d3n14l has joined #openstack-ansible | 12:50 | |
| *** itsuugo has quit IRC | 12:50 | |
| *** itsuugo has joined #openstack-ansible | 12:52 | |
| *** jheroux has joined #openstack-ansible | 12:53 | |
| openstackgerrit | Merged openstack/openstack-ansible-os_neutron: Ensure the nets, subnets and ports are present https://review.openstack.org/370046 | 12:54 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-os_neutron: Update paste, policy and rootwrap configurations 2016-09-08 https://review.openstack.org/367521 | 12:54 |
| *** kylek3h has joined #openstack-ansible | 12:55 | |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-os_neutron: Remove Calico qemu configuration https://review.openstack.org/368286 | 12:55 |
| *** drifterza has quit IRC | 13:04 | |
| *** schwicht has joined #openstack-ansible | 13:04 | |
| *** afred312 has joined #openstack-ansible | 13:04 | |
| *** d3n14l has left #openstack-ansible | 13:05 | |
| openstackgerrit | Andy McCrae proposed openstack/openstack-ansible-os_swift: Add check for swift_hash_path_ variables https://review.openstack.org/369353 | 13:06 |
| openstackgerrit | Merged openstack/openstack-ansible-openstack_openrc: Add alias for OS_ENDPOINT_TYPE https://review.openstack.org/370009 | 13:07 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible: Update SHA for openstack_openrc role https://review.openstack.org/370109 | 13:13 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible: Update SHA for openstack_openrc role https://review.openstack.org/370109 | 13:13 |
| odyssey4me | andymccr evrardjp ^ that should do it | 13:13 |
| *** afred312 has quit IRC | 13:14 | |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible: Update SHA for openstack_openrc role https://review.openstack.org/370109 | 13:14 |
| openstackgerrit | Marc Gariépy proposed openstack/openstack-ansible-os_keystone: Remove requiretty for sudo on centos https://review.openstack.org/368850 | 13:19 |
| openstackgerrit | Merged openstack/openstack-ansible-galera_server: Prevent galera_cluster_name from changing. https://review.openstack.org/369547 | 13:23 |
| *** hblah has joined #openstack-ansible | 13:23 | |
| *** poopcat has quit IRC | 13:26 | |
| openstackgerrit | Travis Truman (automagically) proposed openstack/openstack-ansible-os_aodh: Ansible 2.1.1 role testing https://review.openstack.org/353563 | 13:30 |
| *** KLevenstein has joined #openstack-ansible | 13:37 | |
| openstackgerrit | Travis Truman (automagically) proposed openstack/openstack-ansible-os_glance: Ansible 2.1.1 role testing https://review.openstack.org/353589 | 13:37 |
| openstackgerrit | Travis Truman (automagically) proposed openstack/openstack-ansible-os_ceilometer: Ansible 2.1.1 role testing https://review.openstack.org/353561 | 13:38 |
| *** TxGirlGeek has joined #openstack-ansible | 13:39 | |
| *** itsuugo has quit IRC | 13:42 | |
| *** markvoelker has joined #openstack-ansible | 13:43 | |
| *** itsuugo has joined #openstack-ansible | 13:44 | |
| *** poopcat has joined #openstack-ansible | 13:44 | |
| *** tomjosekal has quit IRC | 13:48 | |
| openstackgerrit | Andy McCrae proposed openstack/openstack-ansible-galera_server: Prevent galera_cluster_name from changing. https://review.openstack.org/370133 | 13:49 |
| Matias | stevelle: got again the stuck queues from yesterday, now for another reason | 13:49 |
| Matias | stevelle: it looks like the py-amqp is unable to handle rates in the order of 100 messages per second without blowing itself up | 13:50 |
| *** itsuugo has quit IRC | 13:50 | |
| Matias | stevelle: that when using ssl | 13:51 |
| *** itsuugo has joined #openstack-ansible | 13:52 | |
| Matias | odyssey4me: I remember vaguely about ssl being disabled in OSA's rabbitmq and reenabled some days ago, does it had something to do with this? | 13:53 |
| *** afred312 has joined #openstack-ansible | 13:56 | |
| *** sdake has joined #openstack-ansible | 13:58 | |
| *** hj-hpe has joined #openstack-ansible | 13:59 | |
| mgariepy | odyssey4me, I would like to tests https://review.openstack.org/#/c/320542/ gate-openstack-ansible-os_glance-ansible-func-centos-7-nv with https://review.openstack.org/#/c/368850/ is is possible ? | 14:02 |
| odyssey4me | mgariepy not right now - you'd have ot test it yourself | 14:04 |
| mgariepy | ok how do I do that ? | 14:04 |
| odyssey4me | mgariepy build a centos-7 instance/vm, then pull down the repo | 14:05 |
| *** hblah has quit IRC | 14:05 | |
| odyssey4me | then look at the contents of run_tests.sh and tox.ini and you'll see how the test executes | 14:05 |
| odyssey4me | execute the steps one by one, and once you have the roles cloned down, put the edit from the other role into the role clone | 14:05 |
| odyssey4me | then continue the execution | 14:06 |
| odyssey4me | it's painful :/ | 14:06 |
| mgariepy | ok, i'll wait for mhayden input on https://review.openstack.org/#/c/368850/ then | 14:09 |
| mgariepy | it will be easier haha | 14:09 |
| * mhayden ganders | 14:09 | |
| cloudnull | mornings | 14:10 |
| mgariepy | the sudoers stuff might need to be moved in the keystone role tho. | 14:10 |
| shasha_tavil | morning everyone | 14:10 |
| openstackgerrit | Merged openstack/openstack-ansible-os_swift: Add check for swift_hash_path_ variables https://review.openstack.org/369353 | 14:11 |
| mhayden | mgariepy: i'm +2 on that | 14:12 |
| *** jperry has joined #openstack-ansible | 14:12 | |
| *** kstev has quit IRC | 14:13 | |
| openstackgerrit | Merged openstack/openstack-ansible-repo_server: Set permissions on the openstackgit directory https://review.openstack.org/368997 | 14:14 |
| *** spedione|AWAY is now known as spedione | 14:16 | |
| openstackgerrit | Jean-Philippe Evrard proposed openstack/openstack-ansible-os_neutron: Add container logs to the build https://review.openstack.org/370150 | 14:17 |
| *** rk4n has joined #openstack-ansible | 14:17 | |
| openstackgerrit | Jean-Philippe Evrard proposed openstack/openstack-ansible-os_neutron: Add container logs to the build https://review.openstack.org/370150 | 14:18 |
| evrardjp | could this one get some votes ^ ? | 14:19 |
| openstackgerrit | Travis Truman (automagically) proposed openstack/openstack-ansible-lxc_container_create: Address ansible_ssh_* var deprecation https://review.openstack.org/370156 | 14:21 |
| *** kstev has joined #openstack-ansible | 14:21 | |
| *** rk4n has quit IRC | 14:21 | |
| odyssey4me | evrardjp sure, once we see that it does what it says on the tin :p | 14:22 |
| odyssey4me | automagically stevelle I saw in the IRC logs something about SSL connectivity issues relating to SSLv3 | 14:23 |
| automagically | For Rabbit? | 14:23 |
| odyssey4me | and a suggestion was made to implement https://github.com/openstack/oslo.messaging/blob/master/oslo_messaging/_drivers/impl_rabbit.py#L56 as an option in all our configs, with the value of TLSv1_2 or osmething | 14:24 |
| odyssey4me | *something | 14:24 |
| odyssey4me | automagically yeah | 14:24 |
| odyssey4me | lemme try and find the log on eavesdrop | 14:24 |
| stevelle | odyssey4me: the oslo.messaging driver for rabbit defaults to SSLv3, yes | 14:24 |
| automagically | Yeah, I believe it was Matias talking about issues with message consumption just dying after awhile | 14:24 |
| odyssey4me | http://eavesdrop.openstack.org/irclogs/%23openstack-ansible/%23openstack-ansible.2016-09-14.log.html#t2016-09-14T01:27:53 | 14:25 |
| stevelle | that too, automagically. either way I don't think we want to use SSLv3 | 14:25 |
| Matias | I'm not sure if it defaults to SSLv3 or if the log message is only some artifact of the connection failure | 14:25 |
| automagically | Ah, okay, thanks for digging up the log | 14:26 |
| odyssey4me | so it appears that python is defaulting to SSLv3 | 14:26 |
| cloudnull | ++ to change to TLS1.* | 14:26 |
| odyssey4me | and we need to override that | 14:26 |
| * automagically putting it on my todo list | 14:26 | |
| odyssey4me | and this will likely have to be done across all roles | 14:26 |
| * automagically which is quite long currently | 14:26 | |
| odyssey4me | so I guess we need a namespaced var per role, and a group var global var to switch it | 14:26 |
| stevelle | main thing automagically is making sure we raise visibility and decide if we want to address it | 14:26 |
| odyssey4me | ok, best thing right now is to raise a bug | 14:27 |
| mgariepy | odyssey4me, mhayden for the sudo stuff, we might be better putting it on in main.yml as if the fernet key stuff is skipped, the next tasks that need to sudo will break. | 14:27 |
| automagically | stevelle: Do we have enough information for a bug | 14:27 |
| stevelle | automagically: probably | 14:27 |
| automagically | odyssey4me: ++ | 14:27 |
| odyssey4me | Matias could you register a bug please? | 14:27 |
| odyssey4me | at the very least we can register it as a known issue and include the config override to address it | 14:27 |
| evrardjp | so, question (for the tests repo), do you see a reason why the bind mounts should be defined globally vs locally (i.e. dependent of the role that will be tested) ? | 14:28 |
| odyssey4me | that will certainly be q cuick and easy way to address it for now, until we get the time to automate it in code | 14:28 |
| * lbragstad lurks at the mention of fernet | 14:28 | |
| Matias | i'd first inspect the ssl negotiation using tcpdump to see if it is really SSLv3 | 14:28 |
| odyssey4me | Matias for now, go ahead and register the bug with what we know now - you can add to the bug any further triage that you do | 14:28 |
| Matias | ok | 14:29 |
| odyssey4me | that at least ensures that it hits our bug triage process and is tracked in some way other than a passing conversation | 14:29 |
| Matias | well, there is a bug surely, py-amqp is unstable like hell with ssl enabled in the ceilometer containers | 14:29 |
| Matias | but i'm not sure if it is really SSLv3 or not | 14:29 |
| Matias | the log message seems to say so, but the python bugtracker says SSLv3 was discontinued in py2.7 | 14:30 |
| odyssey4me | mgariepy uh? which review are you talking about? | 14:30 |
| odyssey4me | Matias SSLv3 was discontinued everywhere | 14:30 |
| openstackgerrit | Nish Patwa(IRC: nishpatwa) proposed openstack/openstack-ansible: [install-guide] Merge install guide configure content into a single page https://review.openstack.org/369665 | 14:30 |
| mgariepy | https://review.openstack.org/#/c/368850/ | 14:30 |
| Matias | we digged a little in kombu and py-amqp code yesterday, and it seems to use py2.7 default settings for ssl if you don't specify anything | 14:32 |
| openstackgerrit | Travis Truman (automagically) proposed openstack/openstack-ansible-pip_install: Ansible 2.1.1 role testing https://review.openstack.org/370170 | 14:32 |
| openstackgerrit | Nish Patwa(IRC: nishpatwa) proposed openstack/openstack-ansible: [install-guide] Merge install guide configure content into a single page https://review.openstack.org/369665 | 14:33 |
| Matias | i'll open the bug after lunch, in a few hours | 14:33 |
| Matias | bbl | 14:33 |
| automagically | Thanks Matias | 14:33 |
| Matias | s/hours/minutes/ | 14:33 |
| mgariepy | http://paste.openstack.org/show/576193/ | 14:33 |
| mgariepy | odyssey4me, mhayden ^^ | 14:33 |
| odyssey4me | mgariepy yes, if there are other elements which require the use of sudo then the sudoers file drop should be done much earlier and unconditionally | 14:33 |
| odyssey4me | mgariepy considering that both fernet and credentials make use of it, and I think the db sync also does - this needs to be unconditionally done | 14:34 |
| *** spotz is now known as spotz_zzz | 14:34 | |
| cloudnull | ++ | 14:35 |
| odyssey4me | mgariepy part of keystone_post_install perhaps? | 14:35 |
| cloudnull | a little late to this convo but cant the sudoers things be done in the lxc_container_create template? | 14:35 |
| mgariepy | keystone_pre_ ? | 14:35 |
| openstackgerrit | Marc Gariépy proposed openstack/openstack-ansible-os_keystone: Remove requiretty for sudo on centos https://review.openstack.org/368850 | 14:36 |
| odyssey4me | if we want to apply it to all containers, then we should rather do it in the cache creation in the lxc_hosts role | 14:36 |
| mgariepy | i can resubmit with post if you like | 14:36 |
| cloudnull | that's what i mean odyssey4me, in the lxc_hosts role. | 14:36 |
| cloudnull | not container create. | 14:36 |
| odyssey4me | cloudnull the question is whether this is only needed by some services, or all of them | 14:37 |
| cloudnull | ubuntu does not have requiretty in their default config | 14:37 |
| cloudnull | only cent does | 14:37 |
| odyssey4me | also, to be fair, if the role is executed against an on-metal host then it needs this config | 14:37 |
| openstackgerrit | Merged openstack/openstack-ansible-pip_install: Make pip_install honour pip_install_options https://review.openstack.org/369121 | 14:37 |
| cloudnull | so we could just make one standard template and use that across the distros | 14:38 |
| cloudnull | from within the container that is | 14:38 |
| automagically | Quick and easy review for any cores who have a minute: https://review.openstack.org/#/c/370156/ | 14:38 |
| odyssey4me | cloudnull for the moment I'm happy as-is - this is surgical, and security changes should be as minimal and surgical as possible | 14:39 |
| automagically | ty cloudnull | 14:39 |
| automagically | Amen odyssey4me | 14:39 |
| odyssey4me | when it gets out of hand and becomes something done all over the place, we can reconsider | 14:39 |
| cloudnull | ok. | 14:39 |
| mgariepy | is keystone_pre_install with the user creation stuff ok ? | 14:39 |
| odyssey4me | yeah, I think pre or post install will do just fine | 14:40 |
| odyssey4me | looking through install now to verify that statement properly | 14:40 |
| *** jamesden_ has joined #openstack-ansible | 14:41 | |
| *** jperry has quit IRC | 14:41 | |
| cloudnull | requiretty is not really a security change even though its in the sudoers file. https://bugzilla.redhat.com/show_bug.cgi?id=1020147#c7 -- from RHEL folks. it's likely this is just getting in the container due to how the base image was created. | 14:41 |
| openstack | bugzilla.redhat.com bug 1020147 in sudo "default requiretty is problematic and breaks valid usage" [High,Closed: rawhide] - Assigned to dkopecek | 14:41 |
| cloudnull | IE https://bugzilla.redhat.com/show_bug.cgi?id=1020147#c5 | 14:42 |
| odyssey4me | cloudnull yeah, but we need to cater for on_metal deployments too - so changing the container doesn't really help much | 14:42 |
| mgariepy | most of the os-* will need sudoers file. | 14:42 |
| odyssey4me | it can only help by changing the task from an implementation to a verification | 14:42 |
| cloudnull | what mgariepy said. | 14:42 |
| cloudnull | we can add the surgical bits to the openstack-hosts role | 14:43 |
| cloudnull | which will then cater to onmetal | 14:43 |
| cloudnull | but i get the point of it's good enough for now. | 14:43 |
| cloudnull | just playing devils advocate. | 14:43 |
| *** jamesde__ has joined #openstack-ansible | 14:43 | |
| odyssey4me | fair enough - so two changes - one to openstack_hosts, the other to lxc_hosts... only for redhat though... and all we're doing is disabling requiretty for root? | 14:44 |
| mgariepy | i would rather not having to add the sudoers.j2 and tasks on all the roles. and some tests needs it as well. | 14:44 |
| odyssey4me | mhayden need your input here, automagically jmccrory thoughts? | 14:44 |
| * mhayden is reading now | 14:45 | |
| odyssey4me | one question, of course, would then be - what do we name the sudoers file? | 14:45 |
| odyssey4me | or do we change the implementation to be different - perhaps a lineinfile change or something | 14:45 |
| mgariepy | openstack-ansible ? | 14:45 |
| *** jamesden_ has quit IRC | 14:45 | |
| cloudnull | sudoers | 14:46 |
| mgariepy | as in /etc/sudoers.d/openstack-ansible ? | 14:46 |
| cloudnull | odyssey4me: that said, we could add -t to our base ssh-command in https://github.com/openstack/openstack-ansible/blob/master/playbooks/inventory/group_vars/all.yml#L411 | 14:46 |
| cloudnull | which would then always set a tty | 14:46 |
| sdake | is jessie around? | 14:47 |
| cloudnull | ohai sdake. | 14:48 |
| cloudnull | jessie == odyssey4me | 14:48 |
| mhayden | cloudnull: ah good point on the -t | 14:48 |
| cloudnull | lots of ways to skin this cat. | 14:49 |
| sdake | hey cloudnull | 14:51 |
| openstackgerrit | Merged openstack/openstack-ansible-lxc_container_create: Address ansible_ssh_* var deprecation https://review.openstack.org/370156 | 14:51 |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible-lxc_container_create: Change the container prep using more intelligent commands https://review.openstack.org/354120 | 14:51 |
| *** jwitko has joined #openstack-ansible | 14:53 | |
| odyssey4me | cloudnull yeah, that's an option - although I don't think that will help the rsync jobs for key sync... although those should have their own requiretty/sudo bits | 14:54 |
| *** itsuugo has quit IRC | 14:54 | |
| cloudnull | yea, we'd have to add -t to all the things that use ssh outside of ansible | 14:55 |
| cloudnull | so that's a less supportable solution overall. | 14:55 |
| cloudnull | IMO, removing it from the container and adding a lineinfile option for openstack_hosts is the best way forward. | 14:55 |
| * cloudnull my 2cents | 14:55 | |
| *** itsuugo has joined #openstack-ansible | 14:56 | |
| mhayden | mgariepy: what are your thoughts on using -t as cloudnull suggested? | 14:57 |
| openstackgerrit | Nish Patwa(IRC: nishpatwa) proposed openstack/openstack-ansible: [DOCS] Added HAProxy example to Production environment document. https://review.openstack.org/370187 | 14:57 |
| *** rk4n has joined #openstack-ansible | 14:58 | |
| cloudnull | nishpatwa_: ^ nice! | 14:58 |
| cloudnull | asettle: rromans: ^ | 14:58 |
| nishpatwa_ | Thanks..:) cloudnull | 14:58 |
| *** jamesde__ has quit IRC | 14:58 | |
| mgariepy | I think any solution that will fix all the roles at onces it better than individual patches. | 14:58 |
| odyssey4me | mgariepy ++ | 14:59 |
| openstackgerrit | Nish Patwa(IRC: nishpatwa) proposed openstack/openstack-ansible: [docs] Merge install guide configure content into a single page https://review.openstack.org/369665 | 14:59 |
| odyssey4me | mgariepy so I think the key sync jobs which run as non-root will need a special sudoers file regardless, and those should be per rsync implementation | 14:59 |
| odyssey4me | but if we can implement something that works for ansible in a simple, clean way - then let's do it | 15:00 |
| cloudnull | mgariepy: ++ | 15:00 |
| mgariepy | will the -t option fix tests that need becomes as well ? | 15:00 |
| odyssey4me | mgariepy no, I don't think it will :/ | 15:02 |
| *** TxGirlGeek has quit IRC | 15:02 | |
| *** BjoernT has joined #openstack-ansible | 15:03 | |
| odyssey4me | cloudnull the dynamic inventory is giving me grief - I need some advise on how to resolve this | 15:04 |
| *** TxGirlGeek has joined #openstack-ansible | 15:05 | |
| odyssey4me | I've made changes in https://review.openstack.org/369485 to make it easier to designate each os-infra bit to live on a host instead of forcing the use of affinities... but now I'm seeing that all the groups I've broken out have the host included as a target in groups like 'horizon_all' :/ ... are you able to see how to fix this at a glance? | 15:05 |
| *** michaelgugino has joined #openstack-ansible | 15:06 | |
| *** david-lyle_ has joined #openstack-ansible | 15:06 | |
| openstackgerrit | Merged openstack/openstack-ansible: Update SHA for openstack_openrc role https://review.openstack.org/370109 | 15:06 |
| *** david-lyle_ has quit IRC | 15:06 | |
| odyssey4me | basically I want to be able to put 'horizon_hosts' into the openstack_user_config/conf.d and it'll build a horizon container on the designated host | 15:06 |
| odyssey4me | cores - we need a second review of https://review.openstack.org/369784 | 15:07 |
| openstackgerrit | Marc Gariépy proposed openstack/openstack-ansible: Add -t option to create a pseudo TTY https://review.openstack.org/370197 | 15:09 |
| *** TxGirlGeek has quit IRC | 15:09 | |
| *** weezS has joined #openstack-ansible | 15:09 | |
| BjoernT | Folks just a heads up. If you are not using galera-server 10.0.27+maria-1~trusty or higher you run into nasty SST race conditions per https://jira.mariadb.org/browse/MDEV-9423, https://jira.mariadb.org/browse/MDEV-10497 and https://bugs.launchpad.net/percona-xtrabackup/+bug/1570560. | 15:09 |
| openstack | Launchpad bug 1570560 in Percona XtraBackup 2.4 "xtrabackup-v2 SST is broken on MariaDB 10.1/Galera w/ binary logs enabled" [High,New] - Assigned to Sergei Glushchenko (sergei.glushchenko) | 15:09 |
| *** TxGirlGeek has joined #openstack-ansible | 15:09 | |
| *** weezS has quit IRC | 15:10 | |
| *** hughmFLEXin has joined #openstack-ansible | 15:10 | |
| cloudnull | odyssey4me: il look at it | 15:12 |
| cloudnull | *i'll | 15:12 |
| cloudnull | BjoernT: good lookin out. | 15:12 |
| openstackgerrit | Michael Gugino proposed openstack/openstack-ansible-os_nova: [DOCS] Update docs for mixed lxd hypervisors https://review.openstack.org/365098 | 15:13 |
| BjoernT | yeah I pulled my hair out yesterday, took me like 5 attempts to get a database restored which only worked if I did manual flush prior to SST | 15:13 |
| hughmFLEXin | cloudnull: morning | 15:13 |
| cloudnull | hughmFLEXin: o/ | 15:13 |
| cloudnull | and an upgrade to the latest 10.0.x fixed it | 15:14 |
| cloudnull | ? | 15:14 |
| openstackgerrit | Merged openstack/openstack-ansible-memcached_server: Lower verbosity on memcached server https://review.openstack.org/369784 | 15:15 |
| cloudnull | Matias: ^ | 15:15 |
| BjoernT | I hope it does, I had manually fixed the situation with manual flushes. https://jira.mariadb.org/browse/MDEV-10497 seems to address this issue quite well. | 15:16 |
| openstackgerrit | Travis Truman (automagically) proposed openstack/openstack-ansible-os_rally: Remove duplicate container prep from role test https://review.openstack.org/370201 | 15:16 |
| *** rk4n has quit IRC | 15:17 | |
| *** TxGirlGeek has quit IRC | 15:20 | |
| *** TxGirlGeek has joined #openstack-ansible | 15:20 | |
| hughmFLEXin | cloudnull: does OSA have test plays to validate the deployment? | 15:21 |
| cloudnull | we have tempest | 15:21 |
| cloudnull | which can be used to validate a deployment | 15:22 |
| *** jperry has joined #openstack-ansible | 15:22 | |
| cloudnull | but tempest isn't great as it'll leave crufty bits around. | 15:22 |
| hughmFLEXin | gotcha... | 15:22 |
| cloudnull | which is just a tempest-ism not really an issue with the role. | 15:22 |
| hughmFLEXin | so most of my previous deploys use flat network for the floating (external) network. What is segmentation id? | 15:24 |
| *** TxGirlGeek has quit IRC | 15:24 | |
| hughmFLEXin | for some reason i went with vxlan... | 15:24 |
| cloudnull | segmentation_id == vlan tag | 15:24 |
| *** TxGirlGeek has joined #openstack-ansible | 15:24 | |
| odyssey4me | cloudnull palendae ok, we seem to have an inventory bug here - I'll fix up the patch, and you'll see what works and what doesn't | 15:24 |
| cloudnull | we support vxlan, flat, and vlan all at the same time. | 15:25 |
| cloudnull | so in general we use vlan for provider networks and vxlan for tenant networks | 15:25 |
| hughmFLEXin | ah | 15:26 |
| cloudnull | so admin creates a couple provider networks for cluster use, maybe attach them to a router if thats needed, and share the networks so all tenant can attach to them. | 15:26 |
| cloudnull | then tenants create vxlan networks and routers which attach to the shared provider networks giving the tenant access to floating IPs | 15:27 |
| openstackgerrit | Major Hayden proposed openstack/openstack-ansible-security: [Docs] More cleanup https://review.openstack.org/370213 | 15:27 |
| *** adrian_otto has joined #openstack-ansible | 15:27 | |
| cloudnull | that's a fairly basic, but common setup. | 15:27 |
| *** TxGirlGeek has quit IRC | 15:27 | |
| *** rk4n has joined #openstack-ansible | 15:27 | |
| *** TxGirlGeek has joined #openstack-ansible | 15:28 | |
| hughmFLEXin | im spoiled by mirantis fuel doing all the networking for me... | 15:28 |
| palendae | odyssey4me, You're referring to https://review.openstack.org/#/c/369485/ ? | 15:29 |
| odyssey4me | palendae yep | 15:30 |
| odyssey4me | so the issue now is that with the patch in the current state, the host ends up in all the groups, for example horizon_all | 15:31 |
| palendae | 'the host' being aio1? | 15:31 |
| odyssey4me | palendae yep | 15:31 |
| *** TxGirlGeek has quit IRC | 15:32 | |
| *** TxGirlGeek has joined #openstack-ansible | 15:32 | |
| palendae | In the horizon scenario, wouldn't you want it there? | 15:33 |
| odyssey4me | if I change 'horizon_containers', 'horizon_hosts' in the end.d entry to 'dashboard_containers' and 'dashboard_hosts' and obviously fix the conf.d entry then it all works the right way | 15:34 |
| odyssey4me | for example, this is a working inventory: https://gist.github.com/evrardjp/165e00f19fbb553fbb3221d1d622ddc9#file-openstack_inventory-json-L1223-L1233 | 15:34 |
| cloudnull | hughmFLEXin: yea. we're all about choose your own misadventure. :) | 15:34 |
| logan- | maybe we should be using the 'service' group instead of the 'service_all' group | 15:34 |
| cloudnull | hughmFLEXin: but maybe this will help https://github.com/os-cloud/osic-ref-impl/blob/master/post-deployment-setup.sh | 15:35 |
| logan- | it seems the only difference is the physical host inclusion in _all? is that a bug or intentional for some reason? :) | 15:35 |
| cloudnull | its a simple script to create networks, secgroup rules, and upload images | 15:35 |
| *** TxGirlGeek has quit IRC | 15:36 | |
| odyssey4me | palendae this is the output from the horizon config in that patch: http://pastebin.com/e15FxcCA | 15:36 |
| *** TxGirlGeek has joined #openstack-ansible | 15:37 | |
| hughmFLEXin | cloudnull: ill take a look thanks | 15:37 |
| odyssey4me | we target horizon_all in our playbooks... that gets both the horizon container and the container host, which from the keystone output in a working inventory you can see is not what it should be | 15:37 |
| odyssey4me | and it seems that for some reason the name of the group is what makes it work or not work, so there is some sort of bad assumption happening here | 15:38 |
| palendae | Sure | 15:38 |
| logan- | I observed that same behavior with unbound yesterday odyssey4me so yeah I'm thinking the same thing.. it seems like when the service name is the same as the container group name it does that. | 15:38 |
| palendae | Oh, right | 15:38 |
| palendae | Yes. | 15:38 |
| palendae | evrardjp had this problem with haproxy | 15:38 |
| openstackgerrit | Merged openstack/openstack-ansible-ops: Updated the readme file to add missing commands. https://review.openstack.org/369548 | 15:39 |
| *** admin0 has quit IRC | 15:39 | |
| hughmFLEXin | cloudnull: qq, is the "range" in user_config for vlan and vxlan require physical switch config for those tags, or are those container-based network tags only | 15:39 |
| odyssey4me | hmm, of course - there's a group called 'horizon' (ie the service name) | 15:39 |
| odyssey4me | that's from the component_skel | 15:40 |
| palendae | We ran into this before when the service name is name the same as container group | 15:40 |
| *** TxGirlGeek has quit IRC | 15:40 | |
| cloudnull | hughmFLEXin: for vxlan no, those are tenant specific ranges and are only used internally. | 15:40 |
| *** pcaruana has quit IRC | 15:40 | |
| cloudnull | for vlan yes. you should have those vlans trunked | 15:41 |
| hughmFLEXin | got it | 15:41 |
| palendae | odyssey4me, The intent was to allow the services to be installed on the hosts, right? | 15:41 |
| openstackgerrit | Merged openstack/openstack-ansible-openstack_hosts: Convert role testing to use Ansible 2.1.1 https://review.openstack.org/369524 | 15:41 |
| cloudnull | if flat then you can use a native vlan if you have that setup. | 15:41 |
| evrardjp | yes I remember having that | 15:41 |
| odyssey4me | palendae so I don't really see how this has anything to do with a property in the dict | 15:42 |
| jamesdenton | does anyone know if v14 has any mechanism to create bgp_dragent.ini for the BGP agent? Or if there's a way to arbitrarily create that file w/ respective config via existing openstack-ansible mechanisms? | 15:42 |
| odyssey4me | basically if I change an entry called 'horizon_containers' to 'dashboard_containers' and 'horizon_hosts' to 'dashboard_hosts' then it suddenly applies the group memberships in the right way... | 15:42 |
| odyssey4me | if for some reason the service property is being used and '_containers' or '_hosts' appended to it, then we're being daft | 15:43 |
| hughmFLEXin | cloudnull: hah, is it too late to go back and add the flat network back into the user_config? | 15:43 |
| cloudnull | no | 15:43 |
| odyssey4me | I can see how we possibly might be grabbing part of the component skel, then appending _... to it | 15:43 |
| odyssey4me | ie a bad assumption | 15:43 |
| openstackgerrit | Major Hayden proposed openstack/openstack-ansible-ops: Remove osa-differ https://review.openstack.org/370234 | 15:44 |
| cloudnull | just add it in, rerun the lxc-container-create play | 15:44 |
| evrardjp | jamesdenton: bgp-dragent is part of osa | 15:44 |
| hughmFLEXin | cloudnull: will do... | 15:44 |
| palendae | The problem before was that using a host named 'haproxy' collided with the existing dict entry | 15:44 |
| evrardjp | it works with overrides likes the others | 15:44 |
| palendae | So this is different | 15:44 |
| palendae | The '_all' groups are implicit | 15:44 |
| evrardjp | jamesdenton: the name of the override is neutron_bgp_dragent_ini_overrides | 15:44 |
| palendae | Looks like we're also creating '_hosts' groups | 15:45 |
| jamesdenton | evrardjp ahh perfect. much thanks! And to enable the service_plugin just add a neutron_conf override? | 15:45 |
| odyssey4me | palendae yeah, so what I think might be going on here is that we're not taking the env.d entries as the source of truth - we're instead adding some assumptions about things that should be there | 15:45 |
| *** BjoernT has quit IRC | 15:45 | |
| odyssey4me | so for instance we're being forgiving about where someone forgot to add a _hosts group, so we create one | 15:46 |
| *** eon` has quit IRC | 15:47 | |
| *** eon` has joined #openstack-ansible | 15:47 | |
| evrardjp | jamesdenton: put in your variables neutron_bgp: True | 15:48 |
| jamesdenton | ahh ok | 15:48 |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible-os_aodh: Ansible 2.1.1 role testing https://review.openstack.org/353563 | 15:49 |
| *** TxGirlGeek has joined #openstack-ansible | 15:49 | |
| openstackgerrit | Merged openstack/openstack-ansible-pip_install: Ansible 2.1.1 role testing https://review.openstack.org/370170 | 15:49 |
| evrardjp | jamesdenton: simple, right? :p | 15:50 |
| jamesdenton | it is! Does only 1 agent get installed? | 15:50 |
| evrardjp | well my knowledge stops here I guess | 15:50 |
| evrardjp | just let me check | 15:50 |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible-pip_install: Add role linking to tox tests https://review.openstack.org/370241 | 15:52 |
| openstackgerrit | Merged openstack/openstack-ansible-os_neutron: Remove Calico qemu configuration https://review.openstack.org/368286 | 15:52 |
| *** BjoernT has joined #openstack-ansible | 15:52 | |
| evrardjp | jamesdenton: it looks like it's gonna be installed when neutron_bgp_dragent is in the group names ... let me check a second more | 15:53 |
| jamesdenton | k | 15:53 |
| *** TxGirlGeek has quit IRC | 15:53 | |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-os_neutron: Add functional testing for Calico https://review.openstack.org/368287 | 15:53 |
| *** TxGirlGeek has joined #openstack-ansible | 15:53 | |
| jamesdenton | evrardjp do we know if anyone has done an implementation and peered with an external router? | 15:54 |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible-repo_server: Convert role testing to use Ansible 2.1.1 https://review.openstack.org/369782 | 15:54 |
| evrardjp | jamesdenton: guess what | 15:54 |
| evrardjp | we have docs for that! | 15:54 |
| evrardjp | http://docs.openstack.org/developer/openstack-ansible-os_neutron/configure-network-services.html#bgp-dynamic-routing-service-optional | 15:54 |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible-os_ceilometer: Ansible 2.1.1 role testing https://review.openstack.org/353561 | 15:55 |
| jamesdenton | evrardjp good deal! i'll mess around with it and try and make it go. thanks for the pointers! | 15:55 |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible-os_sahara: Convert role testing to use Ansible 2.1.1 https://review.openstack.org/353612 | 15:55 |
| evrardjp | jamesdenton: well I never tried myself , but don't hesitate to ping | 15:56 |
| evrardjp | me | 15:56 |
| jamesdenton | sure | 15:56 |
| jamesdenton | thanks! | 15:56 |
| openstackgerrit | Travis Truman (automagically) proposed openstack/openstack-ansible-repo_build: Ansible 2.1.1 role testing https://review.openstack.org/370245 | 15:57 |
| evrardjp | jamesdenton: I think the doc is too simple, there is probably a patchset or two to be done there : | 15:58 |
| evrardjp | :p | 15:58 |
| jamesdenton | yeah i don't see any mention of neutron_bgp: True | 15:58 |
| jamesdenton | but i'll figure it out | 15:58 |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible-os_keystone: Add role linking to tox tests https://review.openstack.org/370246 | 15:58 |
| openstackgerrit | Alexandra Settle proposed openstack/openstack-ansible: [docs] Merge install guide configure content into a single page https://review.openstack.org/369665 | 15:59 |
| automagically | cloudnull: The role link changes are because of the ordering of items within the ANSIBLE_ROLE_PATH var correct? | 16:01 |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible-os_zaqar: Add role linking to tox tests https://review.openstack.org/370248 | 16:01 |
| automagically | i.e. there is some risk that the role could be found in the first path | 16:01 |
| openstackgerrit | Steve Lewis (stevelle) proposed openstack/openstack-ansible: Configure Ceilometer middleware for Gnocchi-Swift https://review.openstack.org/366279 | 16:01 |
| cloudnull | yes. | 16:01 |
| automagically | Nice catch, subtle problem that would cause some real issues if triggered | 16:01 |
| cloudnull | also if there's an existing clone of the role on the disk it'll use that instead | 16:01 |
| cloudnull | so yea, it's a minimal rick | 16:02 |
| cloudnull | *risk | 16:02 |
| cloudnull | but could cause some issues. | 16:02 |
| cloudnull | also some of our changes already have it | 16:02 |
| cloudnull | i was just doing an audit to make sure they all do | 16:02 |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible-os_zaqar: Add role linking to tox tests https://review.openstack.org/370248 | 16:03 |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible-os_barbican: Ansible 2.1.1 role testing https://review.openstack.org/353567 | 16:04 |
| evrardjp | jamesdenton: apparently you don't have to set the variable, it's auto set if you follow the docs! | 16:04 |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible-os_designate: Ansible 2.1.1 role testing https://review.openstack.org/353587 | 16:06 |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible-os_rally: Add role linking to tox tests https://review.openstack.org/370253 | 16:07 |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible-os_cinder: Ansible 2.1.1 update https://review.openstack.org/353569 | 16:08 |
| *** rk4n has quit IRC | 16:08 | |
| *** TxGirlGeek has quit IRC | 16:09 | |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible-os_neutron: Ansible 2.1.1 role testing https://review.openstack.org/353608 | 16:09 |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible-os_gnocchi: Ansible 2.1.1 role testing https://review.openstack.org/353590 | 16:09 |
| *** TxGirlGeek has joined #openstack-ansible | 16:09 | |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible-os_horizon: Ansible 2.1.1 role testing https://review.openstack.org/353596 | 16:10 |
| palendae | odyssey4me, https://github.com/openstack/openstack-ansible/blob/master/playbooks/inventory/dynamic_inventory.py#L323-L327 is the line that's adding entries to the _all groups | 16:10 |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible-memcached_server: Add role linking to tox tests https://review.openstack.org/370261 | 16:11 |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible-os_tempest: Ansible 2.1.x role testing https://review.openstack.org/353616 | 16:12 |
| *** galstrom_zzz is now known as galstrom | 16:12 | |
| cloudnull | so automagically so i think that covers all of them now. | 16:12 |
| *** TxGirlGeek has quit IRC | 16:13 | |
| hughmFLEXin | hmm even with flat, cant seem to create the external network | 16:13 |
| odyssey4me | palendae and I see the _all group name assumption here: https://github.com/openstack/openstack-ansible/blob/master/playbooks/inventory/dynamic_inventory.py#L294 | 16:14 |
| palendae | Right | 16:14 |
| *** TxGirlGeek has joined #openstack-ansible | 16:14 | |
| palendae | It's doing the right thing for neutron | 16:14 |
| palendae | But neutron has neutron_all and network_all | 16:14 |
| palendae | Keystone has keystone_all and identity_all | 16:14 |
| odyssey4me | yep, as long as the physical grouop names are different, it works | 16:14 |
| odyssey4me | hmm, it has both names... that tells us something | 16:15 |
| palendae | Both names? | 16:15 |
| jamesdenton | hughmFLEXin what is the error? | 16:15 |
| odyssey4me | so there must be a collision between the component/container_skel names and the physical_skel name | 16:15 |
| hughmFLEXin | jamesdenton: Error: Failed to create network admin_floating_net | 16:16 |
| palendae | Most likely | 16:16 |
| odyssey4me | this is why they have to, right now, be different | 16:16 |
| jamesdenton | hughmFLEXin very descriptive :) | 16:16 |
| hughmFLEXin | jamesdenton: i know right | 16:16 |
| palendae | Similar to what I was saying for haproxy, but differnt keys | 16:16 |
| hughmFLEXin | jamesdenton: cli doesn't help either, it throws up ssl error | 16:16 |
| jamesdenton | hughmFLEXin Ok that was from Horizon? Check the neutron server log. You ought to see the reason there | 16:17 |
| odyssey4me | palendae yeah, there I understand what was going on - because we create a name based on the prefix before the '_', and that clashed with a host name | 16:17 |
| palendae | Right | 16:17 |
| palendae | Here something else is colliding | 16:17 |
| palendae | I'm not exactly sure what though | 16:17 |
| odyssey4me | lemme write this up as a bug at least | 16:17 |
| odyssey4me | I know how to progress my patch, but it is going to require implementing some group names I'd rather do without. | 16:18 |
| jamesdenton | hughmFLEXin If you're getting SSL errors with CLI, try 'neutron --insecure <command' | 16:18 |
| odyssey4me | If we can find a solution before release, that'd be great - because then we don't have to carry the names for a cycle. | 16:18 |
| hughmFLEXin | jamesdenton: tried that too | 16:18 |
| palendae | I'll poke at it | 16:18 |
| hughmFLEXin | Connection failure that may be retried. | 16:18 |
| hughmFLEXin | END return value: 1 | 16:18 |
| odyssey4me | thanks palendae | 16:18 |
| palendae | Raises concerns about https://review.openstack.org/#/c/369599/ though | 16:19 |
| palendae | In case someone wants to name a group 'lxc_hosts' | 16:19 |
| hughmFLEXin | jamesdenton: from neutron-server log https://gist.github.com/bbyhuy/d0ed9dc6b044a8f6f4e1b2cdc20280f7 | 16:19 |
| palendae | But I don't want people having to manually specify that, either, cause it's purely redundant | 16:19 |
| jamesdenton | hughmFLEXin checking... | 16:19 |
| *** TxGirlGeek has quit IRC | 16:20 | |
| odyssey4me | palendae I would recommend adding a check for that, and erroring out if someone has that name specified in any configs | 16:20 |
| palendae | Yeah | 16:20 |
| odyssey4me | it should be a dynamic group, only - and never overwritteb | 16:20 |
| odyssey4me | *overwritten | 16:20 |
| *** TxGirlGeek has joined #openstack-ansible | 16:20 | |
| jamesdenton | hughmFLEXin that's it? It appears both Horizon and CLI are unable to properly communicate with Neutron API i guess? | 16:20 |
| jamesdenton | hughmFLEXin I assume only your public endpoints are HTTPS? Can you confirm with 'openstack endpoint list'? If so, check to see that the openrc file you're using has OS_ENDPOINT_TYPE set to 'internalURL' and not public. Does the 'net-list' command work OK? | 16:22 |
| hughmFLEXin | export OS_AUTH_URL=http://172.29.236.10:5000/v3 | 16:23 |
| jamesdenton | What about OS_ENDPOINT_TYPE | 16:23 |
| hughmFLEXin | internalURL | 16:24 |
| hughmFLEXin | openstack user list works, network list gives ssl error | 16:24 |
| openstackgerrit | Merged openstack/openstack-ansible: [docs] Merge install guide configure content into a single page https://review.openstack.org/369665 | 16:24 |
| *** karimb has quit IRC | 16:25 | |
| *** dragonmaster has joined #openstack-ansible | 16:25 | |
| *** hybridpollo has joined #openstack-ansible | 16:26 | |
| *** TxGirlGeek has quit IRC | 16:26 | |
| jamesdenton | do you mind posting that ssl error somewhere? | 16:27 |
| *** TxGirlGeek has joined #openstack-ansible | 16:27 | |
| hughmFLEXin | jamesdenton: its super descriptive....not...https://gist.github.com/bbyhuy/f3d87785a60a3be35b7786842409177f | 16:27 |
| jamesdenton | hmm, try the neutron client instead | 16:28 |
| jamesdenton | 'neutron net-list' | 16:28 |
| hughmFLEXin | ah that one actually worked | 16:29 |
| openstackgerrit | Travis Truman (automagically) proposed openstack/openstack-ansible-tests: Fix container destroy host group https://review.openstack.org/370282 | 16:29 |
| jamesdenton | try creating your external provider network with the neutron client | 16:30 |
| kylek3h | odyssey4me: Do you have time for a os_trove question? | 16:30 |
| jamesdenton | hughmFLEXin I don't know that all features have been baked into openstack client yet for networking. I, too, get the SSL error in v14 | 16:30 |
| *** TxGirlGeek has quit IRC | 16:32 | |
| *** TxGirlGeek has joined #openstack-ansible | 16:33 | |
| odyssey4me | kylek3h sure | 16:33 |
| odyssey4me | palendae https://bugs.launchpad.net/openstack-ansible/+bug/1623584 | 16:33 |
| openstack | Launchpad bug 1623584 in openstack-ansible "Dynamic inventory clash between host group name and container group name" [Undecided,New] | 16:33 |
| palendae | Thanks | 16:33 |
| odyssey4me | not sure if the title is a good description, feel free to change at will | 16:33 |
| kylek3h | odyssey4me: I started playing around it with and I'm finding a lot of issues...is it actively being developed? | 16:34 |
| odyssey4me | I've noted the sample config to replicate it, the expected output and also the workaround with its output | 16:34 |
| odyssey4me | kylek3h yeah, it has basically just been started and has the basic plumbing down | 16:34 |
| kylek3h | odyssey4me: Ok..is it for Newton or Ocata? | 16:34 |
| odyssey4me | so it's open for patches by anyone who wants to see it included in Ocata | 16:34 |
| odyssey4me | yeah, it's way too late to get it into Newton - it needs to be integrated by Ocata Milestone 2 to be released in Ocata | 16:35 |
| hughmFLEXin | jamesdenton: ah much more helpful | 16:35 |
| *** retreved has quit IRC | 16:35 | |
| hughmFLEXin | BadRequest: Invalid input for operation: Flat provider networks are disabled. | 16:35 |
| kylek3h | ok, i can throw up a few patches for what I've seen but i haven't gotten it all working in an AIO yet. | 16:35 |
| kylek3h | the patches would be baby steps | 16:35 |
| odyssey4me | kylek3h that'd be fantastic :) ideally we can get it up and working with functional testing ASAP | 16:36 |
| cloudnull | kylek3h: doit() | 16:36 |
| odyssey4me | ++ | 16:36 |
| kylek3h | sure...i'll just have to find good points to break off a review...got a lot of patches now. | 16:36 |
| kylek3h | thanks for the info. | 16:36 |
| *** TxGirlGeek has quit IRC | 16:37 | |
| *** TxGirlGeek has joined #openstack-ansible | 16:38 | |
| *** kylek3h is now known as kylek3h_away | 16:38 | |
| jamesdenton | hughmFLEXin yeah, it's probably either that the configuration wasn't laid down in ml2 config or neutron server didn't get restarted. In the ml2_conf.ini in server container you should see '[ml2]\n type_drivers = flat,vlan,vxlan' or some combination. Flat should be there. Then you would see '[ml2_type_flat]\n flat_networks = *' | 16:39 |
| jamesdenton | without that it won't work | 16:39 |
| jamesdenton | neutron server service may need to be restarted for the changes to take effect | 16:39 |
| hughmFLEXin | jamesdenton: alright let me make the changes and see what happens | 16:40 |
| hughmFLEXin | jamesdenton: so the type_drivers is there | 16:41 |
| *** Matias has quit IRC | 16:41 | |
| hughmFLEXin | but the flat_networks = is missing the asterisk | 16:41 |
| jamesdenton | ok. so you would either need to specify the wildcard or the name(s) of the provider bridge labels that have been defined | 16:41 |
| openstackgerrit | Matt Thompson proposed openstack/openstack-ansible-ceph_client: Restart OS services when symlinks are created https://review.openstack.org/370296 | 16:42 |
| *** TxGirlGeek has quit IRC | 16:42 | |
| automagically | Would like to get https://review.openstack.org/#/c/370282/ reviewed and merged as its blocking getting rabbitmq_server role testing up to Ansible 2.1.1 <— cloudnull and odyssey4me | 16:42 |
| jamesdenton | i'm sorry, i don't know the openstack-ansible overrides and bits offhand | 16:42 |
| *** TxGirlGeek has joined #openstack-ansible | 16:42 | |
| jamesdenton | hughmFLEXin also, here's the bug related to the openstack client issues: https://bugs.launchpad.net/python-openstackclient/+bug/1592368 | 16:42 |
| openstack | Launchpad bug 1592368 in python-openstackclient "Endpoint type is ignored for network commands" [Undecided,Fix released] - Assigned to Stuart McLaren (stuart-mclaren) | 16:42 |
| *** michaelgugino has quit IRC | 16:42 | |
| cloudnull | automagically: done | 16:42 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible: Implement scenario capability for AIO https://review.openstack.org/369485 | 16:42 |
| jamesdenton | it's ignoring the OS_INTERFACE override in the rc file. That override may also not be properly configured, though | 16:42 |
| hughmFLEXin | jamesdenton: would it be the net_name? or container_bridge? | 16:43 |
| jamesdenton | that's a bug for another time with OSA | 16:43 |
| jamesdenton | no, it would be something similar to what's been defined for vlan. In fact, it would be the same label by default (ie. vlan) | 16:43 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible: Implement scenario capability for AIO https://review.openstack.org/369485 | 16:44 |
| hughmFLEXin | jamesdenton: awkwardly enough vlan only shows the vlan_ranges | 16:44 |
| *** M00nr41n has joined #openstack-ansible | 16:44 | |
| jamesdenton | honestly, though, flat networks may be problematic. i have not used them, since on the compute nodes the br-vlan bridge is used and vlan subinterfaces hang off that and are placed into brq bridges. The bridge br-vlan, without tags, can['t be placed in another bridge | 16:45 |
| jamesdenton | yeah - vlan is just a name for what is effectively bond1 or whatever the actual provider interface is. That's why a wildcard will suffice. It means any provider label that is defined anywhere | 16:45 |
| jamesdenton | hughmFLEXin since you should now be able to create external network with the neutron client, try using a VLAN and see if that goes | 16:46 |
| hughmFLEXin | interesting, i tried net-create and used vlan and it worked | 16:46 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible: Implement scenario capability for AIO https://review.openstack.org/369485 | 16:46 |
| hughmFLEXin | without specifying seg_id | 16:46 |
| jamesdenton | yeah it's going to choose one from the range | 16:47 |
| *** TxGirlGeek has quit IRC | 16:47 | |
| jamesdenton | which may or may not be desirable behavior, depending on the circumstances | 16:47 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible: Implement scenario capability for AIO https://review.openstack.org/369485 | 16:47 |
| jamesdenton | It's Complicated™ | 16:47 |
| *** TxGirlGeek has joined #openstack-ansible | 16:47 | |
| hughmFLEXin | jamesdenton: hah...looked up openstack in the thesasurus, Complicated was a synonym | 16:47 |
| jamesdenton | It's flexible. :P | 16:48 |
| hughmFLEXin | jamesdenton: so the network is actually a little different than when I last spoke with you last week. When you gave me the interface file to throw everything on 1 eth | 16:48 |
| jamesdenton | oh right | 16:49 |
| hughmFLEXin | eth3 (10g) is public outbound, while eth2(10g) is everything OSA | 16:49 |
| openstackgerrit | Matt Thompson proposed openstack/openstack-ansible-ceph_client: Restart OS services when symlinks are created https://review.openstack.org/370296 | 16:49 |
| odyssey4me | automagically cloudnull the role linking will actually make role test execution a bit easier | 16:50 |
| hughmFLEXin | so really the only difference is i left eth3 static ip in the interface file and threw yours below it | 16:50 |
| odyssey4me | we can just target the role name as it normally is, instead of toxinidir | 16:50 |
| automagically | odyssey4me: Yeah, we should be able to remove the -e role_name business now | 16:50 |
| odyssey4me | we should not that somewhere actually | 16:50 |
| odyssey4me | *note | 16:50 |
| odyssey4me | somewhere in the test role | 16:51 |
| odyssey4me | ie note that the execution of the role with the role name relies on the role link mechanism being in place via tox | 16:51 |
| cloudnull | odyssey4me: ++ | 16:51 |
| automagically | LaunchPad wishlist bug? | 16:51 |
| automagically | I just opened https://bugs.launchpad.net/openstack-ansible/+bug/1623592 | 16:52 |
| openstack | Launchpad bug 1623592 in openstack-ansible "RabbitMQ upgrade test should be a scenario test" [Wishlist,New] | 16:52 |
| automagically | Cause there is a lot of cleanup stuff that we should really start tracking | 16:52 |
| openstackgerrit | Jean-Philippe Evrard proposed openstack/openstack-ansible-os_neutron: Add container logs to the build https://review.openstack.org/370150 | 16:53 |
| openstackgerrit | Merged openstack/openstack-ansible-os_neutron: Update paste, policy and rootwrap configurations 2016-09-08 https://review.openstack.org/367521 | 16:53 |
| *** TxGirlGeek has quit IRC | 16:53 | |
| odyssey4me | yeah, we should use bugs as TODO items, really | 16:54 |
| jamesdenton | hughmFLEXin and so far everything is OK? If eth3 is intended to be managed by Neutron, then you can't have a static IP on it. You can create eth3.vlan interfaces and put an IP on that, though. Then Neutron can create other vlan interfaces off that interface for provider networks | 16:54 |
| *** TxGirlGeek has joined #openstack-ansible | 16:54 | |
| odyssey4me | don't triage them, so the triage meeting exposes them to anyone looking for work | 16:54 |
| *** galstrom is now known as galstrom_zzz | 16:54 | |
| automagically | odyssey4me: Ah, fair enough | 16:54 |
| automagically | I put it back to Undecided | 16:55 |
| *** jcannava has joined #openstack-ansible | 16:55 | |
| *** TxGirlGeek has quit IRC | 16:56 | |
| *** TxGirlGeek has joined #openstack-ansible | 16:57 | |
| hughmFLEXin | jamesdenton: hmmm, if i was the networking guy i would have a smarter reply, unfortunately, ill have to wait for him to come in | 16:58 |
| openstackgerrit | Travis Truman (automagically) proposed openstack/openstack-ansible-tests: Fix container destroy host group https://review.openstack.org/370282 | 16:58 |
| cloudnull | automagically: done -- odyssey4me: ^ | 16:59 |
| odyssey4me | https://bugs.launchpad.net/openstack-ansible/+bug/1623597 | 16:59 |
| openstack | Launchpad bug 1623597 in openstack-ansible "[doc] Ensure that the reason for using the role link mechanism in role tests is documented" [Undecided,New] | 16:59 |
| jamesdenton | hughmFLEXin no worries. Sounds like you have somewhere to go now with the neutron client. Sorry Horizon isn;t working. What version are you using, anyway? v13? | 17:00 |
| hughmFLEXin | 14b3 | 17:00 |
| odyssey4me | automagically we should also add https://github.com/openstack/openstack-ansible/blob/master/playbooks/lxc-containers-destroy.yml#L49 | 17:01 |
| hughmFLEXin | horizon seems to work (except for creating that vlan network), im just 95% sure the VM's won't have connectivity | 17:01 |
| odyssey4me | although, I'm not entirely sure why we don't just grab the whole of https://github.com/openstack/openstack-ansible/blob/master/playbooks/lxc-containers-destroy.yml | 17:01 |
| automagically | odyssey4me: Good catch | 17:01 |
| hughmFLEXin | uploading images and all that jazz seem to be fine | 17:01 |
| stevelle | cloudnull: any idea why this isn't working? https://review.openstack.org/#/c/366931 patch set 4 should have worked afaict but I might be blind to something obvious | 17:01 |
| odyssey4me | that said, why don't we just use the integrated repo's playbook | 17:01 |
| jamesdenton | hughmFLEXin Dangerous must be your middle name | 17:02 |
| hughmFLEXin | jamesdenton: the docs specify 14! | 17:02 |
| hughmFLEXin | i told cloudnull that yesterday! | 17:02 |
| irtermite | cloudnull: lives too far in the future | 17:02 |
| cloudnull | odyssey4me: i think it needs to be http://cdn.pasteraw.com/6cxtwtadca2do7lw688xctobj6v0gl9 | 17:03 |
| irtermite | he will never come back. we miss him in the now | 17:03 |
| *** adrian_otto has quit IRC | 17:03 | |
| *** TxGirlGeek has quit IRC | 17:03 | |
| automagically | odyssey4me: Yeah, good point. Maybe I’ll give a go. The downside is every test run will end up cloning the integrated repo | 17:03 |
| automagically | AFK for a bit | 17:03 |
| cloudnull | hughmFLEXin: we have docs per release though i dont believe openstack does a good job letting folks know that master == unstable | 17:04 |
| cloudnull | http://docs.openstack.org/developer/openstack-ansible/ | 17:04 |
| cloudnull | if folsk just go to http://docs.openstack.org/developer/openstack-ansible/install-guide/index.html you get master | 17:05 |
| cloudnull | while stable is http://docs.openstack.org/developer/openstack-ansible/mitaka/ | 17:05 |
| openstackgerrit | Merged openstack/openstack-ansible-repo_server: Convert role testing to use Ansible 2.1.1 https://review.openstack.org/369782 | 17:05 |
| cloudnull | asettle rromans: is there a banner header or something that can be added? | 17:05 |
| irtermite | cloudnull: I think hughmFLEXin is talking about "Replace TAG with the current stable release tag : 14.0.0.0b3" | 17:06 |
| cloudnull | when on master docs so folks know they're woking on unstable ? | 17:06 |
| *** jperry has quit IRC | 17:06 | |
| odyssey4me | stevelle I think you meant http://cdn.pasteraw.com/6cxtwtadca2do7lw688xctobj6v0gl9 for stevelle ? | 17:06 |
| irtermite | http://docs.openstack.org/developer/openstack-ansible/install-guide/deploymenthost.html | 17:06 |
| cloudnull | odyssey4me stevelle^ | 17:06 |
| *** weezS has joined #openstack-ansible | 17:06 | |
| *** jperry has joined #openstack-ansible | 17:06 | |
| stevelle | cloudnull: both syntax are supposed to work and I did have that inline syntax working when I modeled it :( | 17:07 |
| cloudnull | irtermite: i think that's generated http://docs.openstack.org/developer/openstack-ansible/install-guide/deploymenthost.html#installing-source-and-dependencies | 17:07 |
| cloudnull | stevelle: i think for 1.9 it works and 2.1 it doesn't | 17:07 |
| irtermite | generated or not... that's the (default) doc that pops up for people. If they don't know any better... they will follow. | 17:07 |
| irtermite | <3 | 17:07 |
| hughmFLEXin | </3 | 17:07 |
| irtermite | hahahaa | 17:07 |
| irtermite | we broke hughmFLEXin | 17:08 |
| cloudnull | https://github.com/openstack/openstack-ansible/blob/master/doc/source/install-guide/deploymenthost.rst#installing-source-and-dependencies | 17:08 |
| cloudnull | irtermite: hughmFLEXin ^ that value just inherits whatever most recent tag is, | 17:08 |
| irtermite | The doc should probably have a second line that lists all versions? | 17:09 |
| irtermite | one sec... | 17:09 |
| cloudnull | yea. something. | 17:09 |
| asettle | cloudnull: I actually think odyssey4me removed that banner | 17:09 |
| hughmFLEXin | yea...cloudnull didn't you file a bug on that yesterday | 17:09 |
| asettle | But the docs aren't unstable? Or do you mean the branch is unstable | 17:10 |
| asettle | What flexy hughmFLEXin just said | 17:10 |
| cloudnull | odyssey4me: ^ | 17:10 |
| cloudnull | I did | 17:10 |
| cloudnull | docs are not unstable | 17:10 |
| irtermite | cloudnull: you filed already? I was about to grab, edit, pull... | 17:10 |
| cloudnull | the checkout is | 17:10 |
| cloudnull | I filed a bug | 17:10 |
| cloudnull | i didn't fix it | 17:10 |
| irtermite | kk | 17:10 |
| openstackgerrit | Merged openstack/openstack-ansible-os_zaqar: Add role linking to tox tests https://review.openstack.org/370248 | 17:11 |
| asettle | cloudnull: sorry, i'm not sure I understand your question/issue then | 17:11 |
| *** Matias has joined #openstack-ansible | 17:11 | |
| cloudnull | peeople reading the docs from master take |my_conf_val| and try to use it | 17:11 |
| cloudnull | which in this case is 14.0.0.0b3 | 17:11 |
| asettle | Oh *facepalm* right, right. | 17:11 |
| cloudnull | or the beta release of newton | 17:12 |
| Matias | did someone talk to me? my screen+weechat pack died | 17:12 |
| irtermite | ^^ asettle so, you could leave that, but add a sub-note to use stable | 17:12 |
| irtermite | and maybe even list others | 17:12 |
| cloudnull | ^ the problem there is that others may not apply to the current docs | 17:12 |
| irtermite | fair | 17:13 |
| cloudnull | I think a banner or something similar to let folks know master docs == headache is a better solution | 17:13 |
| asettle | That, exactly. So, I can add a note later. I'm probably about to leave work. | 17:13 |
| irtermite | then, it should have both current and stable | 17:13 |
| irtermite | oh nvm, that won't work | 17:13 |
| irtermite | the rest of the doc won't make sense... | 17:13 |
| irtermite | hahaa I know... | 17:14 |
| asettle | Either odyssey4me will have a solution, or I'll implement one tonight :) | 17:14 |
| cloudnull | thanks asettle: go home :) | 17:14 |
| asettle | thanks cloudnull :) | 17:14 |
| asettle | I swear I'll do the thing with the note | 17:14 |
| * asettle scribbles it on a post-it | 17:14 | |
| irtermite | The headache banner should read... "You are about to deploy unstable. Please click {here} to escape this land and restart in stable land." | 17:14 |
| asettle | irtermite: you should write it ;) | 17:14 |
| cloudnull | asettle: no worries. whenever's clever | 17:15 |
| irtermite | asettle: just cloned | 17:15 |
| irtermite | can do | 17:15 |
| asettle | HAHAH OMG DONT WRITE "STABLE LAND" :P | 17:15 |
| *** Attilla has quit IRC | 17:15 | |
| irtermite | heheheh | 17:15 |
| irtermite | but, but but | 17:15 |
| asettle | Hahahah you're fired from the docs role | 17:15 |
| cloudnull | hahahaha | 17:15 |
| irtermite | dang it | 17:16 |
| asettle | So much promise *shakes head* | 17:16 |
| irtermite | rise so high so quick. fall harder | 17:16 |
| cloudnull | that's my constant threat. someone write the docs or deal with the nonsense I create. | 17:16 |
| irtermite | cloudnull: so many times... so many times... | 17:17 |
| asettle | cloudnull: I mean, it works, it keeps me in a job. | 17:17 |
| openstackgerrit | Paulo Matias proposed openstack/openstack-ansible-repo_build: Drop privileges before running the git clone script https://review.openstack.org/366280 | 17:17 |
| asettle | But there is an awful lot of shit that happened pre-me. | 17:17 |
| asettle | Okay, I really gotta go. I can feel my bottom melding with the seat. | 17:17 |
| asettle | o/ | 17:17 |
| cloudnull | have a good one asettle | 17:17 |
| *** asettle has quit IRC | 17:17 | |
| *** adrian_otto has joined #openstack-ansible | 17:17 | |
| *** asettle has joined #openstack-ansible | 17:18 | |
| openstackgerrit | Major Hayden proposed openstack/openstack-ansible-security: [Docs] More cleanup https://review.openstack.org/370213 | 17:18 |
| cloudnull | odyssey4me: Matias: mgariepy: we may need a sudoers solution to make https://review.openstack.org/366280 work | 17:19 |
| *** adrian_otto has quit IRC | 17:19 | |
| cloudnull | idk but potentially related. | 17:19 |
| *** adrian_otto has joined #openstack-ansible | 17:20 | |
| Matias | cloudnull: for the test? | 17:20 |
| openstackgerrit | Merged openstack/openstack-ansible-os_aodh: Ansible 2.1.1 role testing https://review.openstack.org/353563 | 17:20 |
| cloudnull | to use become on cent systems may require a tty | 17:20 |
| cloudnull | which is dropped by default. | 17:20 |
| cloudnull | it might work out of the box | 17:21 |
| cloudnull | we we're just talking about it this morning though | 17:21 |
| evrardjp | https://review.openstack.org/#/c/370150/ | 17:21 |
| cloudnull | evrardjp: can we add that to the common tests repo ? | 17:22 |
| cloudnull | so that it happens for all of the containers we create? | 17:22 |
| *** jlockwood has joined #openstack-ansible | 17:22 | |
| evrardjp | cloudnull: I just asked this question a little above | 17:22 |
| openstackgerrit | Major Hayden proposed openstack/openstack-ansible-ops: Remove osa-differ https://review.openstack.org/370234 | 17:22 |
| cloudnull | oh , i missed it sorry | 17:23 |
| evrardjp | cloudnull: I think it could work if we never override this | 17:23 |
| hughmFLEXin | cloudnull: https://s12.postimg.org/3m9h615m5/Screen_Shot_2016_09_14_at_10_18_13_AM.png | 17:23 |
| evrardjp | no my bad it was just one question that disappeared in the flow | 17:23 |
| evrardjp | so | 17:23 |
| evrardjp | either we do a include in the tests repo | 17:23 |
| cloudnull | hughmFLEXin: great success! | 17:23 |
| cloudnull | :) | 17:23 |
| openstackgerrit | Major Hayden proposed openstack/openstack-ansible-ops: Remove osa-differ https://review.openstack.org/370234 | 17:23 |
| evrardjp | but then no override in each role | 17:23 |
| hughmFLEXin | unfortunately won't have external access, but this is a good first step lol | 17:24 |
| odyssey4me | cloudnull yeah, I asked evrardjp to work on that in the os_neutron repo so long so that we can diagnose the issue - if we could push a patch for the common tests repo and consume it for all container setups then that would be amazing | 17:24 |
| cloudnull | evrardjp: we can add it to the tests repo as a test-var | 17:24 |
| *** TxGirlGeek has joined #openstack-ansible | 17:24 | |
| evrardjp | (except if we go through a different variable name, which starts to be ugly) | 17:24 |
| odyssey4me | yeah, actually, that could work | 17:24 |
| hughmFLEXin | cloudnull: Matias: jamesdenton: thanks for all the help though | 17:24 |
| evrardjp | cloudnull: yes, that was my first thought | 17:24 |
| openstackgerrit | Adam Reznechek proposed openstack/openstack-ansible-os_nova: Update PowerVM driver documentation https://review.openstack.org/364627 | 17:24 |
| evrardjp | that's why I asked: do you think we would have to override that at some point for a role | 17:25 |
| evrardjp | then I thought of many different weird shit to overlap inventories | 17:25 |
| odyssey4me | no, but if we do - then we can always override with the inclusion of the default content | 17:25 |
| jamesdenton | hughmFLEXin no problem. Good luck living on the edge! ;) | 17:25 |
| openstackgerrit | Adam Reznechek proposed openstack/openstack-ansible-os_nova: [DOCS] Update PowerVM driver documentation https://review.openstack.org/364627 | 17:25 |
| odyssey4me | we already have a mechanism for the centralised tests to override those test vars | 17:25 |
| odyssey4me | it just uses -e for CLI overrides | 17:26 |
| evrardjp | yes, fine for me | 17:26 |
| *** asettle has quit IRC | 17:26 | |
| evrardjp | so I'll put that in test repo too | 17:26 |
| odyssey4me | if we need something more complex than that, let's cross that bridge when we get there | 17:26 |
| odyssey4me | the idea is to keep things as simple as possible to make it easy to use | 17:26 |
| cloudnull | yea i think adding it to test-vars.yml makes sense. | 17:27 |
| cloudnull | which will pick up logs for all containers as we move tests to that rep o | 17:27 |
| openstackgerrit | Merged openstack/openstack-ansible-ceph_client: Restart OS services when symlinks are created https://review.openstack.org/370296 | 17:28 |
| cloudnull | odyssey4me: https://review.openstack.org/#/c/370253/ | 17:28 |
| cloudnull | automagically: ^ | 17:28 |
| cloudnull | for the role linking bits | 17:28 |
| hughmFLEXin | jamesdenton: have you configured any setups where one of the eth interfaces are taken up by PXE? | 17:29 |
| Matias | hughmFLEXin: I have PXE running in the same interface as br-mgmt | 17:30 |
| openstackgerrit | Merged openstack/openstack-ansible-memcached_server: Add role linking to tox tests https://review.openstack.org/370261 | 17:31 |
| hughmFLEXin | Matias: is your pxe also providing the dhcp for external connectivity? | 17:31 |
| Matias | hughmFLEXin: yes | 17:31 |
| cloudnull | in my setup i have pxe running on em1 | 17:31 |
| jamesdenton | hughmFLEXin yeah it should be a problem to have that same interface (or vlan interface) handle DHCP/TFTP | 17:31 |
| jamesdenton | *should NOT be a problem, rather | 17:32 |
| hughmFLEXin | hm well fk | 17:33 |
| Matias | hughmFLEXin: https://web-dev.ufscar.br/0bin/paste/EcjpfjGN#yeXZbAvmrJaaOG6t5Ku2jgeSLvJlqPZy94s6wMgQmfs | 17:34 |
| *** adrian_otto has quit IRC | 17:34 | |
| jamesdenton | Whatever VLAN is used for PXE must be the native VLAN to the PXE client | 17:34 |
| jamesdenton | (in most cases, anyway) | 17:35 |
| *** michaelgugino has joined #openstack-ansible | 17:35 | |
| hughmFLEXin | Matias: how is your br-ex being utilized? | 17:38 |
| Matias | hughmFLEXin: I use it to add the external network to neutron | 17:38 |
| openstackgerrit | Jean-Philippe Evrard proposed openstack/openstack-ansible-os_neutron: [DO NOT MERGE] More logging for neutron https://review.openstack.org/370335 | 17:39 |
| Matias | neutron net-create --provider:network_type flat --provider:physical_network external --router:external=True public | 17:39 |
| hughmFLEXin | ah so you defined br-ex in your global overrides for flat? | 17:39 |
| Matias | br-ex is declared as a flat network, named external, in provider_networks (openstack_user_config.yml) | 17:39 |
| Matias | yes | 17:40 |
| hughmFLEXin | named == net_name? | 17:40 |
| Matias | yes | 17:40 |
| hughmFLEXin | doood its all starting to make sense | 17:40 |
| Matias | hughmFLEXin: http://paste.openstack.org/show/576306/ | 17:41 |
| * hughmFLEXin sobs quietly | 17:41 | |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible: Remove assumption that the neutron_lbaas var is set https://review.openstack.org/368905 | 17:41 |
| jamesdenton | Matias I assume that the br-bgp bridge is used for a dedicated peering network for Neutron BGP? | 17:42 |
| *** michaelgugino has quit IRC | 17:42 | |
| Matias | jamesdenton: yes | 17:42 |
| openstackgerrit | Merged openstack/openstack-ansible-os_nova: [DOCS] Update PowerVM driver documentation https://review.openstack.org/364627 | 17:42 |
| *** michaelgugino has joined #openstack-ansible | 17:43 | |
| jamesdenton | great. I was intending on doing to same thing. Didn't see any docs related to that configuration, necessarily. Do you know of anything? | 17:43 |
| Matias | jamesdenton: I followed this docs when setting up: https://cloudbau.github.io/openstack/neutron/networking/2016/05/17/neutron-ipv6.html | 17:43 |
| Matias | you can ask if you need any detail | 17:43 |
| jamesdenton | ok fair enough. And you just added the OSA bits to get the interface in the agents container | 17:44 |
| Matias | in my case, I'm setting up the eth13 ip address using a custom playbook I run after OSA, because we are setting up an ipv6 address for it | 17:44 |
| jamesdenton | <thumbs up> | 17:44 |
| Matias | which is not supported by the current inventory script | 17:44 |
| openstackgerrit | Merged openstack/openstack-ansible-repo_build: Ansible 2.1.1 role testing https://review.openstack.org/370245 | 17:44 |
| Matias | but you could just use an ipv4 address | 17:45 |
| *** thorst_ has quit IRC | 17:45 | |
| Matias | in this case you could just use ip_from_q in provider_networks | 17:45 |
| Matias | and declare a queue with a small range | 17:45 |
| hughmFLEXin | man...people willingly want to be networking engineers? | 17:46 |
| *** thorst_ has joined #openstack-ansible | 17:46 | |
| jamesdenton | Matias is there only 1 speaker that can advertise the networks? Is it one speaker per scope? Or do you know if you can configure multiple speakers for HA | 17:46 |
| Matias | jamesdenton: http://paste.openstack.org/show/576307/ | 17:46 |
| Matias | the last line setups it for HA | 17:47 |
| Matias | 3 speakers in my case | 17:47 |
| jamesdenton | yeah i was intending on creating a new cidr and everything | 17:47 |
| evrardjp | Matias: would you mind submitting this to the docs? | 17:47 |
| Matias | heck | 17:47 |
| Matias | I will need to change the bgp session password | 17:47 |
| jamesdenton | perfect | 17:47 |
| Matias | just published it | 17:47 |
| Matias | LOL | 17:47 |
| jamesdenton | lol | 17:47 |
| evrardjp | shit lol | 17:47 |
| * jamesdenton promises to forget | 17:47 | |
| Matias | paste.openstack.org will never forget :P | 17:48 |
| evrardjp | true | 17:48 |
| jamesdenton | nice work Matias. I haven't gotten that far into it :( | 17:48 |
| Matias | evrardjp: I submitted a link to the cloudbau doc to the docs | 17:48 |
| evrardjp | have fun with the change of passwords :p | 17:48 |
| Matias | I was going to write up my own experience later, but the only difference is the multiple speaker setup, I think | 17:49 |
| * odyssey4me enters the password into is dictionaries :) | 17:49 | |
| Matias | which is just adding more than one speaker | 17:49 |
| jamesdenton | yeah. but what could also be helpful is the peer-side configuration too, depending on what you're using. And of course your openstack_user_config changes and playbook runs | 17:50 |
| jamesdenton | there's a lot of hand-wavey black magic docs out there | 17:50 |
| *** retreved has joined #openstack-ansible | 17:50 | |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible: Implement scenario capability for AIO https://review.openstack.org/369485 | 17:51 |
| *** electrofelix has quit IRC | 17:51 | |
| * jamesdenton commissions Matias for Neutron book v3 | 17:51 | |
| Matias | 179/tcp closed bgp | 17:51 |
| Matias | yokatta ne | 17:52 |
| Matias | the bgp port is not open for the external world :P | 17:52 |
| Matias | the password is just additional security | 17:52 |
| hughmFLEXin | if i have br-vxlan set as the internal network in horizon, technically i should be able to ping from one of the controller nodes right? | 17:54 |
| *** adrian_otto has joined #openstack-ansible | 17:54 | |
| *** thorst_ has quit IRC | 17:54 | |
| Matias | cloudnull: http://logs.openstack.org/80/366280/3/check/gate-openstack-ansible-repo_build-ansible-func-ubuntu-trusty/012f4bb/console.html | 17:55 |
| Matias | ops | 17:55 |
| Matias | cloudnull: http://logs.openstack.org/80/366280/3/check/gate-openstack-ansible-repo_build-ansible-func-ubuntu-trusty/012f4bb/console.html#_2016-09-14_17_32_04_601405 | 17:55 |
| Matias | "sudo: unknown user: {{" this is funny, it is not even trying to replace for the variable | 17:56 |
| Matias | I wonder if including the variable in test-vars.yml would help | 17:56 |
| openstackgerrit | Merged openstack/openstack-ansible-os_rally: Add role linking to tox tests https://review.openstack.org/370253 | 17:58 |
| odyssey4me | Matias yes, if you're doing that in the test-install-server.yml playbook (which doesn't know the role defaults), then you must | 17:58 |
| odyssey4me | Matias but ideally that fix should work without touching the playbook, surely? | 17:58 |
| odyssey4me | anyway, keep on trucking - let's see if we can find a solution there | 17:59 |
| *** rk4n has joined #openstack-ansible | 18:00 | |
| Matias | odyssey4me: it's because the playbook precaches 2 git repos as root after running repo_server | 18:00 |
| Matias | unless I move them to be precached before repo_server | 18:00 |
| odyssey4me | Matias yep - and that is how we do it in the integrated repo too | 18:00 |
| odyssey4me | Matias if there's a pre-cached git folder on the host, we bind mount it: https://github.com/openstack/openstack-ansible/blob/master/playbooks/repo-server.yml#L39 | 18:01 |
| Matias | odyssey4me: the cloudnull's permission fix is done in the repo_server role, it needs to be run before repo_install if there any git repos owner by root | 18:01 |
| *** adrian_otto has quit IRC | 18:01 | |
| Matias | if there are * | 18:02 |
| Matias | owned by root* | 18:02 |
| odyssey4me | Matias then we either symlink it or synchronise it - depending on the location of the git repo cache: https://github.com/openstack/openstack-ansible/blob/master/playbooks/repo-build.yml#L81-L101 | 18:02 |
| *** adrian_otto has joined #openstack-ansible | 18:02 | |
| odyssey4me | the repo_server role is executed before repo_build | 18:02 |
| *** BjoernT has quit IRC | 18:03 | |
| *** drifterza has joined #openstack-ansible | 18:03 | |
| Matias | odyssey4me: in test-install-server.yml, it is not | 18:04 |
| odyssey4me | Matias ah, I see in the tests for repo_build we don't do it in the same order | 18:04 |
| Matias | yep | 18:04 |
| Matias | the precache is in the post_tasks | 18:04 |
| odyssey4me | Matias ok, then perhaps we should rather move those tasks to pre_tasks | 18:04 |
| Matias | ok, let me try | 18:04 |
| stevelle | odyssey4me: for changes on the "shared-apache-host" topic is that something you expect to merge for Newton? | 18:04 |
| Matias | I will use the infra to run the tests | 18:04 |
| * Matias lazy to run them locally | 18:04 | |
| odyssey4me | Matias ah no - I see what those tests were for and where they're targeting... your approach is right | 18:04 |
| odyssey4me | stevelle not highly motivated, no - but if you want to get them in then you can happily motivate them along | 18:05 |
| stevelle | odyssey4me: just checking because I may introduce another use of mod_wsgi in Newton b/c of upstream changes | 18:05 |
| stevelle | didn't want to pile on ork | 18:05 |
| cloudnull | automagically: are you working on any of these currently https://review.openstack.org/#/q/status:open+topic:bp/ansible-2-1-support ? | 18:05 |
| *** retreved has quit IRC | 18:05 | |
| Matias | odyssey4me: which of the approaches is right? post_tasks with become or pre_tasks? | 18:06 |
| odyssey4me | stevelle no worries - do your thing, I'll rvise my patches at a later time | 18:06 |
| cloudnull | i wanted to help out in getting more of those up and running and dont want to step on works in progress. | 18:06 |
| *** retreved has joined #openstack-ansible | 18:06 | |
| odyssey4me | cloudnull I think that andymccr mentioned that he's working on nova & swift | 18:07 |
| cloudnull | ok | 18:07 |
| odyssey4me | but he's trying to solve the centos networking issue at the same time | 18:07 |
| odyssey4me | Matias leave the patch as is - just add the var to the test vars so that the playbook knows which user to use | 18:09 |
| *** rk4n has quit IRC | 18:09 | |
| *** adrian_otto has quit IRC | 18:10 | |
| openstackgerrit | Steve Lewis (stevelle) proposed openstack/openstack-ansible-os_ceilometer: Serve the ceilometer-api using mod_wsgi https://review.openstack.org/370352 | 18:10 |
| *** rk4n has joined #openstack-ansible | 18:10 | |
| *** adrian_otto has joined #openstack-ansible | 18:11 | |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible-lxc_container_create: Change the container prep using more intelligent commands https://review.openstack.org/354120 | 18:12 |
| *** rk4n has quit IRC | 18:12 | |
| *** drifterza has quit IRC | 18:13 | |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible-lxc_container_create: Change the container prep using more intelligent commands https://review.openstack.org/354120 | 18:13 |
| *** poopcat has quit IRC | 18:18 | |
| *** galstrom_zzz is now known as galstrom | 18:19 | |
| *** itsuugo has quit IRC | 18:19 | |
| *** poopcat has joined #openstack-ansible | 18:19 | |
| *** itsuugo has joined #openstack-ansible | 18:20 | |
| openstackgerrit | Jean-Philippe Evrard proposed openstack/openstack-ansible-os_neutron: [DO NOT MERGE] More logging for neutron https://review.openstack.org/370335 | 18:21 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible: Add -t option to create a pseudo TTY https://review.openstack.org/370197 | 18:21 |
| *** galstrom is now known as galstrom_zzz | 18:21 | |
| odyssey4me | Matias want me to tweak https://review.openstack.org/366280 for you? | 18:23 |
| openstackgerrit | Paulo Matias proposed openstack/openstack-ansible-repo_build: Drop privileges before running the git clone script https://review.openstack.org/366280 | 18:24 |
| odyssey4me | ah, you did it already | 18:24 |
| Matias | odyssey4me: if this time it doesn't work feel free to fix, please | 18:25 |
| *** retreved has quit IRC | 18:26 | |
| odyssey4me | andymccr take a peek at https://review.openstack.org/#/c/354120/ for some ideas regarding the tests play for network setup | 18:29 |
| odyssey4me | cloudnull did you see the regressions mentioned in https://review.openstack.org/365458 ? | 18:31 |
| cloudnull | no | 18:32 |
| cloudnull | https://review.openstack.org/#/c/363022/ that was with this change | 18:33 |
| cloudnull | ? | 18:33 |
| cloudnull | rather after that change? | 18:33 |
| odyssey4me | cloudnull nope - that issue was in a previous change | 18:33 |
| odyssey4me | cloudnull that's why I didn't block that patch due to that regression | 18:34 |
| cloudnull | ok | 18:36 |
| cloudnull | so the regression was before then | 18:36 |
| * cloudnull looking | 18:36 | |
| odyssey4me | I suspect https://github.com/openstack/openstack-ansible-plugins/commit/c8e25f50e4e5bf4ed6071745542de334a7dc80bd, considering that we had to do https://github.com/openstack/openstack-ansible-plugins/commit/523afbd0142dcb40ef3265499944f178c601dddf immediately after it | 18:38 |
| openstackgerrit | Merged openstack/openstack-ansible-os_keystone: Add role linking to tox tests https://review.openstack.org/370246 | 18:40 |
| odyssey4me | Matias good news, I see that https://review.openstack.org/366280 has passed 2 of 3 func tests - the other one is just waiting for a node | 18:44 |
| odyssey4me | cloudnull ^ you may wish to re-add your vote there | 18:44 |
| hughmFLEXin | Matias: do you think there is anyway i can salvage the current deployment, maybe do some magic on the network switch? | 18:45 |
| openstackgerrit | Merged openstack/openstack-ansible: Configure Ceilometer middleware for Gnocchi-Swift https://review.openstack.org/366279 | 18:49 |
| *** jwitko has quit IRC | 18:49 | |
| *** hughmFLEXin has quit IRC | 18:50 | |
| odyssey4me | anyone around that can respond on the openstack-dev ML about http://lists.openstack.org/pipermail/openstack-dev/2016-September/103671.html ? | 18:51 |
| *** hughmFLEXin has joined #openstack-ansible | 18:52 | |
| *** thorst_ has joined #openstack-ansible | 18:53 | |
| cloudnull | odyssey4me: yes he'll need to redeploy | 18:54 |
| cloudnull | i can reply later today. | 18:54 |
| cloudnull | right now im looking into the plugin issues | 18:54 |
| odyssey4me | awesome, thanks cloudnull | 18:54 |
| cloudnull | if cinder volumes are running in containers and they need to attach to isci that wont work | 18:55 |
| odyssey4me | yep, we recently added notes all over the place about that, but I don't think we backported them to Mitaka | 18:55 |
| stevelle | Matias: https://review.openstack.org/#/c/366279/ merged. Thanks for poking the bear :) | 18:55 |
| *** thorst_ has quit IRC | 19:00 | |
| openstackgerrit | Steve Lewis (stevelle) proposed openstack/openstack-ansible-os_ceilometer: WIP Serve the ceilometer-api using mod_wsgi https://review.openstack.org/370352 | 19:00 |
| *** e-vad has quit IRC | 19:02 | |
| *** AnarchyAo has joined #openstack-ansible | 19:03 | |
| cloudnull | odyssey4me: do we have a bug for the regression ? | 19:05 |
| *** TxGirlGeek has quit IRC | 19:05 | |
| cloudnull | also we may need to backport the fix | 19:05 |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible-plugins: Fix regression for proprietary packages https://review.openstack.org/370384 | 19:06 |
| odyssey4me | cloudnull no we don't have a bug for it | 19:07 |
| cloudnull | ok | 19:07 |
| cloudnull | ^ fix if you dont mind giving that a spin | 19:07 |
| odyssey4me | lemme test that out | 19:07 |
| *** pcaruana has joined #openstack-ansible | 19:07 | |
| odyssey4me | cloudnull oh actually we do have a bug for it | 19:08 |
| odyssey4me | cloudnull https://bugs.launchpad.net/openstack-ansible/+bug/1622275 | 19:09 |
| openstack | Launchpad bug 1622275 in openstack-ansible "repo_build_wheel_selective: no breaks repo_build" [High,Confirmed] - Assigned to Jesse Pretorius (jesse-pretorius) | 19:09 |
| odyssey4me | shall I tweak the commit msg quickly | 19:09 |
| odyssey4me | ? | 19:09 |
| cloudnull | go right ahead | 19:09 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-plugins: Fix regression for proprietary packages https://review.openstack.org/370384 | 19:09 |
| *** kylek3h_away is now known as kylek3h | 19:19 | |
| openstackgerrit | Merged openstack/openstack-ansible-repo_build: Drop privileges before running the git clone script https://review.openstack.org/366280 | 19:19 |
| odyssey4me | Matias ^ w00t - thanks for figuring that out and helping get it right! | 19:19 |
| odyssey4me | cloudnull when you have a bit of time, please take a look through https://review.openstack.org/369485 - to try and combat the consistent timeouts I'd like to break apart the current integrated job a bit and need that to proceed | 19:22 |
| *** adrian_otto has quit IRC | 19:22 | |
| odyssey4me | jmccrory automagically cloudnull d34dh0r53 stevelle mattt | 19:24 |
| odyssey4me | hughsaunders andymccr mhayden evrardjp asettle rromans the RC1 deadline submission is tomorrow, so we need everything ready to do a role pin by the US morning to ensure that we can merge it in time for a release request by COB | 19:24 |
| *** sdake_ has joined #openstack-ansible | 19:29 | |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-repo_build: Update py_pkgs output data for binary package lookup https://review.openstack.org/365458 | 19:31 |
| *** sdake has quit IRC | 19:33 | |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-repo_build: Update py_pkgs output data for binary package lookup https://review.openstack.org/365458 | 19:33 |
| mgariepy | cloudnull, odyssey4me, mhayden http://paste.openstack.org/show/576315/ | 19:34 |
| mgariepy | -t doesnt do the job afterall... | 19:34 |
| mgariepy | opps, need to remove -T.. | 19:35 |
| odyssey4me | cloudnull yes, that does it - very happy - can you review https://review.openstack.org/365458 as it's now final | 19:35 |
| odyssey4me | we need another core review on https://review.openstack.org/370384 to fix the regression please | 19:36 |
| openstackgerrit | Nish Patwa(nishpatwa_) proposed openstack/openstack-ansible: [DOCS] Added HAProxy example to Production environment document. https://review.openstack.org/370187 | 19:40 |
| *** dragonmaster has quit IRC | 19:42 | |
| *** asettle has joined #openstack-ansible | 19:43 | |
| openstackgerrit | Steve Lewis (stevelle) proposed openstack/openstack-ansible-os_ceilometer: WIP Serve the ceilometer-api using mod_wsgi https://review.openstack.org/370352 | 19:46 |
| SamYaple | Can I get some help with this issue? http://paste.openstack.org/show/576318/ . not sure where to start troubleshooting further now | 19:48 |
| SamYaple | Liberty, 12.2.3 | 19:48 |
| *** hughmFLEXin has quit IRC | 19:48 | |
| mgariepy | still doesn't work whitout the -T (and the socket being kill for ssh.) http://paste.openstack.org/show/576319/ | 19:48 |
| openstackgerrit | Merged openstack/openstack-ansible-plugins: Fix regression for proprietary packages https://review.openstack.org/370384 | 19:48 |
| *** hughmFLEXin has joined #openstack-ansible | 19:48 | |
| openstackgerrit | Logan V proposed openstack/openstack-ansible: Aodh should inherit service_region https://review.openstack.org/370405 | 19:50 |
| *** johnmilton has quit IRC | 19:53 | |
| *** can8dnSix has joined #openstack-ansible | 19:55 | |
| *** thorst_ has joined #openstack-ansible | 19:58 | |
| *** pcaruana has quit IRC | 20:00 | |
| openstackgerrit | Merged openstack/openstack-ansible-security: [Docs] More cleanup https://review.openstack.org/370213 | 20:00 |
| cloudnull | mgariepy: bummer about -t | 20:04 |
| *** thorst_ has quit IRC | 20:04 | |
| cloudnull | so then I guess the next best option is to update the base lxc container image and update the openstack-hosts repo to ensure the requiretty option is removed | 20:05 |
| *** pjm6 has quit IRC | 20:05 | |
| *** gouthamr has quit IRC | 20:07 | |
| *** can8dnSix has quit IRC | 20:08 | |
| *** e-vad has joined #openstack-ansible | 20:10 | |
| *** can8dnSix has joined #openstack-ansible | 20:13 | |
| *** Attilla has joined #openstack-ansible | 20:14 | |
| odyssey4me | SamYaple it looks like you're using Ansible 2.x with Liberty? | 20:17 |
| openstackgerrit | Merged openstack/openstack-ansible: Add collect_statistics_interval, rates_mode in rabbitmq.config template https://review.openstack.org/363046 | 20:17 |
| SamYaple | odyssey4me: oh I am not sure. is that no a thing? let me check | 20:17 |
| odyssey4me | SamYaple yeah, Ansible 2.x is only supported from Newton forward - Liberty/Mitaka need Ansible 1.9.4 | 20:18 |
| SamYaple | odyssey4me: let me check if thats what the env is using | 20:18 |
| odyssey4me | SamYaple the actual failure though is a mismatch between the repo's wheels and the requirements of the component it's trying to build a venv for (Ironic) | 20:19 |
| odyssey4me | SamYaple ie Ironic needs things that aren't provided in the Liberty u-c file, or something to that affect - are you using a mix of versions? something like Liberty OSA with Mitaka Ironic? | 20:19 |
| SamYaple | odyssey4me: not intentionally I promise | 20:21 |
| SamYaple | but this is a recycled lab, so probably | 20:21 |
| odyssey4me | SamYaple when using the repo, you have to ensure that the upstream branch versions are the same and the SHA's are from that branch... if you want to use a later branch service then you have to activate developer_mode for that service so that there is no venv built for it... and even then I'm not sure that it'll work (untested code path) | 20:21 |
| SamYaple | also, am using ansible 2.x, will roll back | 20:21 |
| michaelgugino | is anyone else here having issues running tox on OSX? | 20:22 |
| odyssey4me | yeh, Ansible 2.x *might* work - but it's an untested set and we've done a ton of stuff for Ansible 2.1 to work - so it's probably best not to go there | 20:22 |
| michaelgugino | pip is bound and determined to install cryptography from source instead of using the wheel. | 20:22 |
| SamYaple | odyssey4me: since i have only git cloned and checked out tag 12.2.3 for openstack-ansible, im not sure where the other stuff is coming from | 20:22 |
| SamYaple | michaelgugino: is wheel installed? :) | 20:23 |
| michaelgugino | yes | 20:23 |
| SamYaple | gotta ask the simple questinos | 20:23 |
| odyssey4me | SamYaple check whether /etc/ansible is populated with anything on the deploy host - and remove it all | 20:23 |
| SamYaple | odyssey4me: it is and will do | 20:23 |
| SamYaple | odyssey4me: do i need to reset the env? remove all containers and what not? | 20:24 |
| SamYaple | or can i just run the playbooks again | 20:24 |
| odyssey4me | SamYaple hmm, it all depends on whether Ansible executed any of those roles | 20:24 |
| odyssey4me | probably best to use lxc-containers-destroy to ditch the containers | 20:24 |
| odyssey4me | the hosts should be fine | 20:24 |
| SamYaple | is lxc-containers-destroy a playbook? | 20:25 |
| odyssey4me | yep - you'll find it in /opt/openstack-ansible/playbooks | 20:25 |
| Attilla | odyssey4me: And I thought I was online a lot ;-) | 20:26 |
| SamYaple | odyssey4me: thanks, will be back in a bit should more trouble arise | 20:27 |
| *** galstrom_zzz is now known as galstrom | 20:27 | |
| odyssey4me | cloudnull Matias oh dear: http://logs.openstack.org/05/370405/1/check/gate-openstack-ansible-openstack-ansible-aio-ubuntu-trusty/a62f511/console.html#_2016-09-14_20_17_40_862812 | 20:27 |
| odyssey4me | Attilla tomorrow is the deadline to request the first RC, so today will be a long hour day | 20:28 |
| Attilla | odyssey4me: good luck, I have to present at OpenstackUK | 20:28 |
| Attilla | tomorrow | 20:28 |
| openstackgerrit | Steve Lewis (stevelle) proposed openstack/openstack-ansible-os_gnocchi: Install required packages in identity only mode https://review.openstack.org/370415 | 20:29 |
| Matias | wat | 20:29 |
| cloudnull | that looks like an ansible trace | 20:29 |
| *** galstrom is now known as galstrom_zzz | 20:30 | |
| cloudnull | like "shit broke son" | 20:30 |
| hughmFLEXin | lmao | 20:30 |
| Matias | is that a symbolic link? | 20:30 |
| cloudnull | i believe so . | 20:31 |
| cloudnull | which should be ok because we added follow | 20:31 |
| Matias | it appears to be a link to an inexistant file | 20:31 |
| Matias | maybe follow is what broke it | 20:31 |
| Matias | the question is why do we have a link to an inexistant file there | 20:32 |
| odyssey4me | Matias cloudnull so to clue you in - for OpenStack CI we use a pre-existing folder as a git cache to help reduce CI time | 20:32 |
| odyssey4me | the pre-existing folder is symlinked on the host to the data folder for the container, then the data folder for the container is bind mounted | 20:32 |
| odyssey4me | ie https://github.com/openstack/openstack-ansible/blob/master/playbooks/repo-server.yml#L23-L49 | 20:33 |
| cloudnull | Matias: so maybe we need to add a specific "become: true" there? | 20:35 |
| Matias | I think some file exists inside /opt/git/openstack/packetary-specs/doc/source in the gate box which is a symbolic link to ../../policy | 20:35 |
| Matias | but as ../../policy does not exist, follow: true causes the ansible task to fail | 20:36 |
| mhayden | hah, i think i just bumped into that proprietary package business | 20:36 |
| odyssey4me | ah, yes - we have no control over the data that goes into that folder | 20:37 |
| cloudnull | mhayden: is that still an issue? | 20:37 |
| cloudnull | from the fix earlier | 20:37 |
| mhayden | cloudnull: Could not find a version that satisfies the requirement nuage_openstack_neutron (from -r /var/www/repo/os-releases/13.3.1/requirements.txt (line 36)) (from versions: ) | 20:37 |
| mhayden | i don't have the fix applied, obviously | 20:37 |
| cloudnull | oh. | 20:37 |
| cloudnull | is that master? | 20:37 |
| mhayden | mitaka | 20:37 |
| cloudnull | :'( | 20:38 |
| odyssey4me | oh dear - so yes you were right cloudnull ... time for backports | 20:38 |
| Matias | odyssey4me: I think the file module has no way to skip this kind of error, perhaps the only way to get around is to remove follow: true | 20:38 |
| odyssey4me | Matias yes, or convert that task into a shell command | 20:38 |
| Matias | why do we have follow: true? because /var/www/repo/openstackgit might be a symlink? | 20:39 |
| odyssey4me | yes, because it is a symlink | 20:39 |
| openstackgerrit | Matt Thompson proposed openstack/openstack-ansible: Restart OS services when symlinks are created https://review.openstack.org/370418 | 20:39 |
| openstackgerrit | Nolan Brubaker proposed openstack/openstack-ansible: Add lxc_host dynamic group to inventory. https://review.openstack.org/369599 | 20:39 |
| odyssey4me | that's why https://github.com/openstack/openstack-ansible-repo_server/commit/f1705ace9f7c1297db5a1432616f3e055a31adf4 added a condition to check what sort of directory it is | 20:39 |
| Matias | does the realpath come into the stat? | 20:40 |
| Matias | we could recurse into the realpath without follow | 20:40 |
| odyssey4me | so we should add that condition back, and do something different for a symlink - or replace it all with something that works with both options | 20:40 |
| odyssey4me | being able to use a pre-cached git repo is important for offline installs, or staged repositories | 20:41 |
| odyssey4me | and useful for gating too | 20:41 |
| Matias | there is a realpath filter in jinja2 | 20:41 |
| Matias | we could use it | 20:41 |
| odyssey4me | realpath sounds like it may be an option | 20:42 |
| odyssey4me | no idea whether it is picked up or not | 20:42 |
| Matias | we don't really need to follow more than one level of symlinks | 20:42 |
| odyssey4me | lnk_source | 20:42 |
| odyssey4me | no we don't | 20:42 |
| *** gouthamr has joined #openstack-ansible | 20:44 | |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible-plugins: Fix proprietary regression in py_pkgs.py https://review.openstack.org/370420 | 20:44 |
| Matias | let me do a quick PR | 20:44 |
| odyssey4me | thanks Matias | 20:44 |
| odyssey4me | Matias we need to cater for conditions where the dir doesn't exist, the dir exists and is a folder, and the dir exists but is a link | 20:44 |
| odyssey4me | cloudnull it may exist in liberty too? | 20:44 |
| cloudnull | yea... | 20:45 |
| cloudnull | was just about to do it there too | 20:45 |
| odyssey4me | cloudnull are we able to ditch the distro package stuff out of that review? | 20:46 |
| cloudnull | we could... | 20:46 |
| odyssey4me | this needs to be as surgical as possible | 20:46 |
| cloudnull | i'll need to rewrite parts of the patch | 20:47 |
| odyssey4me | it looks like I have test vars to validate for the repo_build role in mitaka, so we do have some regression protection there | 20:47 |
| openstackgerrit | Steve Lewis (stevelle) proposed openstack/openstack-ansible-os_gnocchi: Install required packages in identity only mode https://review.openstack.org/370415 | 20:47 |
| odyssey4me | if you can keep the key sorting in, that'd be helpful for this and future patches | 20:48 |
| cloudnull | i'd rather do a full backport for mitaka | 20:48 |
| cloudnull | and bring in only the bits needed for liberty | 20:48 |
| odyssey4me | yeah, maybe we should just do that | 20:48 |
| Matias | realpath does not assume the path already exists | 20:48 |
| Matias | so it seems like a good option | 20:48 |
| openstackgerrit | Steve Lewis (stevelle) proposed openstack/openstack-ansible-os_gnocchi: Install required packages in identity only mode https://review.openstack.org/370415 | 20:48 |
| odyssey4me | ok, is that other review a full backport? | 20:48 |
| cloudnull | it is | 20:48 |
| odyssey4me | ok, let me put up a test PR for the repo build to validate the results | 20:48 |
| cloudnull | im looking at liberty now to see if its even needed. | 20:49 |
| cloudnull | its not needed in liberty | 20:50 |
| cloudnull | liberty doesn't have propriatary packages as a var | 20:50 |
| cloudnull | so nothing to do there. | 20:50 |
| odyssey4me | cloudnull doesn't seem so - we only have the optional stuff: https://github.com/openstack/openstack-ansible/blob/liberty/playbooks/roles/os_neutron/defaults/main.yml#L380 | 20:51 |
| cloudnull | so i think liberty is all good | 20:51 |
| odyssey4me | yeah, good thing | 20:51 |
| odyssey4me | mhayden a review of https://review.openstack.org/365458 would be marvellous | 20:52 |
| mhayden | oh just a little one-liner there, eh? | 20:53 |
| odyssey4me | just a data validation of a patch that's already in the plugins repo :p | 20:54 |
| adreznec | Just a nice, quick review :) | 20:54 |
| mattt | mhayden: have fun! | 20:54 |
| mattt | :) | 20:54 |
| openstackgerrit | Steve Lewis (stevelle) proposed openstack/openstack-ansible: Gnocchi identities created before Swift playbook https://review.openstack.org/366931 | 20:55 |
| mhayden | odyssey4me: this is just so we have good, representative data for testing, right? | 20:56 |
| mhayden | i figured this wasn't manually generated | 20:57 |
| *** johnmilton has joined #openstack-ansible | 20:58 | |
| *** can8dnSix has quit IRC | 20:59 | |
| odyssey4me | mhayden yes - I generated it from the plugins changes for that patch | 20:59 |
| mhayden | got it | 20:59 |
| odyssey4me | it was a tool to validate that the output from the patches was good | 20:59 |
| odyssey4me | I'm about to do the same for mitaka | 20:59 |
| mhayden | odyssey4me: doesn't https://review.openstack.org/370420 need to go in first for mitakatown? | 21:01 |
| odyssey4me | mhayden nope - I'm putting a review up to show what the data looks like when we use that patch | 21:01 |
| *** TxGirlGeek has joined #openstack-ansible | 21:01 | |
| odyssey4me | two side-by-side reviews for validation that we aren't going to break things | 21:01 |
| mhayden | odyssey4me: you are a smart feller | 21:02 |
| mhayden | and i'm not saying that because cloudnull told me to | 21:02 |
| *** hybridpollo has quit IRC | 21:03 | |
| *** thorst_ has joined #openstack-ansible | 21:03 | |
| *** jheroux has quit IRC | 21:03 | |
| *** admin0 has joined #openstack-ansible | 21:04 | |
| openstackgerrit | Michael Gugino proposed openstack/openstack-ansible-ceph_client: Support deployment of extra ceph.conf files https://review.openstack.org/370431 | 21:04 |
| * mhayden is headed out for soccer coaching time | 21:05 | |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-repo_build: Update py_pkgs output data for updated lookup plugin https://review.openstack.org/370432 | 21:05 |
| *** psilvad has quit IRC | 21:05 | |
| odyssey4me | ugh, it's not pretty because the previous data wasn't sorted - and it's all in json :/ | 21:06 |
| odyssey4me | maybe I should just convert this to yaml too | 21:06 |
| *** askb has joined #openstack-ansible | 21:06 | |
| *** thorst_ has quit IRC | 21:09 | |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-repo_build: Update py_pkgs output data for updated lookup plugin https://review.openstack.org/370432 | 21:10 |
| odyssey4me | that looks better | 21:11 |
| odyssey4me | cloudnull looks good to me - we're even fixing some other regressions (lower case package names) | 21:12 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-plugins: Fix proprietary regression in py_pkgs.py https://review.openstack.org/370420 | 21:13 |
| openstackgerrit | Merged openstack/openstack-ansible-repo_build: Update py_pkgs output data for binary package lookup https://review.openstack.org/365458 | 21:13 |
| odyssey4me | cloudnull perhaps you can validate my thoughts that the data output looks good? https://review.openstack.org/370432 | 21:14 |
| odyssey4me | then we'll need one other core to make a call on both patches - I've added a dep from the one to the other | 21:15 |
| Matias | odyssey4me: {{ _git_folder.stat.exists and _git_folder.stat.islnk | ternary(_git_folder.stat.lnk_source, repo_service_home_folder + '/repo/openstackgit') }} | 21:15 |
| Matias | odyssey4me: should I break this into multiple lines? | 21:15 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-repo_build: Update py_pkgs output data for updated lookup plugin https://review.openstack.org/370432 | 21:15 |
| Matias | code style question | 21:15 |
| odyssey4me | Matias you'll need to add () around the stuff before the ternary filter | 21:16 |
| odyssey4me | but no - one line should do | 21:16 |
| Matias | odyssey4me: it is working without (), I think | has less precedence than and | 21:16 |
| odyssey4me | yeah, we've seen some odd behaviour between versions - rather be safe | 21:17 |
| Matias | ok | 21:17 |
| odyssey4me | it's possibly working because you have a link - try it without a folder at all :) | 21:17 |
| odyssey4me | but anyway - short version, add the () | 21:17 |
| Matias | nevermind, my test with nonexistant dir just failed | 21:17 |
| Matias | () is really needed | 21:17 |
| hughmFLEXin | Matias: does this look correct? https://gist.github.com/bbyhuy/ae3a3a0e6aca420a904ac48ccca60529 | 21:18 |
| Matias | odyssey4me: LOL it created a dir named /root/False | 21:19 |
| Matias | hughmFLEXin: yes, but weren't you going to use dhcp? | 21:19 |
| hughmFLEXin | sorta, its a DHCP that allows for assign static ip to nodes based on mac-addr | 21:21 |
| Matias | hmm cobbler does just like that | 21:21 |
| hughmFLEXin | yea im using Stacki | 21:21 |
| *** galstrom_zzz is now known as galstrom | 21:21 | |
| Matias | I choose cobbler because there was a ready-to-go inventory script for ansible | 21:21 |
| Matias | however I needed to modify the inventory script anyway because it didn't implement variable inheritance | 21:22 |
| hughmFLEXin | in your setup do you use br-vlan or br-vxlan for your internal openstack network | 21:22 |
| hughmFLEXin | since br-ex is the floating net | 21:22 |
| Matias | I currently don't use br-vlan | 21:22 |
| Matias | it is there in case we want to create a pure-vlan network | 21:23 |
| Matias | but by default neutron allocs the project networks into vxlan tunnels | 21:23 |
| *** galstrom is now known as galstrom_zzz | 21:24 | |
| hughmFLEXin | hmm, so i created an internal net using vxlan and would assuming i can at least ping it from my compute node | 21:24 |
| hughmFLEXin | wonder if im missing something | 21:24 |
| Matias | ping what? the internal net's router? | 21:24 |
| hughmFLEXin | internal net vm | 21:25 |
| Matias | ping one vm from another? | 21:25 |
| *** asettle has quit IRC | 21:25 | |
| hughmFLEXin | from compute to vm | 21:25 |
| Matias | no | 21:26 |
| Matias | compute does not see the vm's network | 21:26 |
| Matias | it is just some flat tun | 21:26 |
| *** johnmilton has quit IRC | 21:26 | |
| hughmFLEXin | ah | 21:26 |
| Matias | you might ping it from the neutron-agents container | 21:26 |
| Matias | but you would need to go inside the network namespace | 21:26 |
| Matias | using "ip netns NAME exec" | 21:27 |
| Matias | it creates one namespace per each project network | 21:27 |
| Matias | if you are running HA, you need to check which one of the neutron-agent currently has the master router | 21:27 |
| Matias | then you can get into the namespace and ping | 21:28 |
| Matias | it is easier to just spawn 2 vms and ping between them | 21:28 |
| Matias | unless you are debugging some issue | 21:28 |
| odyssey4me | cloudnull yep, it looks like it works | 21:28 |
| cloudnull | which? | 21:28 |
| hughmFLEXin | hmm k let me add br-ex to the config and re-run lxc-create and try again | 21:29 |
| hughmFLEXin | with floating | 21:29 |
| odyssey4me | cloudnull https://review.openstack.org/#/c/370432/ | 21:29 |
| Matias | odyssey4me: one last test before submitting the PR: ln -s /nonexistant /var/www/repo/openstackgitx/lolol | 21:29 |
| odyssey4me | the updated repo data output | 21:29 |
| cloudnull | excellent | 21:29 |
| Matias | hughmFLEXin: you will not be able to ping a floating IP unless you add a security group allowing for ICMP | 21:29 |
| hughmFLEXin | yea that is already added | 21:29 |
| odyssey4me | interesting to see that mitaka takes around 12 mins while master takes around 8 mins | 21:30 |
| openstackgerrit | Paulo Matias proposed openstack/openstack-ansible-repo_server: Fix error when the git folder contains broken symlinks https://review.openstack.org/370443 | 21:34 |
| Matias | odyssey4me: cloudnull: ^ please do the needful :P | 21:35 |
| odyssey4me | Matias yeah, will monitor the CI output | 21:35 |
| Matias | I hope it solves the issue | 21:35 |
| cloudnull | done. | 21:37 |
| cloudnull | :) | 21:37 |
| Matias | lol thanks | 21:38 |
| odyssey4me | cloudnull any thoughts on https://review.openstack.org/369485 ? | 21:40 |
| cloudnull | thank you Matias. it'll be nice to see this fixed. | 21:40 |
| Matias | i'm going out for the evening.. need to do some tasks unrelated to openstack tonight.. I'll try to come back later, otherwise see you tomorrow.. still need to open that rabbit+ssl bug, will that as soon as possible | 21:41 |
| Matias | thank you guys | 21:41 |
| odyssey4me | thanks Matias :) | 21:41 |
| odyssey4me | have a good evening | 21:41 |
| hughmFLEXin | thx Matias bb | 21:41 |
| cloudnull | odyssey4me: LGTM | 21:42 |
| odyssey4me | cloudnull so I was thinking - a full metal scenario would be good | 21:42 |
| cloudnull | I agree | 21:43 |
| odyssey4me | maybe do a split out of swift so that we have one scenario with swift+telemetry and another with compute + tlemetry | 21:43 |
| odyssey4me | we can do glance + swift in the glance role | 21:43 |
| cloudnull | makes sense to me | 21:44 |
| odyssey4me | it seems like the main code path to do in the integrated gate instead of the roles is the telemetry integration | 21:44 |
| cloudnull | i'd like to have swift basically always be tested. | 21:45 |
| * cloudnull is a swift fanboi | 21:45 | |
| *** michaelgugino has quit IRC | 21:45 | |
| *** AnarchyAo has quit IRC | 21:45 | |
| odyssey4me | doing those above would mean we can leave the default 'aio' scenario for dev only testing and drop it from the gate | 21:45 |
| cloudnull | the more scenarios the better imo. | 21:45 |
| odyssey4me | hopefully that will then bring us under the 90 min build time always | 21:45 |
| cloudnull | ++ | 21:45 |
| odyssey4me | yeah, then we can also do magnum, sahara and others | 21:46 |
| cloudnull | +9000 | 21:47 |
| cloudnull | a zaqar gate would be outstanding | 21:47 |
| odyssey4me | argh - a centralised test repo failure? http://logs.openstack.org/43/370443/1/check/gate-openstack-ansible-repo_server-ansible-func-ubuntu-xenial/0a47e8a/console.html#_2016-09-14_21_46_19_911771 | 21:47 |
| *** admin0 has quit IRC | 21:48 | |
| cloudnull | odyssey4me: looks lke an undefined var | 21:49 |
| odyssey4me | ah, different failure: http://logs.openstack.org/43/370443/1/check/gate-openstack-ansible-repo_server-ansible-func-ubuntu-xenial/0a47e8a/console.html#_2016-09-14_21_45_18_633769 | 21:50 |
| odyssey4me | it worked on trusty, so probably a transient failure | 21:51 |
| odyssey4me | waiting for centos to build, then we can recheck | 21:51 |
| odyssey4me | I do wish that centos minimum size would be increased | 21:52 |
| odyssey4me | minimum nodepool size I mean | 21:52 |
| *** TxGirlGeek has quit IRC | 21:53 | |
| *** asettle has joined #openstack-ansible | 21:53 | |
| *** KLevenstein has quit IRC | 21:56 | |
| mrda | Morning OSA | 22:01 |
| odyssey4me | o/ mrda | 22:02 |
| odyssey4me | any progress on figuring out how to test Ironic in the gate? | 22:02 |
| odyssey4me | we have both role and integrated gate scenario testing available - so I'm quite keen to get all the things tested per commit | 22:02 |
| mrda | o/ | 22:02 |
| cloudnull | ++ | 22:03 |
| cloudnull | that'd be awesome | 22:03 |
| nishpatwa_ | Hi I am facing Gate failure | 22:04 |
| nishpatwa_ | https://review.openstack.org/#/c/370187/ | 22:04 |
| nishpatwa_ | Is there something I can do? | 22:04 |
| *** jperry has quit IRC | 22:05 | |
| csmart | mrda: \o | 22:06 |
| mrda | o/ | 22:06 |
| odyssey4me | nishpatwa_ just hang ten - we have a fix busy being tested right now which should unblock that | 22:06 |
| nishpatwa_ | Sure Thanks | 22:07 |
| odyssey4me | nishpatwa_ I'll do a recheck on your patch once it's unblocked | 22:07 |
| nishpatwa_ | Okay sure odyssey4me | 22:07 |
| *** thorst has joined #openstack-ansible | 22:08 | |
| *** asettle has quit IRC | 22:12 | |
| *** spedione is now known as spedione|AWAY | 22:13 | |
| *** thorst has quit IRC | 22:15 | |
| openstackgerrit | Merged openstack/openstack-ansible: Restart OS services when symlinks are created https://review.openstack.org/370418 | 22:17 |
| *** galstrom_zzz is now known as galstrom | 22:17 | |
| *** galstrom is now known as galstrom_zzz | 22:18 | |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible-lxc_container_create: Change the container prep using more intelligent commands https://review.openstack.org/354120 | 22:20 |
| stevelle | cloudnull: stuck on https://review.openstack.org/370352 -- logs http://paste.openstack.org/show/576564/ any ideas? | 22:26 |
| *** itsuugo has quit IRC | 22:26 | |
| odyssey4me | stevelle so ceilometer did a wholesale switch? | 22:27 |
| stevelle | yeah, that was the last service in the group to switch | 22:27 |
| odyssey4me | maybe dstanek can help? or andymccr if he's around? | 22:27 |
| *** itsuugo has joined #openstack-ansible | 22:28 | |
| stevelle | looking at the find_file method it's probably complaining because there is no _namespace | 22:29 |
| odyssey4me | apache2/uwsgi seems rather heavy handed - why not just switch th uwsgi only? | 22:30 |
| odyssey4me | uwsgi only is kinda equivalent to the current service model I would guess? | 22:30 |
| stevelle | odyssey4me: I'm going with mod_wsgi because the pattern is pervasive in osa | 22:31 |
| odyssey4me | yeah, fair enough | 22:32 |
| *** hj-hpe has quit IRC | 22:33 | |
| odyssey4me | stevelle will it absolutely no longer work with the current model for Newton? | 22:33 |
| odyssey4me | or is this work for post Newton? | 22:33 |
| stevelle | odyssey4me: they murdereated werkzeug | 22:34 |
| odyssey4me | oh, how nice of them | 22:34 |
| *** sdake_ has quit IRC | 22:40 | |
| openstackgerrit | Merged openstack/openstack-ansible-galera_server: Prevent galera_cluster_name from changing. https://review.openstack.org/370133 | 22:47 |
| *** hblah has joined #openstack-ansible | 22:47 | |
| *** itsuugo has quit IRC | 22:52 | |
| *** itsuugo has joined #openstack-ansible | 22:53 | |
| *** adrian_otto has joined #openstack-ansible | 22:56 | |
| *** johnmilton has joined #openstack-ansible | 22:57 | |
| *** jamesdenton has quit IRC | 23:01 | |
| *** weezS has quit IRC | 23:02 | |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible-lxc_container_create: Change the container prep using more intelligent commands https://review.openstack.org/354120 | 23:03 |
| *** itsuugo has quit IRC | 23:10 | |
| *** itsuugo has joined #openstack-ansible | 23:11 | |
| *** thorst has joined #openstack-ansible | 23:12 | |
| openstackgerrit | Merged openstack/openstack-ansible-repo_server: Fix error when the git folder contains broken symlinks https://review.openstack.org/370443 | 23:14 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible: [DOCS] Added HAProxy example to Production environment document. https://review.openstack.org/370187 | 23:14 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible: [TESTING] Ansible Devel https://review.openstack.org/351165 | 23:16 |
| *** thorst has quit IRC | 23:20 | |
| *** pmannidi has joined #openstack-ansible | 23:20 | |
| *** johnmilton has quit IRC | 23:20 | |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible: Remove assumption that the neutron_lbaas var is set https://review.openstack.org/368905 | 23:21 |
| *** jamielennox is now known as jamielennox|away | 23:22 | |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-tests: Fix container destroy host group https://review.openstack.org/370282 | 23:24 |
| *** johnmilton has joined #openstack-ansible | 23:25 | |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible-lxc_container_create: Change the container prep using more intelligent commands https://review.openstack.org/354120 | 23:25 |
| odyssey4me | palendae you around? see comments in https://review.openstack.org/369599 | 23:25 |
| *** kylek3h has quit IRC | 23:27 | |
| stevelle | odyssey4me: https://review.openstack.org/#/c/369485/ seems like it will have some significant upgrade implications | 23:32 |
| stevelle | at least for environments that have tailored their inventory at all | 23:33 |
| odyssey4me | stevelle I tried to ensure that it's additive, and takes nothing away. Have I missed anything? | 23:33 |
| odyssey4me | any existing env.d should work - we'll find out when doing upgrade tests I guess | 23:34 |
| stevelle | probably missed a release note but it's hard to say for sure if anything else is without spending a lot of time on it | 23:35 |
| odyssey4me | stevelle yeah, no release note there - kinda on purpose because it's additive | 23:37 |
| odyssey4me | also because I don't like the new group names and hope we can ditch them asap | 23:37 |
| stevelle | well that's a factor at least | 23:38 |
| odyssey4me | I'm thinking of them as 'experimental' in the hope that palendae can make a magic pill in the inventory | 23:38 |
| odyssey4me | I dunno if you noticed the chat earlier about the issue I found | 23:38 |
| odyssey4me | https://bugs.launchpad.net/openstack-ansible/+bug/1623584 | 23:38 |
| openstack | Launchpad bug 1623584 in openstack-ansible "Dynamic inventory clash between host group name and container group name" [Undecided,Confirmed] - Assigned to Nolan Brubaker (nolan-brubaker) | 23:38 |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible-lxc_container_create: Change the container prep using more intelligent commands https://review.openstack.org/354120 | 23:39 |
| stevelle | ewww | 23:39 |
| *** hblah has quit IRC | 23:40 | |
| odyssey4me | yeah, so for now I'm kinda hoping to keep the new names as dev only, although I know if we release with it we'll need a deprecation cycle | 23:40 |
| odyssey4me | Matias cloudnull good news https://review.openstack.org/370443 is working in the integrated build | 23:40 |
| odyssey4me | it takes 90 secs to complete that task in CI | 23:41 |
| odyssey4me | openstack-infra puts a *lot* of git repositories in the image | 23:41 |
| cloudnull | that aint bad at all | 23:41 |
| odyssey4me | stevelle did you notice that https://review.openstack.org/370420 depends on https://review.openstack.org/370432 ? | 23:42 |
| stevelle | odyssey4me: yeah, thought I voted on that already | 23:43 |
| * stevelle is tired | 23:43 | |
| odyssey4me | stevelle maybe the master version, this is the mitaka one | 23:43 |
| stevelle | I specifically reviewed the mitaka | 23:43 |
| stevelle | I remember giving it the eyeball since not a backport | 23:44 |
| odyssey4me | stevelle sounds like you need an afternoon nap? | 23:44 |
| stevelle | yeah, shifting my schedule to start at 7am again | 23:44 |
| stevelle | by 4:30 or so I'm dragging | 23:45 |
| odyssey4me | a mid day nap of 30-45 mins works wonders for me | 23:45 |
| odyssey4me | if I'm feeling tired I do that when working from home | 23:46 |
| stevelle | when I get my schedule adjusted for a week I'll be fine again | 23:47 |
| odyssey4me | what was your schedule before? | 23:47 |
| stevelle | started about 8:30 or so | 23:48 |
| stevelle | the killer is the morning workout before that | 23:48 |
| stevelle | earlier start means I don't get those crucial light rays in my eyeballs to get me going | 23:49 |
| cloudnull | https://review.openstack.org/#/c/354120/ -- odyssey4me stevelle | 23:51 |
| cloudnull | if you have a moment | 23:51 |
| cloudnull | now lxc container config is done using the container PID instead of the lxc module | 23:51 |
| odyssey4me | cloudnull busy looking through it - looking good so far | 23:52 |
| *** johnmilton has quit IRC | 23:55 | |
| *** johnmilton has joined #openstack-ansible | 23:56 | |
| *** itsuugo has quit IRC | 23:57 | |
| *** itsuugo has joined #openstack-ansible | 23:58 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!
