| *** sdake has quit IRC | 00:01 | |
| *** sdake has joined #openstack-ansible | 00:06 | |
| *** sdake has quit IRC | 00:06 | |
| *** sdake has joined #openstack-ansible | 00:06 | |
| *** asdf1234asdf[m] has quit IRC | 00:06 | |
| *** jhesketh has quit IRC | 00:06 | |
| *** lbragstad has quit IRC | 00:06 | |
| *** chudster_ has quit IRC | 00:06 | |
| *** jamesdenton has quit IRC | 00:06 | |
| *** senk has quit IRC | 00:06 | |
| *** lhinds has quit IRC | 00:06 | |
| *** jmnk has quit IRC | 00:06 | |
| *** flaper87 has quit IRC | 00:06 | |
| *** mudpuppy has quit IRC | 00:06 | |
| *** jiteka- has quit IRC | 00:06 | |
| *** madorn has quit IRC | 00:06 | |
| *** rickflare has quit IRC | 00:06 | |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible-ops master: Add functional tests to elk_metrics_6x https://review.openstack.org/583794 | 00:09 |
|---|---|---|
| *** linshuicheng[m] has quit IRC | 00:10 | |
| *** evrardjp[m] has quit IRC | 00:10 | |
| *** lbragstad[m] has quit IRC | 00:10 | |
| *** taildashef-work has quit IRC | 00:10 | |
| cloudnull | mnaser the change looks fine however I agree w/ logan- ++ reno or leave the default var alone | 00:10 |
| *** keithmnemonic[m] has quit IRC | 00:10 | |
| *** jothibasu[m] has quit IRC | 00:10 | |
| *** markvoelker has quit IRC | 00:10 | |
| mnaser | cloudnull: im going to check but i think i introduced this var this same cycle with another change i did | 00:11 |
| mnaser | so ill verify that because then it'd be a bit silly | 00:11 |
| *** rschulman has quit IRC | 00:11 | |
| *** csmart has quit IRC | 00:11 | |
| cloudnull | ++ | 00:11 |
| cloudnull | if that's the case, then carry on :) | 00:11 |
| mnaser | https://review.openstack.org/#/c/576883/6/defaults/main.yml | 00:11 |
| mnaser | nope its not i lied | 00:11 |
| cloudnull | unless it was a change that was backported, then i fall back to my original thought | 00:12 |
| cloudnull | oh | 00:12 |
| mnaser | voila | 00:14 |
| openstackgerrit | Mohammed Naser proposed openstack/openstack-ansible-os_cinder master: Change deprecated option iscsi_helper to target_helper https://review.openstack.org/584151 | 00:14 |
| cloudnull | lgtm | 00:14 |
| *** jhesketh has joined #openstack-ansible | 00:15 | |
| *** lbragstad has joined #openstack-ansible | 00:15 | |
| *** chudster_ has joined #openstack-ansible | 00:15 | |
| *** jamesdenton has joined #openstack-ansible | 00:15 | |
| *** senk has joined #openstack-ansible | 00:15 | |
| *** lhinds has joined #openstack-ansible | 00:15 | |
| *** jmnk has joined #openstack-ansible | 00:15 | |
| *** flaper87 has joined #openstack-ansible | 00:15 | |
| *** mudpuppy has joined #openstack-ansible | 00:15 | |
| *** jiteka- has joined #openstack-ansible | 00:15 | |
| *** madorn has joined #openstack-ansible | 00:15 | |
| *** rickflare has joined #openstack-ansible | 00:15 | |
| cloudnull | i'll bet the cinder gate is still angry | 00:15 |
| cloudnull | but the change looks good | 00:15 |
| cloudnull | :D | 00:15 |
| mnaser | ill stop making it angry | 00:16 |
| cloudnull | no, dont stop, i think it needs to be made a little more angry :) | 00:21 |
| cloudnull | i think evrardjp was working on a fix for cinderbackup | 00:21 |
| cloudnull | to make it generally happy | 00:21 |
| *** threestrands has joined #openstack-ansible | 00:23 | |
| *** threestrands has quit IRC | 00:23 | |
| *** threestrands has joined #openstack-ansible | 00:23 | |
| *** markvoelker has joined #openstack-ansible | 00:24 | |
| *** cshen has joined #openstack-ansible | 00:31 | |
| *** cshen has quit IRC | 00:36 | |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible-ops master: Add functional tests to elk_metrics_6x https://review.openstack.org/583794 | 00:36 |
| *** poopcat1 has joined #openstack-ansible | 00:44 | |
| *** asdf1234asdf[m] has joined #openstack-ansible | 00:46 | |
| *** exodusftw has quit IRC | 00:46 | |
| *** poopcat has quit IRC | 00:47 | |
| *** exodusftw has joined #openstack-ansible | 00:52 | |
| openstackgerrit | Mohammed Naser proposed openstack/openstack-ansible-os_cinder master: Change deprecated option iscsi_helper to target_helper https://review.openstack.org/584151 | 00:58 |
| *** dave-mccowan has quit IRC | 00:59 | |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible-ops master: Add functional tests to elk_metrics_6x https://review.openstack.org/583794 | 01:16 |
| *** tux_ has quit IRC | 01:33 | |
| *** chyka has joined #openstack-ansible | 01:36 | |
| *** chyka has quit IRC | 01:37 | |
| *** dave-mccowan has joined #openstack-ansible | 01:38 | |
| *** tux_ has joined #openstack-ansible | 01:43 | |
| bgmccollum | should the loop device for nspawn containers not be created if im using LXC? just noticed this hanging around... | 01:45 |
| *** coreywright has quit IRC | 01:45 | |
| *** dave-mccowan has quit IRC | 01:52 | |
| *** Bhujay has joined #openstack-ansible | 01:52 | |
| *** coreywright has joined #openstack-ansible | 02:04 | |
| *** lbragstad[m] has joined #openstack-ansible | 02:11 | |
| *** RossSchulman[m] has joined #openstack-ansible | 02:11 | |
| *** keithmnemonic[m] has joined #openstack-ansible | 02:11 | |
| *** linshuicheng[m] has joined #openstack-ansible | 02:11 | |
| *** jothibasu[m] has joined #openstack-ansible | 02:11 | |
| *** csmart[m] has joined #openstack-ansible | 02:11 | |
| *** evrardjp[m] has joined #openstack-ansible | 02:11 | |
| *** albertcard2 has joined #openstack-ansible | 02:13 | |
| *** yanyao has joined #openstack-ansible | 02:14 | |
| openstackgerrit | bhujay kumar proposed openstack/openstack-ansible-os_cloudkitty master: Correct typo in variable name https://review.openstack.org/584200 | 02:16 |
| *** poopcat1 has quit IRC | 02:16 | |
| logan- | o/ my bad if thats the case mnaser :) | 02:26 |
| *** Bhujay has quit IRC | 02:30 | |
| *** cshen has joined #openstack-ansible | 02:39 | |
| *** adrianreza has joined #openstack-ansible | 02:44 | |
| *** poopcat1 has joined #openstack-ansible | 02:44 | |
| *** cshen has quit IRC | 02:44 | |
| *** albertcard2 has quit IRC | 02:47 | |
| *** threestrands has quit IRC | 02:54 | |
| cloudnull | bgmccollum the machines.raw loop device is created regardless. | 02:57 |
| cloudnull | in lxc we use that for general image storage | 02:57 |
| bgmccollum | OIC thanks | 02:58 |
| cloudnull | in nspawn and lxc w/ the machinectl backend we actually provision the container there | 02:58 |
| cloudnull | if you want to use that directly you can, just set lxc_backend_store (i think thats the var) to machinectl | 02:58 |
| prometheanfire | cloudnull: we need to switch our default now that master is updated | 03:02 |
| cloudnull | ++ | 03:03 |
| *** taildashef-work[ has joined #openstack-ansible | 03:05 | |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible-ops master: Add functional tests to elk_metrics_6x https://review.openstack.org/583794 | 03:08 |
| *** threestrands has joined #openstack-ansible | 03:10 | |
| *** threestrands has quit IRC | 03:10 | |
| *** threestrands has joined #openstack-ansible | 03:10 | |
| *** albertcard2 has joined #openstack-ansible | 03:15 | |
| *** poopcat1 has quit IRC | 03:18 | |
| openstackgerrit | Mohammed Naser proposed openstack/openstack-ansible-os_cinder master: Change deprecated option iscsi_helper to target_helper https://review.openstack.org/584151 | 03:21 |
| openstackgerrit | Mohammed Naser proposed openstack/openstack-ansible-os_cinder master: Change deprecated option iscsi_helper to target_helper https://review.openstack.org/584151 | 03:21 |
| *** kstev has joined #openstack-ansible | 03:26 | |
| cloudnull | mnaser logan- you around ? mind giving my review a look ? | 03:31 |
| mnaser | i am around forever | 03:31 |
| cloudnull | haha | 03:31 |
| mnaser | i dont sleep on flights and im landing 6am est so | 03:31 |
| cloudnull | I know you're pain | 03:31 |
| cloudnull | where you off to ? | 03:31 |
| mnaser | so which revieww? :> | 03:31 |
| cloudnull | home ? | 03:31 |
| mnaser | back home :) | 03:31 |
| cloudnull | https://review.openstack.org/583794 | 03:31 |
| cloudnull | adding testing to the elk bits | 03:31 |
| logan- | looking | 03:32 |
| cloudnull | and that testing should be selective, only running when the elk bits are modified | 03:32 |
| mnaser | oh thats interesting | 03:32 |
| logan- | see my comment in there btw regarding gate pipeline | 03:32 |
| * mnaser the ':' in the job name, not a typo? | 03:32 | |
| cloudnull | nope | 03:33 |
| logan- | mnaser: i was surprised that worked too lol | 03:33 |
| mnaser | thats a super interesting way of 'namespacing' | 03:33 |
| mnaser | logan-: time to try csrf with job names? | 03:33 |
| mnaser | haha | 03:33 |
| cloudnull | i was trying to make a way to signify which thing was running under the ops repo namespace | 03:33 |
| mnaser | i like it | 03:34 |
| mnaser | cloudnull: by the way, i tried using systemd_networkd to configure networking for physical infra in order to deploy OS on top | 03:34 |
| mnaser | but .. weird things started happening for some reason | 03:34 |
| mnaser | example: we have lxcbr0 and br-mgmt, traffic egress lxcbr0, then out br-mgmt (masqueraded) | 03:35 |
| mnaser | it comes back in from br-mgmt and then instead of being sent off using lxcbr0, it actually gets sent to the vm over br-mgmt... | 03:35 |
| mnaser | arp and ip routes point to stuff should be going through lxcbr0.. idk | 03:35 |
| cloudnull | ah logan- so I should add a gate pipeline then | 03:36 |
| logan- | yup | 03:36 |
| * cloudnull will do | 03:36 | |
| cloudnull | mnaser that's odd ? | 03:37 |
| mnaser | very odd | 03:37 |
| mnaser | ah thats a good catch | 03:37 |
| mnaser | logan-: is good at reviews | 03:37 |
| cloudnull | I have networkd running on my hardware lab using https://gist.github.com/cloudnull/69bde16011d1ea94edf926252ea93933#file-playbook-networkd-host-interfaces-yml | 03:38 |
| logan- | been doing alot of zuul lately heh | 03:38 |
| cloudnull | the hostvars are like so https://gist.github.com/cloudnull/69bde16011d1ea94edf926252ea93933#file-openstack_user_config-yml-L417-L429 | 03:38 |
| cloudnull | its not a great way to do all the things. but its working | 03:38 |
| cloudnull | that said tux_ was running into similar issues with networkd on cent7 | 03:39 |
| cloudnull | my lab is running xenial | 03:39 |
| *** kstev has quit IRC | 03:39 | |
| tux_ | cloudnull: networkd doesn't support lxc bridge with CentOS 7 because os systemd bug | 03:40 |
| tux_ | cloudnull: only solution is to upgrade systemd to latest version which is not possible on CentOS7 | 03:41 |
| tux_ | mnaser: ^^ | 03:41 |
| mnaser | oooooh | 03:41 |
| mnaser | thats interesting | 03:41 |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible-ops master: Add functional tests to elk_metrics_6x https://review.openstack.org/583794 | 03:41 |
| tux_ | mnaser: if you using CentOS7 please use ifcfg-* | 03:41 |
| mnaser | tux_: do you have a bug or some keryowrds for me to google | 03:41 |
| mnaser | yeah i ended up writing stuff to use ifcfg- | 03:41 |
| tux_ | i spent lots of hour to fix but no luck | 03:42 |
| cloudnull | however we are using networkd in the integrated gate. | 03:42 |
| mnaser | i'd like to track the bug or maybe push to get the fix backported | 03:42 |
| cloudnull | w/ cent7 | 03:42 |
| cloudnull | so maybe there's a bug, but it is working in the main osa gate | 03:43 |
| mnaser | yeah after many hours i got sick of looking at it :( | 03:43 |
| cloudnull | and within the specific nspawn gates. | 03:43 |
| tux_ | cloudnull: are you saying networkd working with CentOS 7 with LXC bridge ? | 03:43 |
| cloudnull | yes | 03:43 |
| tux_ | cloudnull: how????? | 03:43 |
| logan- | i was looking at feeding osa bridges and stuff to cloud-init config drives in centos but https://bugzilla.redhat.com/show_bug.cgi?id=1536946 nixed that :( | 03:43 |
| openstack | bugzilla.redhat.com bug 1536946 in cloud-init "cloud-init can't handle network config data version 1 for bond" [Medium,Assigned] - Assigned to eterrell | 03:43 |
| cloudnull | thats how all of the main osa tests are run | 03:43 |
| tux_ | cloudnull: tell me your kernel version and systemd version | 03:44 |
| cloudnull | tux_ i don't know that combo in the gate sadly. | 03:45 |
| tux_ | cloudnull: i spend 3 days to find out and it didn't work.. (only issue is lxc bridge) if i don't use lxcbridge everything works great with networkd | 03:45 |
| mnaser | tux_: the answer is latest everything | 03:45 |
| cloudnull | hum | 03:45 |
| tux_ | I have latest version CentOS 7.5 but there is no upgrade for systemd, CentOS 7 still using systemd 2.19 | 03:46 |
| tux_ | i tired to upgrade systemd but found there are so many dependencies which makes upgrade impossible | 03:47 |
| tux_ | cloudnull: when you saying "integrated gate" what does that means ? is this something different technology ? | 03:48 |
| cloudnull | tux_ all changes that happen in the openstack-ansible repo is the integrated gate | 03:48 |
| cloudnull | like https://review.openstack.org/#/c/582814 | 03:48 |
| tux_ | cloudnull: so look like its dummy network which run inside OS | 03:49 |
| cloudnull | while the cent7 gate is failing, its failing on tempest not on a networking issue | 03:50 |
| cloudnull | http://logs.openstack.org/14/582814/1/check/openstack-ansible-deploy-aio_metal-centos-7/dae3b75/job-output.txt.gz#_2018-07-16_02_43_38_996451 | 03:50 |
| cloudnull | yes. its running dummy interfaces for everything that would be physical. | 03:50 |
| cloudnull | the lxcbr0 interface is managed by the lxc-host role | 03:50 |
| tux_ | cloudnull: it may work with internal component but it doesn't support with external communication, I had same issue everything was pinging internal but when my container trying to talk about side it failed to ping | 03:55 |
| tux_ | it was very odd issue after banging my head multiple time i switch to ifcfg-* and i am happy now | 03:55 |
| cloudnull | sadly i dont have a cluster of cent7 nodes to try it. | 03:56 |
| tux_ | may be i will try again and see if i can reproduce that error again | 03:56 |
| tux_ | even i talked to systemd developer and they said systemd 2.19 is very buggy and stay away from it is possible | 03:57 |
| cloudnull | i'd be happy to try and join a screen-share / webex / whatever to try and figure that out on cent | 03:57 |
| cloudnull | that very well could be | 03:57 |
| tux_ | cloudnull: https://maciej.lasyk.info/2016/Dec/16/systemd-231-latest-in-centos-7-thx-to-facebook/ | 03:58 |
| -tux_- So Centos7 currently has systemd version 219 installed which was released on 2015-02-16 (see NEWS). | 03:58 | |
| -tux_- This is a huge problem, as we miss a lot of very important functions related to journald, networkd, machinectl, systemd-nspawn and so on. | 03:58 | |
| *** epalper has joined #openstack-ansible | 03:58 | |
| cloudnull | ++ | 03:58 |
| tux_ | cloudnull: lets do one thing.. i will setup CentOS 7 on my LAB and give it a try to networkd and will share result here.. | 04:00 |
| cloudnull | ++ let us know if there's anything we can do to help | 04:02 |
| tux_ | sure i will, i will be happy if we can fix this on CentOS7 because i want to use networkd badly | 04:03 |
| cloudnull | though if cent7 has these issues w/ systemd there's not a lot we can do. | 04:03 |
| cloudnull | ++ its pretty great :) | 04:03 |
| cloudnull | on xenial that is | 04:03 |
| tux_ | I am 100% sure its systemd issue but worth trying may be i had bad day when i pick that box | 04:03 |
| *** albertcard2 has quit IRC | 04:06 | |
| *** udesale has joined #openstack-ansible | 04:07 | |
| *** mattinator has quit IRC | 04:09 | |
| tux_ | cloudnull: Good night!! i will catch you in morning...! | 04:12 |
| cloudnull | take care | 04:12 |
| *** Bhujay has joined #openstack-ansible | 04:14 | |
| *** yanyao has quit IRC | 04:22 | |
| *** epalper has quit IRC | 04:24 | |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible-ops master: Add functional tests to elk_metrics_6x https://review.openstack.org/583794 | 04:36 |
| *** threestrands has quit IRC | 04:40 | |
| cloudnull | logan- mnaser thanks for having a look at that | 04:40 |
| * cloudnull fingers crossed it gates and collects logs :) | 04:40 | |
| openstackgerrit | ziyu proposed openstack/openstack-ansible-os_sahara master: Add the project source code repository in README https://review.openstack.org/584220 | 04:47 |
| openstackgerrit | ziyu proposed openstack/openstack-ansible-rsyslog_server master: Add the project source code repository in README https://review.openstack.org/584221 | 04:48 |
| openstackgerrit | ziyu proposed openstack/openstack-ansible-rsyslog_server master: Add the project source code repository in README https://review.openstack.org/584221 | 04:50 |
| openstackgerrit | ziyu proposed openstack/openstack-ansible-os_sahara master: Add the project source code repository in README https://review.openstack.org/584220 | 04:51 |
| openstackgerrit | ziyu proposed openstack/openstack-ansible-os_neutron master: Add the project source code repository in README https://review.openstack.org/584222 | 04:55 |
| openstackgerrit | ziyu proposed openstack/openstack-ansible-os_nova master: Add the project source code repository in README https://review.openstack.org/584224 | 05:01 |
| *** tux_ has quit IRC | 05:03 | |
| openstackgerrit | ziyu proposed openstack/openstack-ansible-os_barbican master: Add the project source code repository in README https://review.openstack.org/584225 | 05:05 |
| openstackgerrit | ziyu proposed openstack/openstack-ansible-os_ironic master: Add the project source code repository in README https://review.openstack.org/584226 | 05:09 |
| *** pcaruana has joined #openstack-ansible | 05:09 | |
| *** cshen has joined #openstack-ansible | 05:10 | |
| openstackgerrit | ziyu proposed openstack/openstack-ansible-os_magnum master: Add the project source code repository in README https://review.openstack.org/584227 | 05:12 |
| *** taseer1 has joined #openstack-ansible | 05:15 | |
| *** cshen has quit IRC | 05:15 | |
| *** taseer2 has joined #openstack-ansible | 05:16 | |
| *** Taseer has quit IRC | 05:16 | |
| openstackgerrit | ziyu proposed openstack/openstack-ansible-os_swift master: Add the project source code repository in README https://review.openstack.org/584228 | 05:16 |
| *** taseer1 has quit IRC | 05:19 | |
| *** yanyao has joined #openstack-ansible | 05:29 | |
| *** cshen has joined #openstack-ansible | 05:31 | |
| *** taseer3 has joined #openstack-ansible | 05:31 | |
| *** yanyao has quit IRC | 05:33 | |
| *** taseer2 has quit IRC | 05:34 | |
| *** cshen has quit IRC | 05:35 | |
| *** taseer3 is now known as Taseer | 05:46 | |
| jrosser | cloudnull: those networkd issues feel a little like rp_filter? | 06:03 |
| *** epalper has joined #openstack-ansible | 06:04 | |
| *** taseer1 has joined #openstack-ansible | 06:06 | |
| *** Taseer has quit IRC | 06:08 | |
| *** taseer2 has joined #openstack-ansible | 06:09 | |
| *** taseer1 has quit IRC | 06:11 | |
| *** lvdombrkr has joined #openstack-ansible | 06:22 | |
| *** taseer2 is now known as Taseer | 06:25 | |
| mnaser | jrosser: rp_filter is the reason why it isnt working but why does linux do this weird routing in the first place | 06:27 |
| *** udesale_ has joined #openstack-ansible | 06:28 | |
| *** udesale has quit IRC | 06:30 | |
| *** taseer1 has joined #openstack-ansible | 06:36 | |
| *** cshen has joined #openstack-ansible | 06:38 | |
| *** Taseer has quit IRC | 06:39 | |
| *** lvdombrkr89 has joined #openstack-ansible | 06:49 | |
| *** lvdombrkr has quit IRC | 06:51 | |
| *** taseer2 has joined #openstack-ansible | 06:59 | |
| *** taseer1 has quit IRC | 07:02 | |
| jrosser | mnaser: it is odd, theres no ambiguity in the ip address scheme i assume? | 07:06 |
| mnaser | jrosser: nope, the br-mgmt range was 172.24.0.0/24 and lxcbr0 was the usual 10.something | 07:06 |
| *** DanyC has joined #openstack-ansible | 07:07 | |
| jrosser | and lxcbr nats to the br-mgmt range? | 07:08 |
| hwoarang | good morning | 07:08 |
| openstackgerrit | Markos Chandras (hwoarang) proposed openstack/openstack-ansible master: Add support for using distribution packages for OpenStack services https://review.openstack.org/579770 | 07:08 |
| evrardjp | good morning | 07:08 |
| hwoarang | evrardjp: could you revote on ^^. conflicts again... | 07:08 |
| evrardjp | hwoarang: really? | 07:08 |
| evrardjp | Sorry | 07:08 |
| hwoarang | and i need one more core... mnaser cloudnull perhaps^ | 07:08 |
| evrardjp | have other cores voted before? | 07:09 |
| hwoarang | nope | 07:09 |
| evrardjp | no | 07:09 |
| evrardjp | ok | 07:09 |
| hwoarang | it will get in eventually :) | 07:09 |
| mnaser | ill have a look | 07:09 |
| evrardjp | I'd like at least another | 07:09 |
| mnaser | but i am sad that centos is nv | 07:09 |
| mnaser | but | 07:09 |
| evrardjp | then if you need to rebase I will consider the work ok. | 07:09 |
| odyssey4me | o/ | 07:09 |
| mnaser | im sure we'll get that figured out | 07:09 |
| evrardjp | mnaser: that can be improved | 07:09 |
| evrardjp | later | 07:09 |
| evrardjp | else the patch will become huge :) | 07:09 |
| hwoarang | it nv despite this patch | 07:09 |
| hwoarang | *it is | 07:09 |
| evrardjp | MVP :D | 07:09 |
| mnaser | the work hwoarang has done and involving centos is enough already | 07:10 |
| evrardjp | I am not explaining myself right | 07:10 |
| *** poul has joined #openstack-ansible | 07:10 | |
| evrardjp | ignore my comment it will be easier | 07:10 |
| *** poul has quit IRC | 07:11 | |
| *** poul has joined #openstack-ansible | 07:11 | |
| *** poul_ has joined #openstack-ansible | 07:12 | |
| hwoarang | yeah for centos, it would be nice if the existing basekit job becomes green. then fixing the distro one would be easier | 07:13 |
| hwoarang | both jobs are likely to face similar issues but that's just a speculation right now | 07:13 |
| openstackgerrit | Markos Chandras (hwoarang) proposed openstack/openstack-ansible master: Add support for using distribution packages for OpenStack services https://review.openstack.org/579770 | 07:14 |
| openstackgerrit | Duc Nguyen proposed openstack/openstack-ansible-os_keystone master: Replace port 35357 with 5000 https://review.openstack.org/584256 | 07:14 |
| *** poul has quit IRC | 07:15 | |
| evrardjp | I remember I had a concern a few days ago of a very far fetched thing but I can't remember it hwoarang , sorry | 07:15 |
| evrardjp | it will come back! | 07:16 |
| mnaser | +2 from me. looks good. doesnt break the world | 07:16 |
| evrardjp | #mornings | 07:16 |
| evrardjp | mnaser: thanks | 07:16 |
| mnaser | evrardjp: hwoarang ill trade reviews on https://review.openstack.org/#/c/584151/ and its parent tho :) | 07:17 |
| evrardjp | ofc | 07:17 |
| evrardjp | anything else? | 07:17 |
| mnaser | which should help the basekit centos job pass | 07:17 |
| mnaser | i think | 07:17 |
| mnaser | i dunno. im getting tired | 07:18 |
| mnaser | still cant sleep | 07:18 |
| evrardjp | mnaser: go to bed, more work will be there tomorrow anyway... : p | 07:19 |
| * mnaser is in a plane close over denver right now | 07:19 | |
| mnaser | i cant sleep in planes | 07:19 |
| evrardjp | oh ok | 07:19 |
| *** DanyC has quit IRC | 07:19 | |
| jrosser | mnaser: this makes an interesting read about traffic doing interesting things with bridges https://vincent.bernat.im/en/blog/2017-linux-bridge-isolation | 07:20 |
| evrardjp | WOW | 07:21 |
| evrardjp | SUCCESS on https://review.openstack.org/#/c/574006/ | 07:21 |
| evrardjp | just a timeout | 07:21 |
| evrardjp | wooot | 07:21 |
| evrardjp | what a "milestone"! | 07:22 |
| evrardjp | odyssey4me: can you vote on https://review.openstack.org/#/c/583955/ ? | 07:23 |
| evrardjp | jrosser: vincent bernat is a great writer :) | 07:23 |
| jrosser | yeah its an excellent series of blog posts | 07:25 |
| odyssey4me | weird - why has https://review.openstack.org/#/c/584037/3 not entered gate? | 07:26 |
| evrardjp | I started to follow vincent articles long ago when he started writing about haproxy | 07:26 |
| evrardjp | very in-depth | 07:26 |
| odyssey4me | ah, it depends on a patch which hasn't merged yet | 07:26 |
| evrardjp | yeah | 07:27 |
| evrardjp | my bad | 07:27 |
| *** tosky has joined #openstack-ansible | 07:27 | |
| odyssey4me | another set of eyes on https://review.openstack.org/584052 please to unblock cinder tests | 07:28 |
| evrardjp | odyssey4me: could you also check swift? | 07:28 |
| evrardjp | odyssey4me: swift multi region fails | 07:28 |
| odyssey4me | yeah, will get there | 07:28 |
| odyssey4me | just getting the main ones I have other patches waiting on through, then will get to that one | 07:28 |
| odyssey4me | review for https://review.openstack.org/583672 please - very csimple backport | 07:29 |
| evrardjp | yeah it's blocking httplib2 thing which was helping during my debugging of M3 | 07:29 |
| jrosser | evrardjp: https://vincent.bernat.im/en/blog/2018-multi-tier-loadbalancer | 07:30 |
| jrosser | well theres my answer for massive rgw | 07:30 |
| odyssey4me | another test inventory fix to get magnum working - reviews please: https://review.openstack.org/584042 | 07:30 |
| *** aetaric has quit IRC | 07:31 | |
| *** hamzy_ has joined #openstack-ansible | 07:31 | |
| odyssey4me | and another bit of tidying up: https://review.openstack.org/584038 | 07:31 |
| evrardjp | jrosser: OMG I am old. I remember reading https://vincent.bernat.im/en/blog/2011-lab-site-to-site-vpn | 07:31 |
| *** hamzy has quit IRC | 07:31 | |
| *** Bhujay has quit IRC | 07:32 | |
| *** Bhujay has joined #openstack-ansible | 07:33 | |
| mnaser | wow i am really potato-ing | 07:33 |
| *** udesale__ has joined #openstack-ansible | 07:33 | |
| mnaser | im asking for reviews on a change that's not mine and that i can vote for | 07:33 |
| mnaser | i'll look now odyssey4me | 07:33 |
| odyssey4me | evrardjp: mind if we merge https://review.openstack.org/#/c/583276/1 now to unblock octavia, assuming it passes gate | 07:34 |
| evrardjp | mnaser: what is potato-ing? | 07:34 |
| mnaser | becoming a potato | 07:34 |
| mnaser | i guess? | 07:34 |
| Adri2000 | hello | 07:34 |
| evrardjp | mnaser: yes but it's still not very explicit :p | 07:35 |
| odyssey4me | evrardjp: don't harrangue the fellow mid-flight with lack of sleep :p | 07:35 |
| Adri2000 | any way to use config overrides to put multiples values for the same key? as I understand https://docs.openstack.org/nova/queens/configuration/config.html#pci.alias I'd need to put multiple "alias = ..." in my nova.conf | 07:36 |
| evrardjp | odyssey4me: I am trying to stimulate the person! | 07:36 |
| *** udesale_ has quit IRC | 07:36 | |
| evrardjp | odyssey4me: don't come with fancy words like harrangue | 07:36 |
| odyssey4me | :p | 07:36 |
| evrardjp | ! | 07:36 |
| evrardjp | Adri2000: a set? | 07:36 |
| evrardjp | using "?" | 07:36 |
| mnaser | and we're flying over denver now | 07:36 |
| mnaser | anyone whos flown over denver can tell you how it's a very unpleasent experience lol | 07:36 |
| evrardjp | Adri2000: isn't this a dict ? | 07:37 |
| *** gkadam has joined #openstack-ansible | 07:37 | |
| odyssey4me | wow, it'd be good for https://docs.openstack.org/ansible-config_template/latest/ to get some examples | 07:37 |
| evrardjp | Adri2000: try dict and set, see what works for the nova | 07:37 |
| evrardjp | odyssey4me: these are generated from the code I think | 07:38 |
| evrardjp | we can change the code to add those | 07:38 |
| Adri2000 | evrardjp: in nova.conf directly I know that multiple alias = ... lines works | 07:38 |
| evrardjp | well you'd say it's always generated from the code... | 07:38 |
| evrardjp | some* | 07:38 |
| mnaser | i think you can simply add commas | 07:38 |
| mnaser | for ListOpt | 07:38 |
| mnaser | ex: alias = foo,bar | 07:39 |
| mnaser | is the same as | 07:39 |
| mnaser | alias = foo\nalias = bar | 07:39 |
| Adri2000 | ok let me try that :) | 07:39 |
| odyssey4me | I did think there were more examples, but apparently I'm wrong | 07:39 |
| odyssey4me | That really should be fixed - that module is super-useful, but useless without documentation to expose how to use itproperly | 07:40 |
| evrardjp | hwoarang: question | 07:40 |
| mnaser | i wonder maybe the ceph part has docs odyssey4me ? | 07:40 |
| evrardjp | is there a plan to support some parts of a deploy with distro and some with sources? | 07:40 |
| evrardjp | mnaser: odyssey4me we have docs in the reference that we can copy there | 07:41 |
| hwoarang | evrardjp: i have no such plans :/ | 07:41 |
| evrardjp | hwoarang: ok | 07:41 |
| evrardjp | we just know what the system is | 07:41 |
| hwoarang | it could be possible if these parts are on separate containers. if you are mixing packages on the same host it would be too hard | 07:41 |
| evrardjp | "current state" | 07:41 |
| evrardjp | ok | 07:41 |
| odyssey4me | evrardjp: perhaps andymccr could take it upon himself to add some docs, given he's been doing a ton of work on the module for ceph-ansible | 07:41 |
| evrardjp | he has done a great job :) | 07:42 |
| odyssey4me | ++ | 07:42 |
| Adri2000 | mnaser: evrardjp: foo, bar doesn't work but { foo, bar } works perfectly! | 07:44 |
| mnaser | there ya go | 07:44 |
| mnaser | :> | 07:44 |
| evrardjp | hwoarang: don't you think we should skip the repo_masters group creation (or make it empty) when distro_packages are used? | 07:44 |
| odyssey4me | Adri2000: could you do a bug report for lack of docs for what you wanted to do, then please share what worked in the report too | 07:45 |
| evrardjp | Adri2000: when did I say dict? :) | 07:45 |
| evrardjp | odyssey4me: Adri2000 ++ | 07:45 |
| evrardjp | hwoarang: I will submit a patch | 07:46 |
| hwoarang | ok | 07:48 |
| mnaser | wow, well we're hauling ass right now | 07:48 |
| mnaser | thanks to tailwinds doing over 1000km/h means i get home sooner woo | 07:48 |
| *** hamzy_ has quit IRC | 07:50 | |
| *** CharlesB has joined #openstack-ansible | 07:54 | |
| *** pmannidi has quit IRC | 07:56 | |
| *** electrofelix has joined #openstack-ansible | 07:56 | |
| Adri2000 | evrardjp: :p odyssey4me: yep will do | 07:57 |
| *** rmart04 has joined #openstack-ansible | 07:58 | |
| hwoarang | does it worth having the distro jobs just in the check pipeline? | 07:58 |
| hwoarang | so we can free up some 'gate' resources? | 07:59 |
| hwoarang | i think i had that in the past but evrardjp you said we need in both. sorry if i forget the details :/ there is only so much info i can hold in my head | 07:59 |
| odyssey4me | no need to save resources | 07:59 |
| hwoarang | ok | 07:59 |
| odyssey4me | and yeah, voting checks must also be in gate to ensure that we maintain consistency between patch submission and patch approval | 08:00 |
| odyssey4me | ie it makes sure that when it passes check (code submission) it also passes at the time of merge (the code underneath could have changed beteen submission and approval) | 08:00 |
| *** suggestable has joined #openstack-ansible | 08:01 | |
| hwoarang | when a patch is approved you still run the check pipeline right? | 08:01 |
| odyssey4me | nope, once w is added it goes straight into gate | 08:01 |
| odyssey4me | and because zuul rebases the patch on top of the current head of master, it's important that we match tests that matter in gate | 08:02 |
| hwoarang | hmm i thought that we ran check+gate on W+1. my bad | 08:03 |
| hwoarang | makes sense then | 08:03 |
| evrardjp | hwoarang: we needed both WHEN assuming some can be disabled. if it's an all or nothing situation that's different : ) | 08:03 |
| *** taseer2 is now known as Taseer | 08:03 | |
| evrardjp | I remember saying something about ppl have to be smart and not do dumb thing like install_source = distro and change for some only | 08:04 |
| evrardjp | but instead do install = source, and change some if they need to | 08:04 |
| evrardjp | I meant install_method=distro above. | 08:04 |
| *** chyka has joined #openstack-ansible | 08:04 | |
| evrardjp | I will patch that | 08:04 |
| *** chyka has quit IRC | 08:09 | |
| *** shardy has joined #openstack-ansible | 08:10 | |
| openstackgerrit | Merged openstack/openstack-ansible-tests master: Connect to localhost as remote host for cinder/nova https://review.openstack.org/584033 | 08:15 |
| odyssey4me | evrardjp hwoarang could I get some eyes on https://review.openstack.org/584042 and https://review.openstack.org/584038 please? | 08:26 |
| *** parasitid has quit IRC | 08:26 | |
| evrardjp | sure | 08:28 |
| *** KeithMnemonic has quit IRC | 08:37 | |
| *** KeithMnemonic has joined #openstack-ansible | 08:38 | |
| *** indistylo has joined #openstack-ansible | 08:42 | |
| *** parasitid has joined #openstack-ansible | 08:44 | |
| *** DanyC has joined #openstack-ansible | 08:55 | |
| openstackgerrit | Jean-Philippe Evrard proposed openstack/openstack-ansible master: Preparing group of host is not a "change" https://review.openstack.org/583879 | 09:07 |
| openstackgerrit | Jean-Philippe Evrard proposed openstack/openstack-ansible master: Automatic skip of the repo build for distro installs https://review.openstack.org/584285 | 09:07 |
| *** taseer1 has joined #openstack-ansible | 09:08 | |
| *** Taseer has quit IRC | 09:08 | |
| *** taseer2 has joined #openstack-ansible | 09:09 | |
| openstackgerrit | Jean-Philippe Evrard proposed openstack/openstack-ansible-os_swift master: Remove pypy job https://review.openstack.org/584287 | 09:11 |
| *** taseer3 has joined #openstack-ansible | 09:12 | |
| evrardjp | I'd love seeing this merged: https://review.openstack.org/#/c/581802/ | 09:12 |
| *** taseer1 has quit IRC | 09:12 | |
| *** cshen has quit IRC | 09:13 | |
| evrardjp | odyssey4me: that will be required for the test refactoring work ^ | 09:13 |
| odyssey4me | seems appropriate | 09:14 |
| *** cshen has joined #openstack-ansible | 09:15 | |
| *** taseer2 has quit IRC | 09:15 | |
| *** parasitid has quit IRC | 09:17 | |
| odyssey4me | hmm, do we have an example of how to set a standard job non-voting now? | 09:17 |
| odyssey4me | I'd like to temporarily disable the octavia functional test, get these fixes in to make it work again, then enable it again. | 09:18 |
| odyssey4me | ah: https://github.com/openstack/openstack-ansible-os_neutron/blob/master/zuul.d/project.yaml#L33 | 09:19 |
| Bhujay | wondering will these code work when i am running ceilometer container in a separate host which is not hosting rabbit containers ? https://github.com/openstack/openstack-ansible-os_ceilometer/blob/780d8d477cfcbe8c317dac4db68c2413b30dd0f1/templates/ceilometer.conf.j2#L44 | 09:21 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-os_octavia master: Switch compute to use different inventory_hostname https://review.openstack.org/584048 | 09:22 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-os_octavia master: Fixes compute error https://review.openstack.org/583276 | 09:22 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-os_octavia master: Temporarily set functional test to non-voting https://review.openstack.org/584295 | 09:22 |
| Bhujay | got this while running os_ceilometer , i had this error fatal: [infra7_ceilometer_central_container-d460b902]: FAILED! => {"failed": true, "msg": "'glance_rabbitmq_telemetry_host_group' is undefined"} | 09:23 |
| evrardjp | odyssey4me: just redefine the job with voting: false | 09:24 |
| Bhujay | and finally had to take help of this tedious workaround http://paste.openstack.org/show/726321/. Not sure if it need a fix or i am missing something | 09:24 |
| evrardjp | I mean re-list it in the project | 09:24 |
| evrardjp | yeah exactly :) | 09:24 |
| *** jbadiapa has quit IRC | 09:28 | |
| *** cshen has quit IRC | 09:28 | |
| Bhujay | evrardjp, odyssey4me , if you can take a look whenever time permits , not sure if its a bug ^ | 09:30 |
| *** hamzy_ has joined #openstack-ansible | 09:31 | |
| *** hachi has joined #openstack-ansible | 09:31 | |
| openstackgerrit | Merged openstack/openstack-ansible-os_ironic master: Remove localhost from neutron agents group https://review.openstack.org/584038 | 09:34 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-os_octavia master: Make certificate generation host configurable https://review.openstack.org/582641 | 09:35 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-os_octavia master: Execute service setup against a delegated host using Ansible built-in modules https://review.openstack.org/582324 | 09:35 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-os_octavia master: Reduce testing tech debt https://review.openstack.org/582668 | 09:35 |
| *** parasitid has joined #openstack-ansible | 09:35 | |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-os_octavia master: Revert "Temporarily set functional test to non-voting" https://review.openstack.org/584298 | 09:37 |
| openstackgerrit | Merged openstack/openstack-ansible-os_magnum master: Switch compute/storage to use different inventory_hostname https://review.openstack.org/584042 | 09:41 |
| openstackgerrit | Merged openstack/openstack-ansible-os_tempest master: Switch compute to use different inventory_hostname https://review.openstack.org/584037 | 09:42 |
| *** taildashef-work[ has quit IRC | 09:42 | |
| *** lbragstad[m] has quit IRC | 09:42 | |
| *** RossSchulman[m] has quit IRC | 09:42 | |
| *** linshuicheng[m] has quit IRC | 09:42 | |
| *** asdf1234asdf[m] has quit IRC | 09:42 | |
| *** jothibasu[m] has quit IRC | 09:42 | |
| *** csmart[m] has quit IRC | 09:42 | |
| *** keithmnemonic[m] has quit IRC | 09:42 | |
| *** evrardjp[m] has quit IRC | 09:43 | |
| *** jbadiapa has joined #openstack-ansible | 09:43 | |
| *** cshen has joined #openstack-ansible | 09:44 | |
| *** aetaric has joined #openstack-ansible | 09:47 | |
| *** asdf1234asdf[m] has joined #openstack-ansible | 09:48 | |
| *** prasen has joined #openstack-ansible | 09:48 | |
| openstackgerrit | Merged openstack/openstack-ansible-os_tempest master: Disable Volume Backup Tests https://review.openstack.org/583955 | 09:55 |
| *** taseer3 is now known as Taseer | 09:55 | |
| evrardjp | I would be happy if we could discuss the cleanup of the ansible_local multiple when conditions, by having ppl voting on what they think is best readable here: https://etherpad.openstack.org/p/osa-readability-test | 09:57 |
| evrardjp | odyssey4me: hwoarang jrosser d34dh0r53 cloudnull logan- jmccrory mnaser ^ | 09:58 |
| *** gillesMo has joined #openstack-ansible | 09:59 | |
| *** hachi has quit IRC | 10:02 | |
| openstackgerrit | Markos Chandras (hwoarang) proposed openstack/openstack-ansible-os_neutron master: tasks: providers: Rename 'ovs' configuration file https://review.openstack.org/583919 | 10:03 |
| hwoarang | oh that needs to free up some brain capacity | 10:04 |
| *** jbadiapa is now known as jbadiapa|away | 10:06 | |
| hwoarang | evrardjp: is there a typo in D? is not " ? | 10:08 |
| hwoarang | not empty? | 10:08 |
| evrardjp | yeah | 10:09 |
| evrardjp | that's the equivalent of default(False, True) | bool for example | 10:09 |
| *** cshen has quit IRC | 10:09 | |
| evrardjp | | default(... | 10:09 |
| odyssey4me | evrardjp: I wonder if that might be equivalent of None? But yeah, D looks good to me as-is. | 10:10 |
| odyssey4me | could perhaps do != there as well I guess | 10:11 |
| evrardjp | I will ofc try this | 10:11 |
| evrardjp | but it was to give an idea | 10:11 |
| evrardjp | because C is very compelling | 10:11 |
| odyssey4me | none is different | 10:12 |
| evrardjp | None is different indeed | 10:12 |
| odyssey4me | C is very vebose though, and easy to make a mistake with | 10:12 |
| evrardjp | not sure why this came up | 10:12 |
| evrardjp | oh you brought it :p | 10:13 |
| odyssey4me | there is also https://github.com/openstack/openstack-ansible-plugins/commit/80a1262252db951e40749d37f5348aa8907039c6 | 10:13 |
| evrardjp | odyssey4me: well, funny story, that's the reason I wanted to get rid of. | 10:13 |
| evrardjp | so ansible has in 2.7 a filter named subelements | 10:13 |
| odyssey4me | that filter's not used anywhere I don't think, so we can revert that patch if we're never going to use it | 10:14 |
| evrardjp | so that would partially cover this, and included upstream | 10:14 |
| evrardjp | so for example the test_dict: a: b: c: d: | 10:14 |
| evrardjp | so for example the test_dict: a: b: c: d* | 10:14 |
| evrardjp | you could do test_dict | subelements('c') | 10:15 |
| evrardjp | IIRC | 10:15 |
| *** keithmnemonic[m] has joined #openstack-ansible | 10:15 | |
| *** csmart[m] has joined #openstack-ansible | 10:15 | |
| evrardjp | else | 10:15 |
| *** evrardjp[m] has joined #openstack-ansible | 10:15 | |
| *** linshuicheng[m] has joined #openstack-ansible | 10:15 | |
| *** jothibasu[m] has joined #openstack-ansible | 10:15 | |
| evrardjp | json_query a.b.c | 10:15 |
| *** RossSchulman[m] has joined #openstack-ansible | 10:15 | |
| *** lbragstad[m] has joined #openstack-ansible | 10:15 | |
| hwoarang | jrosser: that was a good reading indeed (re bridge isolation) | 10:15 |
| evrardjp | here it's just the conditional that make it weird | 10:15 |
| *** taildashef-work[ has joined #openstack-ansible | 10:15 | |
| evrardjp | for the example given in the patch you said odyssey4me it would easily be rewritten in json_query as: {{ (ansible_local | json_query('openstack_ansible.swift.venv_tag')) == swift_venv_tag }} | 10:17 |
| evrardjp | so there is 0 difference there, and no reason to have get_nested | 10:18 |
| evrardjp | I was just asking for clarity of the thing when conditions are more complex. | 10:18 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-plugins master: Revert "Add a get_nested filter" https://review.openstack.org/584304 | 10:20 |
| evrardjp | I truly believe jinja is missing an "empty" test, which is the same as | default(True, True) | 10:21 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-plugins master: Revert "Add a get_nested filter" https://review.openstack.org/584304 | 10:22 |
| odyssey4me | evrardjp: agreed | 10:22 |
| *** hachi has joined #openstack-ansible | 10:23 | |
| evrardjp | odyssey4me: I wanted to do it! :) | 10:23 |
| evrardjp | please keep that topic name for all the cleanups :) | 10:24 |
| jrosser | evrardjp: is D missing a " ? | 10:27 |
| evrardjp | ? | 10:27 |
| hwoarang | jrosser: it's actual two ' | 10:27 |
| hwoarang | ;p i got confused by that too | 10:28 |
| hwoarang | *actually | 10:28 |
| evrardjp | let's rewrite this as "" | 10:28 |
| jrosser | ah right, guess "readbility test" is good name there :) | 10:29 |
| evrardjp | hahah | 10:29 |
| evrardjp | well we'll have to review those things | 10:29 |
| evrardjp | better be in sync before starting patching | 10:29 |
| evrardjp | I will remove my vote as I am split between B and D anyway. | 10:30 |
| evrardjp | oh the A is wrong | 10:31 |
| evrardjp | it's "bool" not defined | 10:31 |
| evrardjp | it doesn't matter I don't like it! :p | 10:32 |
| *** rmart04 has quit IRC | 10:32 | |
| *** shardy has quit IRC | 10:56 | |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-os_octavia master: Reduce testing tech debt https://review.openstack.org/582668 | 11:11 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-os_octavia master: Revert "Temporarily set functional test to non-voting" https://review.openstack.org/584298 | 11:11 |
| *** aruns__ has joined #openstack-ansible | 11:13 | |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-os_nova master: Execute service setup against a delegated host using Ansible built-in modules https://review.openstack.org/582315 | 11:14 |
| openstackgerrit | Markos Chandras (hwoarang) proposed openstack/openstack-ansible-os_horizon master: tasks: horizon_post_install: Ensure horizon dashboard directory is accessible https://review.openstack.org/584327 | 11:14 |
| openstackgerrit | Markos Chandras (hwoarang) proposed openstack/openstack-ansible-os_horizon master: tasks: horizon_post_install: Add missing become statements for horizon user https://review.openstack.org/584328 | 11:14 |
| openstackgerrit | Markos Chandras (hwoarang) proposed openstack/openstack-ansible-os_horizon master: tasks: horizon_post_install: Add 'force' parameter when creating symlink https://review.openstack.org/584329 | 11:14 |
| openstackgerrit | Markos Chandras (hwoarang) proposed openstack/openstack-ansible-os_horizon master: Add support for using distribution packages for OpenStack services https://review.openstack.org/584330 | 11:14 |
| *** indistylo has quit IRC | 11:16 | |
| odyssey4me | evrardjp: does https://review.openstack.org/#/c/584049/3 look right to you, or just I screw up a change somewhere? | 11:17 |
| *** indistylo has joined #openstack-ansible | 11:17 | |
| evrardjp | checking | 11:17 |
| *** aruns__ has quit IRC | 11:18 | |
| *** udesale__ has quit IRC | 11:19 | |
| evrardjp | odyssey4me: NoValidConnectionsError: [Errno None] Unable to connect to port 22 on 10.1.3.3 | 11:19 |
| evrardjp | fun with networking. | 11:19 |
| odyssey4me | yeah, do you think my change broke it or do you think that's some sort of other breakage somewhere? | 11:20 |
| * odyssey4me doesn't understand all that crap | 11:20 | |
| evrardjp | I need to see the whole picture, which I don't have with just a patch... IMO we are now using SSH everywhere | 11:21 |
| evrardjp | which is error prone | 11:22 |
| evrardjp | but this can't be true :p | 11:22 |
| odyssey4me | yeah, but there is no other way around it - I tried pretty much every other way I could think of | 11:22 |
| odyssey4me | other roles are fine, even those that use nova | 11:23 |
| odyssey4me | although I expect they use different tests | 11:23 |
| evrardjp | there is only one way to find out | 11:23 |
| evrardjp | (opening code) | 11:23 |
| evrardjp | all the plays are from tests repo | 11:24 |
| evrardjp | good | 11:24 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-os_nova master: Execute service setup against a delegated host using Ansible built-in modules https://review.openstack.org/582315 | 11:24 |
| evrardjp | and the inventory is simple | 11:24 |
| evrardjp | oh now I understand | 11:24 |
| evrardjp | we have physical_host and container_name into group_vars. | 11:25 |
| odyssey4me | I took 'openstack1' out of the 'neutron_linuxbridge_agent' group because I didn't think that actually belonged there, but perhaps it does? | 11:25 |
| odyssey4me | oh, you think that physical_hosts should now be 'compute1' ? | 11:26 |
| evrardjp | no I am not sure it matters, they will still do it right | 11:27 |
| evrardjp | oh wait | 11:27 |
| evrardjp | yeah it doesn't matter | 11:27 |
| odyssey4me | might matter for these vars? https://review.openstack.org/#/c/584049/3/tests/host_vars/compute1.yml | 11:28 |
| *** aruns__ has joined #openstack-ansible | 11:28 | |
| evrardjp | you might want to add ansible_connection=local in the inventory | 11:29 |
| evrardjp | for compute1 | 11:29 |
| odyssey4me | nope, we do not want connection=local for it | 11:29 |
| evrardjp | why? | 11:30 |
| odyssey4me | that's why this change is happening in the first place | 11:30 |
| evrardjp | else it would ssh | 11:30 |
| odyssey4me | if you use local connection, then it needs become - read the commit msg | 11:30 |
| odyssey4me | we *want* it to ssh, because that way we're in as root | 11:30 |
| evrardjp | mmm yeah. | 11:31 |
| evrardjp | so it's on purpose | 11:31 |
| evrardjp | I see | 11:31 |
| *** indistylo has quit IRC | 11:31 | |
| jamesdenton | mornin' | 11:31 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-os_nova master: Switch compute to use different inventory_hostname https://review.openstack.org/584049 | 11:33 |
| evrardjp | I think the best thing you could do is run it on a machine odyssey4me | 11:33 |
| odyssey4me | trying to put that back to see | 11:33 |
| evrardjp | and find out what's wrong | 11:33 |
| evrardjp | jamesdenton: morning! | 11:33 |
| jamesdenton | hi! | 11:33 |
| jamesdenton | someone said networking. my ears were burning | 11:33 |
| evrardjp | the issue also appears only in tempest | 11:33 |
| evrardjp | hahaha | 11:34 |
| odyssey4me | haha | 11:34 |
| evrardjp | jamesdenton: as long as it's not bleeding | 11:34 |
| odyssey4me | yeah, I suspect that the provider network isn't working... but I don't have enough understanding of how that works to troubleshoot it even if I did build an instance | 11:34 |
| jamesdenton | there are days... | 11:34 |
| evrardjp | https://www.youtube.com/watch?v=eNr0WXQ3Ho4 | 11:35 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-os_nova master: Move MQ vhost/user creation into role https://review.openstack.org/583387 | 11:35 |
| evrardjp | odyssey4me: try without removing the vars of localhost | 11:35 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-os_nova master: Execute service setup against a delegated host using Ansible built-in modules https://review.openstack.org/582315 | 11:36 |
| evrardjp | if it's localhost used as target for the building the networks part you're in a bad place | 11:36 |
| odyssey4me | evrardjp: I thought of that, then saw this: https://github.com/openstack/openstack-ansible-os_nova/blob/master/tests/os_nova-overrides.yml#L19-L22 | 11:36 |
| evrardjp | that's neutron though? | 11:36 |
| evrardjp | haha great. | 11:37 |
| evrardjp | more overrides. | 11:37 |
| odyssey4me | we *really* want to make sure those provider network settings are found | 11:37 |
| odyssey4me | https://github.com/openstack/openstack-ansible-os_nova/blob/master/tests/host_vars/localhost.yml#L22-L29 stays in localhost's vars because the test play that builds the bridges is targeted at localhost | 11:37 |
| evrardjp | odyssey4me: that was my point | 11:38 |
| odyssey4me | so those vars stay in my patch... so I don't understand what you're suggesting | 11:38 |
| *** lvdombrkr has joined #openstack-ansible | 11:39 | |
| *** cshen has joined #openstack-ansible | 11:39 | |
| evrardjp | nothing yet | 11:39 |
| *** lvdombrkr89 has quit IRC | 11:40 | |
| *** hachi has quit IRC | 11:41 | |
| *** lvdombrkr89 has joined #openstack-ansible | 11:41 | |
| odyssey4me | hwoarang & evrardjp could we go through some review quickly to unblock octavia? | 11:41 |
| evrardjp | yeah -- I don't see the nova issue just yet | 11:42 |
| odyssey4me | https://review.openstack.org/584295 then https://review.openstack.org/584048 then https://review.openstack.org/583276 | 11:42 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-os_octavia master: Revert "Temporarily set functional test to non-voting" https://review.openstack.org/584298 | 11:42 |
| odyssey4me | then that one ^ | 11:43 |
| evrardjp | odyssey4me: nothing in your chain works though | 11:43 |
| odyssey4me | https://review.openstack.org/#/c/583276/2 passed the ssl test, so I think it's done | 11:43 |
| odyssey4me | the functional test just timed out | 11:43 |
| *** lvdombrkr has quit IRC | 11:44 | |
| evrardjp | yeah -- I hope it's not on a network issue in tempest :p | 11:45 |
| Dmitriy | I have another question regarding masakari role. It seems, that in 17.0.6 venv for masakari should be built on the repo server. However, build fails with the following error: | 11:45 |
| evrardjp | at some point if all of this doesn't work, should we start reverting this whole become: user: story? | 11:45 |
| Dmitriy | Could not find a version that satisfies the requirement microversion-parse>=0.2.1 (from masakari) (from versions: 0.1.4) | 11:45 |
| Dmitriy | No matching distribution found for microversion-parse>=0.2.1 (from masakari) | 11:45 |
| evrardjp | I am leaving a few more days, but if tempest doesn't work, we'll never be able to reach m3. | 11:46 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-os_octavia master: Move MQ vhost/user creation into role https://review.openstack.org/583388 | 11:46 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-os_octavia master: Make certificate generation host configurable https://review.openstack.org/582641 | 11:47 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-os_octavia master: Execute service setup against a delegated host using Ansible built-in modules https://review.openstack.org/582324 | 11:47 |
| evrardjp | odyssey4me: we are asking you too much there, and it must be super frustrating , I am sorry. | 11:48 |
| odyssey4me | evrardjp: if we revert that, all the work delegating to the deploy host will have to be reverted - so no, I'd rather than run all tests as root and be done with it | 11:48 |
| odyssey4me | getting this all to work is pretty key to that other spec to put all OSA config into user-space (non-root) | 11:48 |
| evrardjp | ok | 11:49 |
| evrardjp | I don't say it's not valuable, don't get me wrong | 11:49 |
| evrardjp | I just say it's maybe a bad timing | 11:49 |
| odyssey4me | this approach has worked for several roles already, so I think this is fine - it's just some weird niggle to figure out | 11:49 |
| Dmitriy | And /opt/venv-build-script.sh fails for masakari as well. http://paste.openstack.org/show/726335/ | 11:50 |
| odyssey4me | wuld like some review for https://review.openstack.org/583366 now that cinder's unblocked | 11:51 |
| odyssey4me | Dmitriy: the great likelihood is that you're trying to pin to a SHA that's out of sync with the global requirements | 11:51 |
| odyssey4me | Dmitriy: pick a SHA for the same date as all the other SHA's | 11:51 |
| *** aruns__ has quit IRC | 11:52 | |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-os_magnum master: Update to use oslo.messaging service for RPC and Notify https://review.openstack.org/579645 | 11:53 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-os_magnum master: Move MQ vhost/user creation into role https://review.openstack.org/583379 | 11:53 |
| *** indistylo has joined #openstack-ansible | 11:54 | |
| Dmitriy | odyssey4me I'm trying to upgrade from 17.0.5 to 17.06. So I've sewitched to it, runned bootstrap-ansible.sh and now I'm trying to run setup-infrastructure.yml. | 11:54 |
| odyssey4me | Dmitriy: given that masakari is not included in the integrated repo, you'll have to tell me more about what settings you've implemented to use it | 11:56 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-os_magnum master: Execute service setup against a delegated host using Ansible built-in modules https://review.openstack.org/582296 | 11:56 |
| *** jbadiapa|away is now known as jbadiapa | 11:58 | |
| openstackgerrit | Guilherme Steinmuller Pimentel proposed openstack/openstack-ansible-os_rally stable/queens: Change rally pip package https://review.openstack.org/584089 | 11:58 |
| Dmitriy | odyssey4me hm, I was working on masakari role, and used it in our production. So I've placed /etc/openstack_deploy/env.d/masakari.yml with description of the service http://paste.openstack.org/show/726336/ | 11:59 |
| hwoarang | evrardjp: you misunderstood me on https://review.openstack.org/#/c/584285/ | 11:59 |
| Dmitriy | Also I've placed installation step in openstack_user_config.yml | 12:00 |
| odyssey4me | Dmitriy: I mean specifically with regards to the git/version related information in your overrides/repo_packages | 12:00 |
| guilhermesp | ^^ ignore the patch above, I was just updating a commit message, but as odyssey4me said, I proposed on master and I passed | 12:00 |
| guilhermesp | mornings :) | 12:00 |
| odyssey4me | Dmitriy: have you set any values anywhere for https://github.com/openstack/openstack-ansible-os_masakari/blob/master/defaults/main.yml#L29-L30 ? | 12:01 |
| Dmitriy | odyssey4me nope | 12:02 |
| odyssey4me | Dmitriy: ok, so that means you're trying to build rocky masakari for a queens deployment | 12:03 |
| Dmitriy | they are excatly like in git | 12:03 |
| odyssey4me | the requirements are not compatible | 12:03 |
| Dmitriy | odyssey4me it seems, that I get the point. but masakari doesn't have any tag/branch except master | 12:03 |
| odyssey4me | so, for it to work right, you'll need to set those two vars in your /etc/openstack_deploy/user_variables.yml file, with the value for masakari_git_install_branch being a SHA that's on the same date as those that match the other openstack services for the release of OSA you're using: https://github.com/openstack/openstack-ansible/blob/17.0.6/playbooks/defaults/repo_packages/openstack_services.yml | 12:04 |
| Dmitriy | no, my fault | 12:04 |
| *** kstev has joined #openstack-ansible | 12:04 | |
| odyssey4me | Dmitriy: I'm talking about the openstack/masakari version here, not the role version | 12:04 |
| odyssey4me | so, for this example we find a SHA for masakari's stable/queens which matches the date 09.06.2018 | 12:05 |
| Dmitriy | odyssey4me: Yep, I've already realised, that I'm not right. Thanks you for pointing me | 12:05 |
| odyssey4me | no worries, happy to help | 12:05 |
| *** prasen has quit IRC | 12:07 | |
| evrardjp | hwoarang: oh good point. I still answered :p | 12:13 |
| evrardjp | ppl shouldn't then | 12:14 |
| *** kstev has quit IRC | 12:16 | |
| hwoarang | i am not sure i understood the comment about manual deployment to be honest :/ I didn't mean to block the patch just thought to point that out | 12:19 |
| openstackgerrit | Merged openstack/openstack-ansible-os_octavia master: Temporarily set functional test to non-voting https://review.openstack.org/584295 | 12:35 |
| openstackgerrit | Merged openstack/openstack-ansible-os_octavia master: Switch compute to use different inventory_hostname https://review.openstack.org/584048 | 12:35 |
| openstackgerrit | Merged openstack/openstack-ansible-os_octavia master: Fixes compute error https://review.openstack.org/583276 | 12:35 |
| evrardjp | hwoarang: if someone deploys by doing os-keystone-install.yml os-glance-install.yml ... | 12:35 |
| evrardjp | or in this case haproxy-install.yml repo-install.yml | 12:36 |
| evrardjp | so that conditional wouldn't appear | 12:36 |
| evrardjp | however | 12:36 |
| evrardjp | I abandonned the thing, because I thought that if ppl start plays manually, there is no reason they aren't smart enough to not run the repo-install | 12:36 |
| *** cshen has quit IRC | 12:39 | |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-repo_build stable/queens: tasks: repo_build_wheels: Download get-pip.py if it does not exist https://review.openstack.org/584355 | 12:45 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-repo_build stable/queens: tasks: repo_build_wheels: Download get-pip.py if it does not exist https://review.openstack.org/584355 | 12:50 |
| *** cronix has joined #openstack-ansible | 12:50 | |
| odyssey4me | evrardjp with https://review.openstack.org/#/c/583925/ merged, https://review.openstack.org/584355 is now necessary | 12:50 |
| hwoarang | evrardjp: ah ok that would make sense... | 12:51 |
| admin0 | hi all .. is it possible to just do horizon and keystone install.. and then add 3 existing OSA installs as regions to this new one :) | 12:52 |
| admin0 | how do I tackle/handle regions | 12:53 |
| evrardjp | admin0: yes, you might want to speak to other ppl with experience but yeah | 12:54 |
| evrardjp | odyssey4me: HOHOHOHOHO https://review.openstack.org/#/c/574006/ | 12:55 |
| evrardjp | green! | 12:55 |
| evrardjp | after a month! | 12:55 |
| evrardjp | hwoarang: could you merge this one in: https://review.openstack.org/#/c/574006/17 | 12:56 |
| evrardjp | after that I will stop being grumpy | 12:56 |
| evrardjp | (I didn't say for how long) | 12:56 |
| evrardjp | next step is bumping pip to 10! | 12:56 |
| evrardjp | wooot | 12:56 |
| evrardjp | and re-bumping the shas | 12:56 |
| hwoarang | ok | 12:57 |
| evrardjp | but at least we have something to release for M3. That was far too old | 12:57 |
| odyssey4me | just in time for the next bump ;) | 12:58 |
| odyssey4me | good work on that evrardjp - that got complicated | 12:58 |
| *** armaan has joined #openstack-ansible | 13:00 | |
| openstackgerrit | caoyuan proposed openstack/openstack-ansible-os_keystone master: Remove the unnecessary space https://review.openstack.org/582791 | 13:04 |
| mgariepy | good morning everyone | 13:04 |
| evrardjp | odyssey4me: well I haven't fixed the things | 13:04 |
| evrardjp | I need to dig into cinder + swift + keystone | 13:04 |
| evrardjp | but it's something I'd like to do when I have more time | 13:04 |
| evrardjp | fix it in devstack too | 13:05 |
| openstackgerrit | Merged openstack/openstack-ansible-os_cinder master: Switch storage to use different inventory_hostname https://review.openstack.org/584052 | 13:12 |
| openstackgerrit | Merged openstack/openstack-ansible-os_cinder master: Change deprecated option iscsi_helper to target_helper https://review.openstack.org/584151 | 13:12 |
| evrardjp | mgariepy: woot: Readability test for you: https://etherpad.openstack.org/p/osa-readability-test | 13:12 |
| *** cshen has joined #openstack-ansible | 13:13 | |
| mgariepy | evrardjp, isn't | not supposed to be there anymore ? | 13:15 |
| mgariepy | | == is ? | 13:15 |
| openstackgerrit | jacky06 proposed openstack/openstack-ansible-os_keystone master: Remove the unnecessary space https://review.openstack.org/582791 | 13:15 |
| mgariepy | or i misunderstood something else. | 13:15 |
| odyssey4me | cores, quick review please - very simple patch: https://review.openstack.org/583672 | 13:17 |
| *** parasitid has quit IRC | 13:18 | |
| mgariepy | done | 13:18 |
| *** parasitid has joined #openstack-ansible | 13:19 | |
| *** CharlesB has quit IRC | 13:20 | |
| odyssey4me | thx mgariepy got another one here if you have a moment: https://review.openstack.org/584304 | 13:24 |
| *** rmart04 has joined #openstack-ansible | 13:24 | |
| evrardjp | mgariepy: I have serious trouble with the "is bool". For me, False or True would return True to is bool :) | 13:25 |
| evrardjp | mgariepy: | == is for tests | 13:26 |
| evrardjp | | should be kept for filters. | 13:26 |
| openstackgerrit | Markos Chandras (hwoarang) proposed openstack/openstack-ansible-os_horizon master: Add support for using distribution packages for OpenStack services https://review.openstack.org/584330 | 13:26 |
| Adri2000 | got another override related question: trying to override enable_router in OPENSTACK_NEUTRON_NETWORK in horizon's local_settings.py: the horizon_config_overrides var can only override HORIZON_CONFIG if I understand correctly :/ any solution apart from adding a new variable to the local_settings.py template? | 13:26 |
| openstackgerrit | Merged openstack/openstack-ansible-pip_install stable/queens: Update get-pip to version 3.3 https://review.openstack.org/583234 | 13:27 |
| mgariepy | ha ok | 13:27 |
| evrardjp | Adri2000: yeah currently local_settings.py is still a legacy thing | 13:27 |
| evrardjp | Adri2000: I'd welcome the contributor that changes it to the new format, if the new format is ready | 13:28 |
| evrardjp | just saying. | 13:28 |
| evrardjp | :D | 13:28 |
| odyssey4me | evrardjp: surely 'is bool' is a type verification of the var, whereas '| bool' returns an evaluation result based on the content of the var? | 13:28 |
| evrardjp | odyssey4me: that's the thing | 13:28 |
| evrardjp | it is not a type verification | 13:28 |
| Adri2000 | evrardjp: you mean there is a new config file format coming in horizon? | 13:28 |
| evrardjp | | bool is a test | 13:28 |
| evrardjp | so it will be written "is bool" in the future. | 13:28 |
| *** armaan has quit IRC | 13:28 | |
| ebbex_ | "lxc_container_create : Generate machine-id" fails on stable/queens, "mesg: ttyname failed: Inappropriate ioctl for device\n/bin/sh: 1: /usr/bin/python: not found\n" | 13:28 |
| evrardjp | I just damn hate it. | 13:28 |
| evrardjp | I want to rename it "is truth" | 13:29 |
| evrardjp | is true | 13:29 |
| evrardjp | or whatever | 13:29 |
| odyssey4me | well, '| bool' is actually evaluating the var content as far as I know | 13:29 |
| evrardjp | is true-ish | 13:29 |
| evrardjp | yeah | 13:29 |
| openstackgerrit | Merged openstack/openstack-ansible-pip_install stable/pike: Update get-pip to version 3.3 https://review.openstack.org/583235 | 13:29 |
| *** hachi has joined #openstack-ansible | 13:29 | |
| evrardjp | so it's an ansible filter that checks if the content is true-ish | 13:29 |
| odyssey4me | so if you have {{ 'foo' | bool }} it will return False, because 'foo' != True | 13:29 |
| evrardjp | so y/yes/True/true matches | 13:29 |
| evrardjp | that's correct odyssey4me | 13:30 |
| evrardjp | odyssey4me: in the future you will have to write it "{{ 'foo' is bool }}" | 13:30 |
| mgariepy | False | bool ? | 13:30 |
| mgariepy | would it test if the thing is a bool ? .. | 13:31 |
| evrardjp | mgariepy: that will be written "{{ False is bool }}" and will return false. | 13:31 |
| odyssey4me | well, meh | 13:31 |
| evrardjp | because False isn't true-ish | 13:31 |
| evrardjp | it's still a boolean | 13:31 |
| evrardjp | but hey | 13:31 |
| evrardjp | I didn't write that. | 13:31 |
| mgariepy | changing it now wouldn't be a great idea either i guess haha | 13:32 |
| evrardjp | also remember that testing on strings could return True :) | 13:33 |
| evrardjp | fun with ansible! | 13:33 |
| evrardjp | so {% if 'False' %} is True-ish | 13:33 |
| evrardjp | but {% if False %} isn't. | 13:33 |
| mgariepy | happy friday ! | 13:34 |
| evrardjp | I meant is True, not True-ish in the meaning of earlier. | 13:34 |
| *** jwitko has joined #openstack-ansible | 13:34 | |
| evrardjp | hahaha | 13:34 |
| evrardjp | but that all makes sense to me :) | 13:34 |
| * odyssey4me pours himself a drink | 13:34 | |
| evrardjp | I just want to have our own filter that is doing the same as | bool but renames things as "trueish" | 13:35 |
| evrardjp | or is somethingliketrue | 13:35 |
| * odyssey4me noticed evrardjp muttering to himself | 13:35 | |
| odyssey4me | Noisy up there, eh? | 13:36 |
| evrardjp | hahahah | 13:36 |
| Adri2000 | evrardjp: would you +1 adding an os_horizon horizon_enable_router variable for the current local_settings.py? :) | 13:38 |
| *** chyka has joined #openstack-ansible | 13:38 | |
| evrardjp | I'd be okay for you Adri2000 | 13:39 |
| guilhermesp | ppl, is this ( https://bit.ly/2LpiPfY ) "poor" in terms of testing or do you think it needs more scenarios? | 13:39 |
| evrardjp | but I'd be even more okay with using yaml/ini | 13:39 |
| Adri2000 | evrardjp: but that'd be require changes in horizon directly right? | 13:40 |
| evrardjp | Adri2000: alternatively you can bring the .py folder | 13:40 |
| evrardjp | Adri2000: I think those merged, right? | 13:40 |
| Adri2000 | no idea | 13:40 |
| evrardjp | Adri2000: well I support the local_settings.d/ should be there | 13:40 |
| evrardjp | let me check for the rest | 13:41 |
| *** goldenfri has joined #openstack-ansible | 13:41 | |
| *** EmilienM is now known as EvilienM | 13:41 | |
| openstackgerrit | Adrien Cunin proposed openstack/openstack-ansible-os_horizon master: New horizon_enable_router variable https://review.openstack.org/584375 | 13:42 |
| Adri2000 | at least, here is the local patch I had to do to move forward ^ :) | 13:43 |
| evrardjp | Adri2000: apparently I was wrong | 13:43 |
| Adri2000 | ah | 13:43 |
| evrardjp | there is no .ini or .yaml format support yet | 13:43 |
| evrardjp | but local_settings.d/ can be brought up though | 13:43 |
| *** chyka has quit IRC | 13:44 | |
| evrardjp | I think it's time to refactor horizon's role code, it starts to be messy | 13:44 |
| Dmitriy | odyssey4me: and how masakari has appeared in building venv? Is venv is been build for every role, that is present in /etc/ansible/roles? On what basis it is added by repo_build role? As https://github.com/openstack/openstack-ansible-repo_build/blob/8cbd722ac4de268cbf6c283f9825958071b6a6bb/tasks/repo_clone_git.yml#L34-L52 still trying to clone master branch, while I've specified branch for masakari in user_variables.yml | 13:45 |
| Dmitriy | evrardjp: do you think if https://bugs.launchpad.net/openstack-ansible/+bug/1778098 needs to be fixed? I may send a patch if this approach looks ok | 13:49 |
| openstack | Launchpad bug 1778098 in openstack-ansible "os_horizon role fails if horizon_custom_themes is specified" [Low,Confirmed] - Assigned to Amy Marrich (amy-marrich) | 13:49 |
| evrardjp | Dmitriy: all bugs need fixing | 13:49 |
| evrardjp | I can discuss that in a minute | 13:50 |
| odyssey4me | Dmitriy: you have to re-specify the repo too otherwise the plugin won't lookup your branch spec | 13:51 |
| *** coreywright has quit IRC | 13:51 | |
| odyssey4me | even if the value is the same, doesn't matter | 13:51 |
| Dmitriy | evrardjp ok. I just want to get if we are going just to patch by creating empty dir (just to make role not to fail) or we would like to allow people to deploy their themes by playbook. | 13:52 |
| guilhermesp | Dmitriy was this bug oppened by you, isn't it? | 13:53 |
| Dmitriy | guilhermesp: Yep, it was. But it was opened before I've discovered gerrit and this IRC) | 13:55 |
| guilhermesp | hahahha | 13:55 |
| guilhermesp | I'm still trying to understand why do we need "Setting horizon permissions" task | 13:55 |
| odyssey4me | Dmitriy: Do you have masakari_git_install_branch set in user_variables.yml? And do you have masakari_git_repo set in the same file? | 13:56 |
| Dmitriy | guilhermesp: actually we don't, but isn't it more beatifull, when all files of horizon are owned by horizon user?) | 13:56 |
| Dmitriy | odyssey4me: masakari_git_install_branch is set, but masakari_git_repo should be taken from role defaults | 13:58 |
| evrardjp | I re-read the bug | 13:58 |
| evrardjp | let me find the triage conversation | 13:58 |
| odyssey4me | Dmitriy: if you don't set masakari_git_repo then the lookup plugin will ignore your other setting | 13:59 |
| odyssey4me | even if the value is the same, that var is the key that unlocks the use of the overrides | 13:59 |
| evrardjp | Dmitriy: the bug is a documentation issue | 14:01 |
| evrardjp | http://eavesdrop.openstack.org/meetings/openstack_ansible_meeting/2018/openstack_ansible_meeting.2018-06-26-16.00.log.html#l-244 | 14:01 |
| evrardjp | spotz: said she would do it | 14:01 |
| evrardjp | You can take over from it it's been a month | 14:01 |
| evrardjp | just tell spotz | 14:01 |
| *** tux_ has joined #openstack-ansible | 14:02 | |
| hwoarang | hmm lxc_aio ubuntu has started to timeout quite frequently lately :/ | 14:03 |
| guilhermesp | Dmitriy just to add into your comment. I was just thinking of add "unecessary" tasks, so I think the focus now is to refactor and make things simples | 14:03 |
| Dmitriy | odyssey4me: oh... ok, I will try. | 14:03 |
| *** coreywright has joined #openstack-ansible | 14:04 | |
| *** tux__ has joined #openstack-ansible | 14:04 | |
| evrardjp | hwoarang: I've noticed it too | 14:05 |
| evrardjp | I think we have a race condition somewhere | 14:06 |
| Dmitriy | guilhermesp: I was just trying to make it more correct in some terms, but didn't take attention on simplicity. I'm pretty new to openstack, so it was a consequence) | 14:06 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-repo_server master: Remove the upstream pypi reverse proxy https://review.openstack.org/584385 | 14:07 |
| *** tux_ has quit IRC | 14:07 | |
| guilhermesp | Dmitriy no problem! :D keep making suggestions, this is the spirit | 14:09 |
| openstackgerrit | Merged openstack/openstack-ansible-os_ironic stable/queens: Remove duplicate key https://review.openstack.org/583672 | 14:09 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-repo_server master: Remove the upstream pypi reverse proxy https://review.openstack.org/584385 | 14:11 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-repo_server stable/queens: Remove the upstream pypi reverse proxy https://review.openstack.org/584387 | 14:11 |
| Dmitriy | evrardjp: so, the workflow of openstack installation should be the following: running setup-hosts, then to place theme into themes directory, and run setup-openstack? | 14:12 |
| *** kstev has joined #openstack-ansible | 14:12 | |
| openstackgerrit | Adrien Cunin proposed openstack/openstack-ansible-os_nova master: Put passthrough_whitelist in the [pci] section https://review.openstack.org/584390 | 14:14 |
| cloudnull | mornings | 14:14 |
| *** cshen has quit IRC | 14:15 | |
| evrardjp | Dmitriy: you can probably do it while doing your configuration | 14:15 |
| evrardjp | then run all the plays | 14:15 |
| evrardjp | morning cloudnull | 14:15 |
| odyssey4me | cloudnull logan- more of those get-pip patches to review: https://review.openstack.org/#/q/topic:bug/1779534-stable/queens+status:open / https://review.openstack.org/#/q/topic:bug/1779534-stable/pike+status:open / https://review.openstack.org/#/q/topic:bug/1779534-stable/ocata+status:open | 14:18 |
| Dmitriy | evrardjp: but on start I don't have a container for horizon, so there isn't a place to put a theme. So it should be done in meanwhile. But ok, in case of fixing docs, this shouldn't be a problem | 14:18 |
| *** FrankZhang has quit IRC | 14:19 | |
| admin0 | \o | 14:20 |
| guilhermesp | Dmitriy I think the way it is now, you do your customizations after deploying horizon. | 14:20 |
| guilhermesp | is there any reason to customize your themes in a fresh install? | 14:20 |
| guilhermesp | I mean, defining your custom themes at the first time you're deploying osa? | 14:21 |
| jrosser | odyssey4me: would it be worth us testing https://review.openstack.org/584387 ? | 14:21 |
| guilhermesp | you can deploy the default first, so you're going to have your libs dir and then you can do your customizations | 14:21 |
| Dmitriy | guilhermesp: do you mean not to specify horizon_custom_themes on fresh install, and place it on the second run? As in case of providing it at the first run, os_horizon fails, because of collectstatic failure | 14:22 |
| guilhermesp | yep Dmitriy | 14:23 |
| Dmitriy | guilhermesp: yep, this can be a workaround (which I used), but it's not really obvious and looks like some bug... | 14:23 |
| odyssey4me | jrosser: busy putting together the test set now | 14:23 |
| odyssey4me | needs an integrated repo patch | 14:23 |
| odyssey4me | almost done | 14:23 |
| guilhermesp | Dmitriy: uhum, spotz is going to deal with it :) | 14:24 |
| jrosser | excellent, whats more helpful for you - http_proxy only or http_proxy with local pip mirror, we can try either | 14:24 |
| evrardjp | Dmitriy: no the plan was to have the files to deploy on the deploy node | 14:25 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible master: Use upstream pypi before the repo is built https://review.openstack.org/584393 | 14:25 |
| evrardjp | And you point to that folder | 14:25 |
| evrardjp | in your config | 14:25 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-repo_server master: Remove the upstream pypi reverse proxy https://review.openstack.org/584385 | 14:25 |
| evrardjp | odyssey4me: can we remove apt mirror can we can we :) | 14:25 |
| odyssey4me | evrardjp: the facility to disable it is already there, so I don't think we should remove it | 14:26 |
| tux__ | In production what is recommended bluestore with LVM or raw OSD disk? | 14:29 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible stable/queens: Use upstream pypi before the repo is built https://review.openstack.org/584394 | 14:29 |
| Dmitriy | evrardjp: yep, that sounds great. But I have some note on using synchronize (which I offered previously) for files delivery - as it's actually rsync, it should have direct access to the containers, which not always the case (for example, we don't have access from deploy host to containers mgmt network). But unarchive works really good, and we're creating archive with git, which doesn't make archive creation a problem. | 14:30 |
| evrardjp | Dmitriy: We should assume access to the mgmt network as this is what's used for ssh | 14:33 |
| evrardjp | that's the purpose of is_container_address | 14:33 |
| evrardjp | IIRC | 14:34 |
| jrosser | thats not strictly true is it? | 14:34 |
| evrardjp | I think it was :) | 14:34 |
| evrardjp | ppl can avoid it | 14:34 |
| jrosser | let me rephrase, it's not mandatory the the mgmt network is the one the deploy host ssh's to the targets over | 14:34 |
| evrardjp | jrosser: oh yeah indeed, but is_container_address was supposed to be the one | 14:35 |
| evrardjp | by default it's br-mgmt | 14:35 |
| Dmitriy | evrardjp: We have access to the mgmt address of management and compute nodes, but not inside of containers, and (it sounds strange) it works. I don't fully understand why, but it did. | 14:35 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible stable/queens: Update get-pip to version 3.3 https://review.openstack.org/584095 | 14:35 |
| evrardjp | ofc ppl can do whatever they want, because the IP used for the node is what matters | 14:35 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible stable/queens: Use upstream pypi before the repo is built https://review.openstack.org/584394 | 14:35 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible stable/queens: Ensure that repo checks use the right values https://review.openstack.org/564422 | 14:35 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible stable/queens: Ensure that the repo servers also use the pypi cache https://review.openstack.org/568502 | 14:35 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible stable/queens: Remove default pip_links value https://review.openstack.org/568637 | 14:35 |
| odyssey4me | jrosser: that's the full stack now - let's see how it goes | 14:36 |
| odyssey4me | if you could test with proxy that'd be great | 14:36 |
| evrardjp | Dmitriy: probably because of our connection plugin, connecting to the physical node of your container | 14:36 |
| Dmitriy | evrardjp the only problem we've ever faced is syncronize... | 14:36 |
| evrardjp | ofc | 14:36 |
| jrosser | odyssey4me: we cal also choose pip mirror or not pip mirror? | 14:37 |
| evrardjp | This is easily solved by putting one of your infra nodes as deploy node though. | 14:37 |
| Dmitriy | evrardjp yep, probably that's the point... | 14:37 |
| jrosser | *can | 14:37 |
| odyssey4me | jrosser: now the repo build pip mirror var still takes effect, but once the repo is built it switches to using pypiserver on the repo container | 14:38 |
| odyssey4me | for the moment, that's the functionality I'm targeting to keep like-for-like capability with pike for queens | 14:38 |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible-ops master: Add functional tests to elk_metrics_6x https://review.openstack.org/583794 | 14:39 |
| odyssey4me | for master if we can stabilise this I can turn back to trying to put the python builds into the roles - once that's done, we can make the repo server optional | 14:39 |
| jrosser | ok, we'll we'll start with just http proxy and see where it gets to, keep it simple | 14:39 |
| *** Bhujay has quit IRC | 14:42 | |
| odyssey4me | evrardjp: tada - nova fixed: https://review.openstack.org/584049 | 14:43 |
| cloudnull | yay! | 14:43 |
| odyssey4me | fokks, can we get another review for https://review.openstack.org/584298 to get octavia functional tests voting again | 14:43 |
| odyssey4me | erm, folks | 14:43 |
| odyssey4me | hahaha | 14:43 |
| odyssey4me | reviews for https://review.openstack.org/583387 on nova too - nice to see that working now :) | 14:44 |
| odyssey4me | xgerman_ cloudnullcould you revisit https://review.openstack.org/#/c/582641/ now that it's passing again - thanks for the help fixing octavia functional tests xgerman_ ! | 14:45 |
| odyssey4me | and https://review.openstack.org/583388 while you're there ;) | 14:45 |
| admin0 | jrosser, you are also using osa + ceph ? is it bluestore ? | 14:45 |
| jrosser | yes, both | 14:45 |
| jrosser | but we deploy ceph entirely independantly of OSA | 14:46 |
| jrosser | with ceph-ansible directly | 14:46 |
| admin0 | can you help me .. i want to do bluestore .. independent ceph-ansible is also fine to get this stalled process going | 14:46 |
| jrosser | i can try, stuartgr knows our ceph the best | 14:47 |
| evrardjp | odyssey4me: great | 14:47 |
| *** armaan has joined #openstack-ansible | 14:47 | |
| tux__ | Folks i am using bluestore but why its using xfs ? http://paste.openstack.org/show/726352/ | 14:48 |
| tux__ | I thought it use raw disk | 14:48 |
| evrardjp | tux__: we rely on cephansible -- have a look at what this mean in their documentation ... | 14:49 |
| mgariepy | tux__, what's in your config ? | 14:52 |
| mgariepy | tux__, what "ceph --version" returns on you osd node? | 14:53 |
| tux__ | madorn: evrardjp: http://paste.openstack.org/show/726353/ | 14:53 |
| tux__ | ceph version 12.2.6 (488df8a1076c4f5fc5b8d18a90463262c438740f) luminous (stable) | 14:53 |
| tux__ | mgariepy: ^^ | 14:53 |
| tux__ | evrardjp: ^^ | 14:54 |
| tux__ | Can you shared your config to see if i am doing something stupid ? | 14:54 |
| *** lvdombrkr89 has quit IRC | 14:55 | |
| devx | o/ | 14:55 |
| *** armaan has quit IRC | 14:57 | |
| openstackgerrit | Jean-Philippe Evrard proposed openstack/openstack-ansible master: Remove group vars hack https://review.openstack.org/584411 | 14:57 |
| mgariepy | tux__, from : http://docs.ceph.com/ceph-ansible/master/osds/scenarios.html seems like xfs is normal | 14:58 |
| *** rpittau_ has quit IRC | 14:59 | |
| tux__ | Look like they are putting journal WAL/DB on XFS filesystem | 14:59 |
| tux__ | data disk is none filesystem base right? | 15:00 |
| mgariepy | blkid /dev/sdb* | 15:00 |
| *** mattinator has joined #openstack-ansible | 15:00 | |
| *** FrankZhang has joined #openstack-ansible | 15:00 | |
| tux__ | mgariepy: http://paste.openstack.org/show/726354/ | 15:01 |
| tux__ | ceph divided disk in two part sdb1 and sdb2 (look like sdb1 is journal) | 15:02 |
| *** lbragstad has quit IRC | 15:03 | |
| *** rmart04 has quit IRC | 15:03 | |
| mgariepy | i use the lvm scenario for my ceph cluster. | 15:06 |
| tux__ | Does anyone has ceph config with LVM please share :) | 15:06 |
| tux__ | mgariepy: can you tell me how you using LVM with Ceph ? | 15:07 |
| tux__ | Do i need to create lvm manually ? and how ? | 15:07 |
| tux__ | We have 4 node cluster and almost 30 OSD disk | 15:07 |
| mgariepy | tux__, http://docs.ceph.com/ceph-ansible/master/osds/scenarios.html | 15:07 |
| tux__ | mgariepy: i am on same page and read it | 15:08 |
| mgariepy | last example | 15:08 |
| tux__ | I have SSD so should i need to create 3 lv on single SSD disk ? 1. WAL 2. DB 3. data ? | 15:09 |
| tux__ | bluestore keep data and journal on same disk | 15:09 |
| tux__ | thats all i want to understand how did you create LVM partition ? | 15:09 |
| tux__ | it would be helpful if you share your config to i get some hint :) | 15:10 |
| mgariepy | tux__, how many ssd do you have? | 15:10 |
| tux__ | 1 node i have 4 SSD (500GB) at present | 15:11 |
| tux__ | in future i will add more when i need more space | 15:11 |
| tux__ | i have total 5 server currently | 15:11 |
| mgariepy | so 7 or 8 osd per ssd ? | 15:11 |
| mgariepy | 4 ssd and 30 hdd ? | 15:11 |
| tux__ | Let me tell you want i have | 15:12 |
| *** lbragstad has joined #openstack-ansible | 15:12 | |
| tux__ | 5x server and 20x physical SSD (500GB) | 15:12 |
| tux__ | I want to give 1 OSD per physical SSD i think that is idle case right? | 15:13 |
| tux__ | This is first cluster no idea what would be good here, what do yo suggests? | 15:13 |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible-ops master: Add functional tests to elk_metrics_6x https://review.openstack.org/583794 | 15:13 |
| tux__ | 1 OSD per physical disk. | 15:14 |
| tux__ | in that case i will have total 20 OSD in my 5 node cluster | 15:14 |
| jrosser | odyssey4me: this has a parent that looks like master? https://review.openstack.org/#/c/584394/ | 15:15 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-os_magnum master: Execute service setup against a delegated host using Ansible built-in modules https://review.openstack.org/582296 | 15:17 |
| *** indistylo has quit IRC | 15:18 | |
| odyssey4me | jrosser: gerrit doesn't think so? | 15:18 |
| jrosser | https://git.openstack.org/cgit/openstack/openstack-ansible/commit/?id=44c98893b48693d17231842e33336e6575bd4a31 ? | 15:19 |
| jrosser | what am i doing wrong? | 15:19 |
| *** dave-mccowan has joined #openstack-ansible | 15:20 | |
| odyssey4me | jrosser: that's definitely the queens patch because I had to combine some things | 15:20 |
| openstackgerrit | Merged openstack/openstack-ansible-repo_build stable/queens: tasks: repo_build_wheels: Download get-pip.py if it does not exist https://review.openstack.org/584355 | 15:21 |
| openstackgerrit | Merged openstack/openstack-ansible-pip_install stable/ocata: Update get-pip to version 3.3 https://review.openstack.org/584109 | 15:21 |
| *** dave-mcc_ has joined #openstack-ansible | 15:23 | |
| *** dave-mccowan has quit IRC | 15:25 | |
| Dmitriy | as everybody is here, can someone take a look at https://review.openstack.org/#/c/583943/ ? | 15:26 |
| jrosser | odyssey4me: could you confirm the order we apply these to stable/queens? | 15:26 |
| odyssey4me | jrosser: as it shows in gerrit, from the bottom up - alternatively just use the 'checkout' download option from gerrit | 15:30 |
| logan- | does anyone use /etc/openstack_deploy/host_vars or group_vars on queens? seems like ansible is not loading the 2nd inventory | 15:31 |
| jamesdenton | @cloudnull Thanks for looking at the patch set for OVN. It's a little sketchy, as you can tell. I didn't want to risk losing my work ;D | 15:31 |
| odyssey4me | so from https://review.openstack.org/#/c/568637/7 - download, click copy on checkout | 15:31 |
| odyssey4me | git fetch https://git.openstack.org/openstack/openstack-ansible refs/changes/37/568637/7 && git checkout FETCH_HEAD | 15:31 |
| odyssey4me | logan- admin0 said the same yesterday | 15:31 |
| odyssey4me | I don't know if it needs an inventory.ini there before it works, or what | 15:31 |
| logan- | yeah i am looking at admin0's env and confirming it | 15:31 |
| logan- | inventory.ini is present etc | 15:31 |
| logan- | ANSIBLE_INVENTORY env contains both inventories | 15:32 |
| odyssey4me | well, then that's a pickle | 15:32 |
| odyssey4me | I wonder if ANSIBLE_INVENTORY can be set to use both, or whether ansible.cfg has to be used | 15:32 |
| logan- | for a different project i have on 2.5 we use ANSIBLE_INVENTORY="inventory1,inventory2,inventory3" etc to stack them | 15:34 |
| logan- | guess ill run some tests with 2.4 and see whats up | 15:34 |
| jrosser | we use group_vars on queens but not host_vars and the group_vars data seems ok | 15:34 |
| logan- | thanks jrosser, thats helpful | 15:35 |
| *** epalper has quit IRC | 15:35 | |
| jrosser | and its 17.0.4 give or take a little bit | 15:35 |
| logan- | confirmed group_vars works | 15:37 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-os_magnum master: Execute service setup against a delegated host using Ansible built-in modules https://review.openstack.org/582296 | 15:37 |
| cloudnull | jamesdenton it looks great! | 15:39 |
| cloudnull | i +/- just a could nits | 15:39 |
| openstackgerrit | Merged openstack/openstack-ansible-os_searchlight master: Remove the unnecessary space https://review.openstack.org/582777 | 15:39 |
| cloudnull | I've not given OVN a try yet but it seems like a good start | 15:39 |
| odyssey4me | xgerman_: these vars, could you help me understand their purpose? is that for octavia to jump on neutron's RPC queue or is that for octavia to have a special RPC queue which neutron listens on? | 15:40 |
| xgerman_ | o/ | 15:40 |
| xgerman_ | which ones? | 15:41 |
| odyssey4me | oh, forgot to paste the link :p | 15:41 |
| odyssey4me | https://github.com/openstack/openstack-ansible-os_octavia/blob/master/defaults/main.yml#L256-L270 | 15:41 |
| logan- | i wonder if the host_vars provided in dynamic_inventory override everything in subsequent inventories | 15:41 |
| openstackgerrit | Merged openstack/openstack-ansible-os_neutron stable/pike: Install openvswitch packages on 'ml2.ovs.dvr' configuration https://review.openstack.org/581041 | 15:42 |
| jamesdenton | cloudnull I will have some questions for you after this next set of patches re: best practices | 15:42 |
| openstackgerrit | Merged openstack/openstack-ansible-tests stable/queens: Update get-pip to version 3.3 https://review.openstack.org/584090 | 15:42 |
| cloudnull | jamesdenton whenever you're ready, fire away :) | 15:43 |
| xgerman_ | odyssey4me: all this is for synchronizing between neutron lbaas and octavia/if you run octavia stand-alone you don’t need that | 15:43 |
| logan- | lol: "A bug was fixed with the inventory path/directory, which was defaulting to the current working directory. This caused group_vars and host_vars to be picked up from the current working directory instead of just adjacent to the playbook or inventory directory when a host list (comma separated host names) was provided as inventory." | 15:44 |
| odyssey4me | xgerman_: ok, so it's not neutrons primary RPC queue, it's actually a totally separate queue dedicated to this function | 15:44 |
| xgerman_ | yes, it is. | 15:44 |
| odyssey4me | well, queue is the wrong word - let's ask this another way... if this was in its own vhost, would that work? I'm not saying it will be, but it possibly could be | 15:45 |
| odyssey4me | the way it's set out right now it actually lives on the same rpc vhost as neutron, which one might say is a security risk given that you could have people doing load balancer things and flooding the neutron rpc queue | 15:46 |
| xgerman_ | so it needs to be accessed by both octavia and neutron — so I was putting it into neutron’s vhost… the access permissions are a bit funky and it didn’t work for a while unless you had the same user everywhere | 15:46 |
| *** electrofelix has quit IRC | 15:47 | |
| odyssey4me | mind if I make the prefix for those all octavia_neutron_* so that they're namespaced? | 15:47 |
| openstackgerrit | Merged openstack/openstack-ansible-nspawn_container_create master: Update qgroup setup to be more explicit https://review.openstack.org/584122 | 15:47 |
| odyssey4me | and we would expect neutron to setup this vhost/user/password, correct? | 15:47 |
| odyssey4me | octavia is a consumer, not the creator | 15:47 |
| openstackgerrit | Merged openstack/openstack-ansible-tests stable/pike: Update get-pip to version 3.3 https://review.openstack.org/584091 | 15:48 |
| xgerman_ | octavia sends to neutron — so the producer | 15:48 |
| xgerman_ | but yes, that all makes sense… but I would make octavia create it since this is really for a specific case we are trying to deprecate | 15:48 |
| odyssey4me | so should we, in effect, have a vhost called something like '/octavia-neutron' instead, and octavia creates it and neutron connects to it if lbaas is enabled? | 15:49 |
| odyssey4me | then this vhost would have its own user/password, although it'll default to the octavia one | 15:50 |
| johnsom | odyssey4me Octavia is a stand alone service that uses neutron/nova/glance/barbican stable APIs. So we should be careful to not label anything that is an Octavia config with "neutron", only neutron-lbaas stuff. | 15:50 |
| johnsom | Only the deprecated neutron-lbaas integrates into neutron. | 15:51 |
| xgerman_ | well, the event streamer is a weird animal | 15:51 |
| odyssey4me | ok, I think that works though - we have other similar vars in other roles where the prefix is the role name the vars is in and then the rest of the var may have a setting for another service to access an API or whatever | 15:52 |
| odyssey4me | eg nova_glance_... | 15:52 |
| xgerman_ | yeah, as I said we are trying to deprecate that so we should condition all of that on if neutron_lbaas is enabled AND configured with octavia | 15:53 |
| *** chyka has joined #openstack-ansible | 15:53 | |
| odyssey4me | so I'm just making sure I move this to something similar, because right now this is namespaced for neutron which can make stuff complicated when you're needing to scale very big and use multiple rpc clusters | 15:53 |
| johnsom | Right, it is super important to make that split that Octavia is not neutron nor a neutron project | 15:53 |
| odyssey4me | ok cool, thanks for the clarity - I'll be adjusting https://review.openstack.org/583388, or perhaps put a patch on top of it, to fix that up. | 15:54 |
| odyssey4me | perhaps on top - given it'll be a significant set of changes | 15:55 |
| odyssey4me | that'll be for next week though | 15:55 |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible-ops master: Add functional tests to elk_metrics_6x https://review.openstack.org/583794 | 15:56 |
| *** udesale__ has joined #openstack-ansible | 16:00 | |
| *** udesale__ has quit IRC | 16:05 | |
| openstackgerrit | Merged openstack/openstack-ansible-tests stable/ocata: Update get-pip to version 3.3 https://review.openstack.org/584092 | 16:08 |
| openstackgerrit | Merged openstack/openstack-ansible-os_octavia master: Revert "Temporarily set functional test to non-voting" https://review.openstack.org/584298 | 16:08 |
| odyssey4me | I'm out for the w/end. Cheers folks! | 16:15 |
| mgariepy | cya odyssey4me | 16:15 |
| jrosser | bye odyssey4me | 16:16 |
| *** suggestable has quit IRC | 16:18 | |
| openstackgerrit | Merged openstack/openstack-ansible-os_nova master: Switch compute to use different inventory_hostname https://review.openstack.org/584049 | 16:19 |
| openstackgerrit | Merged openstack/openstack-ansible-os_nova master: Move MQ vhost/user creation into role https://review.openstack.org/583387 | 16:19 |
| *** poul_ has quit IRC | 16:25 | |
| *** gkadam has quit IRC | 16:29 | |
| openstackgerrit | Merged openstack/openstack-ansible-os_keystone master: Remove the unnecessary space https://review.openstack.org/582791 | 16:31 |
| *** gillesMo has quit IRC | 16:31 | |
| openstackgerrit | Merged openstack/openstack-ansible-os_magnum master: Update to use oslo.messaging service for RPC and Notify https://review.openstack.org/579645 | 16:38 |
| openstackgerrit | Merged openstack/openstack-ansible-os_octavia master: Make certificate generation host configurable https://review.openstack.org/582641 | 16:45 |
| *** cshen has joined #openstack-ansible | 16:53 | |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible-ops master: Move heartbeat from utility_all to kibana https://review.openstack.org/584444 | 16:54 |
| *** DanyC has quit IRC | 16:56 | |
| jrosser | cloudnull: i'd been wondering if it was worth a dedicated set of elastic master nodes | 16:57 |
| jrosser | in light of how heavy the load is if the cluster is recovering, getting those off the log nodes might have some merit | 16:58 |
| *** cshen has quit IRC | 16:58 | |
| openstackgerrit | James Denton proposed openstack/openstack-ansible-plugins master: Updates provider_network plugin to support geneve https://review.openstack.org/584447 | 17:06 |
| *** cshen has joined #openstack-ansible | 17:07 | |
| tux__ | Quick question can we convert ansible output in JSON or any good formate | 17:09 |
| tux__ | its hard to read when Error output is super log and not formated | 17:09 |
| cloudnull | jrosser ++ breaking the elastic nodes out to dedicated gear would help a bunch | 17:11 |
| cloudnull | tux__ there's ara that might be of some use there. | 17:12 |
| openstackgerrit | German Eichberger proposed openstack/openstack-ansible-os_octavia master: Fix quota for Octavia https://review.openstack.org/571368 | 17:12 |
| openstackgerrit | Markos Chandras (hwoarang) proposed openstack/openstack-ansible-os_horizon master: Add support for using distribution packages for OpenStack services https://review.openstack.org/584330 | 17:13 |
| cloudnull | jrosser having masters and coordinators away from your data and ingest nodes makes for a far more stable arch | 17:14 |
| cloudnull | would could use a specialized elastic_master group and override the master node election when that group is present? | 17:14 |
| jrosser | sounds reasonable | 17:16 |
| jrosser | im running a hyperdiverged setup(!) so it would be easy to slide the elk master off onto a quieter node | 17:17 |
| jrosser | i think the only thing i need to keep an eye on is memory usage | 17:18 |
| openstackgerrit | Frank Zhang proposed openstack/openstack-ansible master: Add aio host conf for barbican role https://review.openstack.org/579284 | 17:19 |
| openstackgerrit | Andy Smith proposed openstack/openstack-ansible-os_nova master: Setup oslo.messaging extra packages for optional drivers https://review.openstack.org/584450 | 17:20 |
| cloudnull | jrosser with the retention policy changes, how have things been > | 17:22 |
| cloudnull | ? | 17:22 |
| jamesdenton | cloudnull any suggestions on using the systemd module? | 17:23 |
| cloudnull | did it finally plateau | 17:23 |
| cloudnull | jamesdenton just 's/service/systemd/g' :) | 17:23 |
| jamesdenton | oh, cool | 17:24 |
| jamesdenton | thats easy :D | 17:24 |
| cloudnull | ++ :) | 17:24 |
| jrosser | cloudnull: inhavent checked today, ill take a look later, but its certainly growing | 17:26 |
| jrosser | the search is still "sticky" though | 17:26 |
| cloudnull | you think that's because there's just more to search, or something else going on ? | 17:28 |
| jrosser | i need to see if its io bound | 17:28 |
| jrosser | the disks are very busy going by the lights | 17:28 |
| jrosser | each node has a raidz mirror | 17:29 |
| *** FrankZhang has quit IRC | 17:30 | |
| jrosser | i wouod have thought the search should not slow down because the indexes are split per day | 17:30 |
| jrosser | so for something "today" the amount of data to trawl is fairly constant | 17:31 |
| *** cshen has quit IRC | 17:32 | |
| mgariepy | tux__, https://getara.org/ | 17:33 |
| jrosser | cloudnull: i also thought that a modestly complex query in a cron job would put useful info into the query time stats | 17:33 |
| mgariepy | i think it should be added by default to osa deployment | 17:33 |
| jrosser | deploying ara alongside osa is|was a little difficult | 17:34 |
| openstackgerrit | Merged openstack/openstack-ansible-plugins master: Revert "Add a get_nested filter" https://review.openstack.org/584304 | 17:36 |
| tux__ | mgariepy: thanks!! | 17:36 |
| *** cshen has joined #openstack-ansible | 17:36 | |
| mgariepy | jrosser, that's why it should be installed by default ;) | 17:42 |
| jrosser | agreed :) | 17:44 |
| jrosser | particularly now all the no_log stuff is in.... | 17:45 |
| *** poul_ has joined #openstack-ansible | 17:47 | |
| openstackgerrit | Andy Smith proposed openstack/openstack-ansible-os_neutron master: Setup oslo.messaging extra packages for optional drivers https://review.openstack.org/583021 | 17:49 |
| *** cshen has quit IRC | 17:50 | |
| *** kstev has quit IRC | 17:51 | |
| *** poul_ has quit IRC | 17:55 | |
| admin0 | logan-, thanks .. the ceph-install config ran successfully .. | 17:55 |
| admin0 | so no group vars, no host vars .. direct yml file | 17:55 |
| *** DanyC has joined #openstack-ansible | 17:56 | |
| *** mattinator has quit IRC | 17:57 | |
| openstackgerrit | Andy Smith proposed openstack/openstack-ansible-os_cinder master: Setup oslo.messaging extra packages for optional drivers https://review.openstack.org/582632 | 17:57 |
| logan- | great | 18:01 |
| *** DanyC has quit IRC | 18:02 | |
| *** kstev has joined #openstack-ansible | 18:03 | |
| admin0 | do i need cephx: true ? | 18:08 |
| admin0 | please ignore that question | 18:10 |
| *** mattinator has joined #openstack-ansible | 18:11 | |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible-ops master: Add option to allow a deployer to specific specific master https://review.openstack.org/584462 | 18:12 |
| cloudnull | jrosser idk if that will work for your use case? | 18:12 |
| cloudnull | if you have suggestions on tuning it up feel free to make changes. | 18:13 |
| admin0 | logan-, setup-infra passed :) | 18:13 |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible-ops master: Add functional tests to elk_metrics_6x https://review.openstack.org/583794 | 18:14 |
| jrosser | cloudnull: i'll take a look, will be next week before i can run it i think | 18:14 |
| cloudnull | np | 18:21 |
| cloudnull | I'm busy making it so that we've some testing on all this anyway :) | 18:21 |
| cloudnull | I just need to get the logs to sync and that should be good | 18:22 |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible-ops master: Move heartbeat from utility_all to kibana https://review.openstack.org/584444 | 18:23 |
| openstackgerrit | Andy Smith proposed openstack/openstack-ansible-os_keystone master: Setup oslo.messaging extra packages for optional drivers https://review.openstack.org/584479 | 18:26 |
| *** kstev has quit IRC | 18:27 | |
| *** kstev has joined #openstack-ansible | 18:29 | |
| *** vnogin has joined #openstack-ansible | 18:35 | |
| openstackgerrit | Merged openstack/openstack-ansible stable/pike: Update get-pip to version 3.3 https://review.openstack.org/584098 | 18:35 |
| openstackgerrit | Merged openstack/openstack-ansible stable/ocata: Update get-pip to version 3.3 https://review.openstack.org/584101 | 18:35 |
| openstackgerrit | Merged openstack/openstack-ansible master: Skip provider_networks module if possible https://review.openstack.org/581802 | 18:35 |
| *** vnogin has quit IRC | 18:35 | |
| *** vnogin_ has joined #openstack-ansible | 18:35 | |
| ebbex_ | how or at what point is python installed in the lxc_container_create role? | 18:40 |
| *** vnogin_ has quit IRC | 18:41 | |
| logan- | ebbex_: it is not installed into each individual container, it gets installed into the base container image https://github.com/openstack/openstack-ansible-lxc_hosts/blob/eeb21321f47244e12e572e3b83368645777806c0/vars/ubuntu-16.04.yml#L48-L50 | 18:42 |
| logan- | so inside lxc_hosts instead of lxc_container_create | 18:42 |
| *** NostawRm has quit IRC | 18:44 | |
| *** NostawRm has joined #openstack-ansible | 18:44 | |
| ebbex_ | logan-: Thanks, I'm looking into why python is not installed by the time this patch kicks in; https://review.openstack.org/#/c/545316/ | 18:44 |
| *** poul has joined #openstack-ansible | 18:46 | |
| *** rmart04 has joined #openstack-ansible | 18:47 | |
| *** rmart04 has quit IRC | 18:49 | |
| *** poul has quit IRC | 18:51 | |
| d34dh0r53 | mnaser: did you ping me earlier? sorry my scrollback is screwed up | 18:52 |
| *** jproulx1 has quit IRC | 18:55 | |
| d34dh0r53 | anyone seen this on master? http://paste.openstack.org/show/726367/ | 18:56 |
| cloudnull | d34dh0r53 nope | 18:58 |
| d34dh0r53 | hmm | 18:58 |
| *** poul has joined #openstack-ansible | 19:03 | |
| logan- | would be curious to see the python traceback from -vvv d34dh0r53 | 19:08 |
| d34dh0r53 | logan-, cloudnull: cjloader says it's my ansible version, I've done a bunch of pike deploys on this mnaio so that may very well be the case | 19:08 |
| logan- | ahh | 19:09 |
| *** joabdearaujo has quit IRC | 19:09 | |
| logan- | that would make sense, master connection plugin is incompatible with ansible<2.5 | 19:09 |
| d34dh0r53 | yep, I think that's what it is | 19:09 |
| *** joabdearaujo has joined #openstack-ansible | 19:09 | |
| *** cshen has joined #openstack-ansible | 19:11 | |
| *** poul has quit IRC | 19:13 | |
| cjloader | logan-: i had same error when i deployed originally | 19:13 |
| *** poul has joined #openstack-ansible | 19:13 | |
| *** poul_ has joined #openstack-ansible | 19:13 | |
| *** poul has quit IRC | 19:14 | |
| *** poul has joined #openstack-ansible | 19:14 | |
| openstackgerrit | Merged openstack/openstack-ansible-ops master: Add functional tests to elk_metrics_6x https://review.openstack.org/583794 | 19:15 |
| *** poul_ has quit IRC | 19:18 | |
| openstackgerrit | Andy Smith proposed openstack/openstack-ansible-os_glance master: Setup oslo.messaging extra packages for optional drivers https://review.openstack.org/584496 | 19:23 |
| *** poul has quit IRC | 19:25 | |
| admin0 | n00b question .. when i login to the ceph node and do ceph command .. like ceph df etc .. it gives unable to find keyring .etc | 19:44 |
| logan- | yeah the admin keyring is not copied to the osd nodes by default | 19:44 |
| logan- | log in to the mons and it should work | 19:44 |
| admin0 | aah ! .. got it :) | 19:45 |
| tux__ | If i enable JBOD mode in RAID controller how does OS disk will get mirror (RAID-1) ? Do i need to use software RAID? | 19:52 |
| guilhermesp | as ceph/osa is a hot topic.... ceph/osa users: do you share the same ceph cluster with more then one osa deployment? If so, how do you do that? | 19:53 |
| logan- | guilhermesp: i do not, but it would not be too difficult to do if you use different pool names for each osa environment. you probably only want to use ceph-ansible in one of the osa envs, and then in the other envs just connect to the cluster using the ceph-client settings to connect to an existing cluster | 19:54 |
| guilhermesp | yeah logan- that's what I was thinking.... I want to create another pool using one of my rulesets (easy part). What I imagine is: the computes of the outer cloud needs to connect to ceph public network (I'm using br-storage as ceph public network ) | 19:59 |
| guilhermesp | then I only need to configure the client on these computes and then point nova.conf to use the newest pool that I created | 20:00 |
| logan- | yup, that should work | 20:00 |
| guilhermesp | I'm wondering if these outer computes needs to acess br-mgmt network of the inner cloud | 20:00 |
| logan- | hmm | 20:00 |
| guilhermesp | in order to communicate with ceph_mons | 20:00 |
| logan- | where are your mons | 20:00 |
| logan- | yeah they will need to communicate with mons | 20:00 |
| guilhermesp | so, I will need to plug two networks from inner cloud to outer cloud (br-mgmt and br-storage). I particularly don't think this is beautiful | 20:01 |
| guilhermesp | but seems to be the only way for my architecture | 20:02 |
| guilhermesp | thanks logan- :) | 20:04 |
| tux__ | logan-: I am installing fresh ceph-mon on existing OSA and its throwing this error http://paste.openstack.org/show/726373/ | 20:07 |
| tux__ | logan-: any idea? what i missed here | 20:07 |
| *** poul has joined #openstack-ansible | 20:07 | |
| logan- | looking at the output it seems like ceph-mon is not running, so you'll prob need to login and start looking at ceph-mon and why it isnt running | 20:09 |
| tux__ | hmm | 20:09 |
| logan- | np guilhermesp | 20:09 |
| tux__ | error: monitor data filesystem reached concerning levels of available storage space (available: 3% 321 MB) | 20:11 |
| tux__ | /dev/mapper/rootvg01-lv01 9.8G 9.5G 321M 97% /var/log | 20:12 |
| *** poul has quit IRC | 20:12 | |
| tux__ | my var partition is small | 20:12 |
| tux__ | how do i increase it ? | 20:12 |
| tux__ | in LXC | 20:12 |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible-ops master: Move heartbeat from utility_all to kibana https://review.openstack.org/584444 | 20:13 |
| guilhermesp | huum, logan- Actually I was seeing here that the computes (ceph.conf) of the inner cloud connects to the mon containers through br-storage (ceph-public network), so, I think br-mgmt wouldn't be required for the outer computes | 20:13 |
| guilhermesp | i.e 10.11.18.0/24 is my ceph_public/br-storage. This is one line of the mon's session of my ceph.confg a compute that uses ceph in the inner cloud: mon addr = 10.11.18.120 | 20:13 |
| tux__ | logan-: How do i increase disk size in Ceph-mon container ? Do i need to do by hand? | 20:16 |
| admin0 | tux__, it shares from the host system does it not ? | 20:19 |
| tux__ | it shared from host machine | 20:20 |
| tux__ | admin0: http://paste.openstack.org/show/726374/ | 20:21 |
| tux__ | Host var i have 403G available | 20:21 |
| tux__ | who decide container partition size ? | 20:22 |
| tux__ | from that output look like Host / (root) map with container /var/log :( | 20:24 |
| tux__ | that is bad idea | 20:24 |
| *** kstev has quit IRC | 20:26 | |
| tux__ | why ceph care about /var/log filesystem.. that is interesting | 20:27 |
| *** kstev has joined #openstack-ansible | 20:28 | |
| *** poul has joined #openstack-ansible | 20:36 | |
| jamesdenton | Anyone know how oslo.privsep gets installed on a node? | 20:36 |
| tux__ | http://paste.openstack.org/show/726375/ | 20:42 |
| tux__ | what is wrong here, may be ceph isn't starting because of missing keyring | 20:43 |
| jamesdenton | n/m, venv strikes again | 20:44 |
| *** kstev has quit IRC | 20:46 | |
| *** tux__ has quit IRC | 20:47 | |
| *** poul has quit IRC | 20:48 | |
| openstackgerrit | Andy Smith proposed openstack/openstack-ansible-os_heat master: Setup oslo.messaging extra packages for optional drivers https://review.openstack.org/584513 | 20:51 |
| *** poul has joined #openstack-ansible | 20:52 | |
| *** poul has quit IRC | 20:57 | |
| openstackgerrit | Andy Smith proposed openstack/openstack-ansible master: [WIP] Add qdrouterd role for messaging backend deployment https://review.openstack.org/570261 | 21:01 |
| *** cshen has quit IRC | 21:06 | |
| openstackgerrit | James Denton proposed openstack/openstack-ansible master: Adds haproxy endpoints for networking-ovn support https://review.openstack.org/584069 | 21:07 |
| openstackgerrit | James Denton proposed openstack/openstack-ansible-os_neutron master: Adds networking-ovn support https://review.openstack.org/584067 | 21:08 |
| *** cshen has joined #openstack-ansible | 21:08 | |
| *** jwitko has quit IRC | 21:08 | |
| *** RossSchulman[m] has quit IRC | 21:16 | |
| *** evrardjp[m] has quit IRC | 21:16 | |
| *** linshuicheng[m] has quit IRC | 21:16 | |
| *** jothibasu[m] has quit IRC | 21:16 | |
| *** lbragstad[m] has quit IRC | 21:16 | |
| *** asdf1234asdf[m] has quit IRC | 21:16 | |
| *** csmart[m] has quit IRC | 21:17 | |
| *** taildashef-work[ has quit IRC | 21:17 | |
| *** keithmnemonic[m] has quit IRC | 21:17 | |
| *** poul has joined #openstack-ansible | 21:18 | |
| admin0 | by default osa/ceph makes public-network = mgmt network ? | 21:22 |
| cloudnull | admin0 i think that's right | 21:23 |
| admin0 | this means the instances will use the mgmt for their storage traffic ? and not the br-storage | 21:26 |
| admin0 | is this why we have that 172.29.248.0/22 entry in some examples, as ceph replication network and then br-storage is used as publc | 21:29 |
| cloudnull | that could be... logan- may have some thoughts on that. | 21:30 |
| cloudnull | i just dont know enough to say thats right | 21:30 |
| cloudnull | IMO the instance traffic should not be on br-mgmt | 21:30 |
| cloudnull | though I still coming to terms w/ ceph | 21:30 |
| cloudnull | my lab env is new, and small. | 21:31 |
| *** poul has quit IRC | 21:32 | |
| *** KeithMnemonic has quit IRC | 21:34 | |
| admin0 | public network = vm <-> storage ; cluser network = only between osds for replication, heartbeat etc .. by default i see public network set as 236 and storage as 244 | 21:37 |
| cloudnull | maybe that should be changed in the example or is expected to be changed in config? | 21:38 |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible-ops master: Move heartbeat from utility_all to kibana https://review.openstack.org/584444 | 21:39 |
| jrosser | separating the ceph replication network guves you two things, visibility/accounting in your traffic stats of client vs internal ceph cluster traffic, and the opportunity to put replication on its own interfaces | 21:43 |
| jrosser | the compute nodes should communicate with the ceph public network over br-storage | 21:44 |
| admin0 | how to validate this jrosser | 21:44 |
| *** poul has joined #openstack-ansible | 21:45 | |
| jrosser | it depends what you have built | 21:45 |
| jrosser | traceroute from the computes to the osds | 21:46 |
| jrosser | tcpdump the traffic to verify | 21:46 |
| admin0 | i used the default ceph playbook in osa .. all things are working .. but i get less iops and i suspected networking .. in the config i see public network = 236.x and cluster as 244.x .. so in ceph, public = client - storage communication and cluster = osd-osd commnication which is why i asked | 21:47 |
| jrosser | that feels suspect | 21:47 |
| admin0 | and then i recall seeing 248.x in some examples so asked if that was what it is used for .. | 21:48 |
| jrosser | another reason to use br-storage is you can do jumbo frames, which really shouldnt be on br-mgmt | 21:48 |
| *** tux_ has joined #openstack-ansible | 21:48 | |
| admin0 | in that case, i need to override the config and put public_network on the 244.x and then create a br-replicaiton or something in the OSDs, give it the 248.x and sepcify in the config and run the playbooks again | 21:49 |
| *** poul has quit IRC | 21:49 | |
| jrosser | unless you have an actual separate set of interfaces for the private network i wonder what gain there is tbh | 21:50 |
| * jrosser out for today | 21:50 | |
| admin0 | what will happen if i give the same network for both ? | 21:50 |
| admin0 | public + cluster = 244.x | 21:50 |
| admin0 | i want tthe vms to not use br-mgmt | 21:50 |
| tux_ | If you give both same and if any server crash then cluster network start shuffling data which can overrun your public network | 21:54 |
| tux_ | same amount of ingress traffic will go out from cluster network for replication thats what i think | 21:55 |
| tux_ | if you use single NIC for both then it may be tight during recovery | 21:55 |
| tux_ | if you have 40G that make difference and you can run on same NIC | 21:56 |
| admin0 | i am thinking 2x 40gb for br-storage (bond) and 1x 40gb for br-replication | 21:56 |
| *** poul has joined #openstack-ansible | 22:02 | |
| admin0 | do you guys prefer br-ceph or br-replication ? | 22:02 |
| cloudnull | take care jrosser | 22:02 |
| cloudnull | i too need to step away a few, need to go put some food on the gril | 22:03 |
| cloudnull | take care all | 22:03 |
| admin0 | happy weekend | 22:03 |
| cloudnull | you too admin0 | 22:03 |
| *** asdf1234asdf[m] has joined #openstack-ansible | 22:04 | |
| admin0 | almost there to produce a fully working osa+ceph detail step by step document | 22:04 |
| *** jwitko has joined #openstack-ansible | 22:05 | |
| cloudnull | as for the nic question, in prod environments with a storage component (netapp, ceph, solidfire, whatever) we typically have at least one 10/40G bond dedicated to all storage traffic. then have two vlans one for instance traffice, one for replication. | 22:05 |
| cloudnull | ok now i gotta run :) | 22:05 |
| *** poul has quit IRC | 22:07 | |
| *** cshen has quit IRC | 22:12 | |
| *** tux_ has quit IRC | 22:15 | |
| *** hachi has quit IRC | 22:16 | |
| *** tux_ has joined #openstack-ansible | 22:18 | |
| *** tux_ has quit IRC | 22:23 | |
| admin0 | when using ceph, glance cinder and nova are the only 3 clients of ceph right ? or other components use it also ? | 22:26 |
| *** coreywright has quit IRC | 22:27 | |
| *** cshen has joined #openstack-ansible | 22:31 | |
| *** keithmnemonic[m] has joined #openstack-ansible | 22:31 | |
| *** csmart[m] has joined #openstack-ansible | 22:31 | |
| *** jothibasu[m] has joined #openstack-ansible | 22:31 | |
| *** evrardjp[m] has joined #openstack-ansible | 22:31 | |
| *** linshuicheng[m] has joined #openstack-ansible | 22:31 | |
| *** RossSchulman[m] has joined #openstack-ansible | 22:31 | |
| *** lbragstad[m] has joined #openstack-ansible | 22:31 | |
| *** taildashef-work[ has joined #openstack-ansible | 22:31 | |
| openstackgerrit | Merged openstack/openstack-ansible-ops master: Move heartbeat from utility_all to kibana https://review.openstack.org/584444 | 22:33 |
| *** cshen has quit IRC | 22:36 | |
| admin0 | is it possible to plug an instance to a bridge manually ? | 22:37 |
| admin0 | a container i mean | 22:37 |
| admin0 | when i change my ceph public_network to .244, ceph-mgr process fails saying the container does not have any .244/22 ip range .. which is not present .. .. so is there an easy way exists to add an interface to the ceph-mon-container to also be in the br-storage | 22:39 |
| *** coreywright has joined #openstack-ansible | 22:39 | |
| *** poul has joined #openstack-ansible | 22:46 | |
| admin0 | aah .. group binds :D | 22:47 |
| *** tux_ has joined #openstack-ansible | 22:49 | |
| tux_ | admin0: in my network i setup br-cluster for replication :) because it match with ceph naming convention | 22:58 |
| admin0 | tux_, are you sure you are using br-storage for instance <-> storage traffic | 22:59 |
| admin0 | because my tests show otherwise | 22:59 |
| tux_ | br-storage on compute node for storage traffic | 23:00 |
| admin0 | tux on your ceph.conf, what is the value of public_network ? | 23:01 |
| admin0 | in the mons and in the osd nodes | 23:01 |
| tux_ | admin0: http://paste.openstack.org/show/726378/ | 23:03 |
| *** poul has quit IRC | 23:03 | |
| tux_ | mon and public should be on same network | 23:04 |
| tux_ | BRB | 23:06 |
| *** tux_ has quit IRC | 23:06 | |
| *** cshen has joined #openstack-ansible | 23:08 | |
| *** cshen has quit IRC | 23:13 | |
| *** poul has joined #openstack-ansible | 23:13 | |
| *** tux_ has joined #openstack-ansible | 23:14 | |
| *** poul has quit IRC | 23:18 | |
| admin0 | \o/ | 23:20 |
| admin0 | a small patch and detail document coming up | 23:21 |
| *** poul has joined #openstack-ansible | 23:30 | |
| *** poul has quit IRC | 23:35 | |
| *** chyka has quit IRC | 23:36 | |
| tux_ | admin0: ?? | 23:41 |
| admin0 | nah .. not so simple ... now the network is good, but openstack still puts mon in the mgnt range and not in storage | 23:41 |
| tux_ | delete all your facts and try | 23:42 |
| admin0 | no . this is different | 23:42 |
| tux_ | i had same issue and after clean up facts it fixed my issue | 23:42 |
| admin0 | the default ceph playbooks do not do this | 23:42 |
| admin0 | before every run, i delete the facts | 23:42 |
| tux_ | so you are deploying outside OSA | 23:42 |
| admin0 | no | 23:42 |
| admin0 | with oSA | 23:42 |
| admin0 | if outside osa, it would have been good already | 23:43 |
| tux_ | Do you have br-storage network on ceph-mon container? | 23:43 |
| tux_ | make sure you have br-storage network attached to Ceph-mon | 23:44 |
| admin0 | first thing is in user_variables,you have to override the public_ and cluster_ network .. but as soon as you do that, ceph manager will fail because our mon containers do not come with br-storage IP .. so i had to use the group binds .. that fixed the ceph networking part .. but when i create a new volume, libvirt still maps to br-mgmt ips of mon | 23:44 |
| tux_ | can you paste your user_varibles config ? | 23:45 |
| admin0 | tux you are setting mon_host override in variables ? | 23:46 |
| tux_ | yes | 23:47 |
| admin0 | it should be _ there right ? mon_host | 23:47 |
| admin0 | in the paste you gave | 23:47 |
| *** poul has joined #openstack-ansible | 23:47 | |
| tux_ | http://paste.openstack.org/show/726379/ | 23:48 |
| tux_ | this is what it looks in user_var | 23:48 |
| *** tosky has quit IRC | 23:48 | |
| admin0 | so you have dedicated mons ? | 23:48 |
| tux_ | no its OSA container | 23:49 |
| admin0 | or are those IPs from what lxc-container assigned ? | 23:49 |
| admin0 | ok | 23:49 |
| admin0 | got it . so first run setup-hosts, let it create the container and assign ip.. then use thsoe IPs in the variables | 23:49 |
| tux_ | yes | 23:49 |
| tux_ | you are correct there | 23:49 |
| admin0 | did you also used bind_host to add ceph-mon so that br-storage is connected ot the container ? | 23:50 |
| *** poul has quit IRC | 23:52 | |
| admin0 | i see this key possbility: monitor_address_block: "{{ cidr_networks.storage}}"" | 23:55 |
| admin0 | that way, i think no need to specify the monitors ip | 23:55 |
| admin0 | and it should auto take it | 23:55 |
| tux_ | yes i did bind-host for br-storage | 23:57 |
| admin0 | ok | 23:57 |
| tux_ | http://paste.openstack.org/show/726380/ | 23:58 |
| tux_ | here its | 23:58 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!