| *** shoutm has quit IRC | 00:18 | |
| *** sdake has joined #openstack-ansible | 00:18 | |
| *** shoutm has joined #openstack-ansible | 00:21 | |
| *** waded_ has joined #openstack-ansible | 01:12 | |
| *** waded has quit IRC | 01:12 | |
| *** galstrom_zzz is now known as galstrom | 01:15 | |
| *** shoutm has quit IRC | 01:23 | |
| *** shoutm has joined #openstack-ansible | 01:28 | |
| *** galstrom is now known as galstrom_zzz | 01:33 | |
| *** shoutm has quit IRC | 01:48 | |
| *** shoutm has joined #openstack-ansible | 01:50 | |
| *** prometheanfire has quit IRC | 01:58 | |
| *** woodard has joined #openstack-ansible | 02:01 | |
| *** woodard has quit IRC | 02:05 | |
| *** Mudpuppy has quit IRC | 02:13 | |
| *** palendae_ is now known as palendae | 02:39 | |
| *** galstrom_zzz is now known as galstrom | 02:51 | |
| *** shoutm_ has joined #openstack-ansible | 02:53 | |
| *** galstrom is now known as galstrom_zzz | 02:53 | |
| *** shoutm has quit IRC | 02:54 | |
| *** shoutm_ has quit IRC | 03:03 | |
| *** shoutm has joined #openstack-ansible | 03:03 | |
| *** Mudpuppy has joined #openstack-ansible | 03:14 | |
| *** shoutm_ has joined #openstack-ansible | 03:22 | |
| *** shoutm has quit IRC | 03:24 | |
| *** persia has quit IRC | 03:30 | |
| *** persia has joined #openstack-ansible | 03:32 | |
| *** prometheanfire has joined #openstack-ansible | 03:39 | |
| *** skamithi12 has joined #openstack-ansible | 03:39 | |
| *** skamithi12 has left #openstack-ansible | 03:43 | |
| *** shoutm_ has quit IRC | 03:46 | |
| *** shoutm has joined #openstack-ansible | 04:03 | |
| *** shoutm_ has joined #openstack-ansible | 05:31 | |
| *** shoutm has quit IRC | 05:34 | |
| *** javeriak has joined #openstack-ansible | 06:12 | |
| *** mattoliverau has quit IRC | 06:13 | |
| *** mattoliverau has joined #openstack-ansible | 06:14 | |
| *** javeriak has quit IRC | 06:17 | |
| *** Mudpuppy has quit IRC | 06:17 | |
| *** Mudpuppy has joined #openstack-ansible | 06:47 | |
| *** Mudpuppy has quit IRC | 06:52 | |
| *** javeriak has joined #openstack-ansible | 08:13 | |
| openstackgerrit | Jesse Pretorius proposed stackforge/os-ansible-deployment-specs: Add Liberty Release spec https://review.openstack.org/221189 | 08:17 |
|---|---|---|
| *** javeriak has quit IRC | 08:21 | |
| git-harry | odyssey4me: what is the impact of this - https://bugs.launchpad.net/openstack-ansible/+bug/1487353 | 08:52 |
| openstack | Launchpad bug 1487353 in openstack-ansible trunk "MySQL host address is not set properly for all .my.cnf files" [Low,In progress] - Assigned to Jesse Pretorius (jesse-pretorius) | 08:52 |
| *** gparaskevas has joined #openstack-ansible | 08:53 | |
| odyssey4me | git-harry the impact is that operators can't access galera as they expect to - it's also a change of behavior from juno to kilo | 09:07 |
| odyssey4me | ie juno had the ip set correctly, kilo does not | 09:08 |
| git-harry | odyssey4me: okay, so it has no impact on the actual services | 09:10 |
| odyssey4me | git-harry nope - all the services use the ansible module with the host set as an arg | 09:12 |
| evrardjp | hello everyone | 09:14 |
| evrardjp | I was trying to improve my storage performance this morning, and I grepped mtu on the os-ansible-deployment repo | 09:14 |
| evrardjp | I've seen there was an option to set container's mtu (container_mtu) | 09:15 |
| evrardjp | but I don't see anything that uses that, except lxc creation (so, there is no configuration of the os with that) | 09:15 |
| evrardjp | is that normal? | 09:15 |
| evrardjp | never tried it before | 09:15 |
| evrardjp | I also remember that it was a pain to move on the hosts from a standard 1500 mtu to a 9000 mtu, someone has advices? scripts in /etc/network/interfaces? | 09:17 |
| evrardjp | it was not so simple as adding mtu in /etc/network/interfaces, with bridges | 09:18 |
| odyssey4me | evrardjp as I recall you have to ensure that the entire chain of dependencies has the right mtu | 09:27 |
| evrardjp | indeed | 09:27 |
| evrardjp | the last part of my chain is the host and the container | 09:27 |
| odyssey4me | and you have to use post-up in most cases because the mtu doesn't properly apply on bridges with the simple mtu config option in /interfaces | 09:27 |
| evrardjp | the host causes my pain, and the container I didn't try | 09:27 |
| evrardjp | odyssey4me: that's exactly what I meant by "the host causes my pain" | 09:28 |
| evrardjp | ;) | 09:28 |
| evrardjp | should I stop trying? | 09:28 |
| odyssey4me | the only places where it may be important to do this is for the vxlan interfaces - neutron container and compute hosts I think | 09:29 |
| odyssey4me | but sometimes you may wish to do it for your storage networks too | 09:29 |
| evrardjp | would you care to explain me more? | 09:29 |
| evrardjp | I guess that there is an impact of encapsulation | 09:29 |
| evrardjp | but I'm sure if there is a performance impact | 09:29 |
| evrardjp | what I mean is, yes if you don't pay attention, the system won't work | 09:30 |
| odyssey4me | yes, for vxlan due to the encapsulation and the fact that vm's typically have 1500 mtu set | 09:30 |
| evrardjp | but I'm not sure that increasing mtu will increase performance of vxlans | 09:30 |
| evrardjp | so your advise is to change that default value | 09:30 |
| odyssey4me | linux vm's generally behave quite well if you set a lower mtu value for the dnsmasq service | 09:31 |
| evrardjp | and define container_mtu | 09:31 |
| evrardjp | ok | 09:31 |
| odyssey4me | but not all operating systems listen to that | 09:31 |
| evrardjp | ok | 09:31 |
| evrardjp | so you don't advise to change the mtu at all, or at least not increase it | 09:31 |
| evrardjp | for at least linux OSes guested in the cloud | 09:31 |
| odyssey4me | windows ignores it, so if your network mtu is 1500 and you therefore need the vm's to have a lower mtu - you'd have to hard set it in the images | 09:31 |
| evrardjp | ok | 09:32 |
| odyssey4me | the vxlan network mtu increase is simply to alleviate that issue - not for performance | 09:32 |
| odyssey4me | it's to allow the vm's and virtual networks to use an mtu of 1500, meaning you deal with less support calls | 09:32 |
| odyssey4me | the alternative is to ensure that every image you provide has the default mtu set to 1452 (I think that's the max value) to accommodate the vxlan encapsulation | 09:33 |
| evrardjp | I don't see a problem for setting a larger mtu, but it need to work... and for your last concern, I'm afraid that neutron isn't configured to take the extra bytes I put on my mtu | 09:34 |
| evrardjp | so I guess it's better to stick to 1500 | 09:35 |
| evrardjp | everywhere, except for vms -> 1452 | 09:35 |
| evrardjp | thanks for the info | 09:36 |
| openstackgerrit | Jesse Pretorius proposed stackforge/os-ansible-deployment: Set the Ubuntu mirror used based on the environment https://review.openstack.org/218611 | 09:38 |
| odyssey4me | evrardjp once you've worked it out, it seems to me that this may be a useful documentation entry :) | 09:39 |
| evrardjp | :) | 09:41 |
| *** sdake has quit IRC | 09:42 | |
| git-harry | odyssey4me: does anyone use teardown.sh? | 09:49 |
| *** jlvillal has quit IRC | 09:50 | |
| *** jlvillal has joined #openstack-ansible | 09:50 | |
| mattt | git-harry: yeah, because we have an open review for it atm | 09:50 |
| git-harry | mattt: yeah, I sort of referring to that. | 09:50 |
| git-harry | The script seems like a bad idea in general. I'd argue it should just be deleted | 09:51 |
| git-harry | It seems it's just easier to rebuild | 09:51 |
| *** skamithi12 has joined #openstack-ansible | 09:53 | |
| odyssey4me | git-harry yes, apparently some people do - it's seen as a way to re-do a lab without rekicking | 09:54 |
| odyssey4me | when you're dealing with hardware labs, a rebuild takes a lot of time - teardown is built for that scenario | 09:54 |
| *** gparaskevas has quit IRC | 09:54 | |
| odyssey4me | it's known that it's not perfect, but it's better than nothing | 09:55 |
| git-harry | odyssey4me: I think nothing would be better | 09:55 |
| *** shoutm_ has quit IRC | 09:57 | |
| skamithi12 | odyssey4me: gm. ready to talk about deployment through vagrant. | 10:02 |
| skamithi12 | changing the license on the project now as the first step to apache | 10:03 |
| odyssey4me | skamithi12 welcome! :) | 10:03 |
| evrardjp | git-harry: I'm using it | 10:05 |
| evrardjp | I was using my own script that was destroying the containers before | 10:05 |
| evrardjp | teardown is just another way to do it | 10:06 |
| skamithi12 | odyssey4me: license on the project changed. i looked at an example blueprint https://blueprints.launchpad.net/openstack-ansible/+spec/remove-upstream-repo-dependency is this the kind of format i should follow or you had something else in mind? | 10:11 |
| odyssey4me | skamithi12 so your blueprint has already been registered: https://blueprints.launchpad.net/openstack-ansible/+spec/deploy-with-vagrant | 10:12 |
| odyssey4me | have you configured gerrit yet? | 10:12 |
| odyssey4me | ie configured 'git review' | 10:12 |
| skamithi12 | no not yet. | 10:12 |
| openstackgerrit | Merged stackforge/os-ansible-deployment: Remove double register in swift handlers https://review.openstack.org/220449 | 10:13 |
| odyssey4me | ok, to put the spec together and submit it you'll need to do that | 10:13 |
| odyssey4me | let me give you a basic overview of the process | 10:13 |
| odyssey4me | skamithi12 under 'submitting code' there are some guiding bits which you should ideally work through and understand: https://github.com/stackforge/os-ansible-deployment/blob/master/CONTRIBUTING.rst | 10:14 |
| evrardjp | git-harry: why do you prefer "nothing" instead of teardown? If some stuff are not correctly cleaned, then it's maybe time to register a bug? :d | 10:14 |
| odyssey4me | skamithi12 the spec repo is here: https://github.com/stackforge/os-ansible-deployment-specs | 10:15 |
| odyssey4me | skamithi12 https://github.com/stackforge/os-ansible-deployment-specs/blob/master/specs/template.rst is the template for the spec that you'll need to copy and edit | 10:16 |
| odyssey4me | everyone, meet skamithi12 who'll be working on https://blueprints.launchpad.net/openstack-ansible/+spec/deploy-with-vagrant to bring his work on deploying openstack-ansible with vagrant into the project :) | 10:18 |
| mattt | skamithi12: hey there! thanks in advance for your work here :) | 10:19 |
| evrardjp | hello skamithi12 :) | 10:19 |
| odyssey4me | mancdaz with regards to https://bugs.launchpad.net/openstack-ansible/+bug/1491882 we saw that some time ago with neutron on compute nodes as I recall | 10:20 |
| openstack | Launchpad bug 1491882 in openstack-ansible "some services don't deploy cleanly on bare_metal if there are other services on same host" [Undecided,New] | 10:20 |
| odyssey4me | I can't remember how it was resolved, but it seems that the conversion of cinder_volume to default to on_metal has left that as a new issue. | 10:21 |
| evrardjp | skamithi12: good luck with vagrant-libvirt then :) | 10:21 |
| git-harry | evrardjp: teardown scripts on complicated systems such as this are hard to maintain and generally get little attention. For testing they are not a fair test when compared to a new build and I've personally been bitten too many times in the past from this type of cleanup script. | 10:21 |
| evrardjp | yup I understand | 10:22 |
| *** jlvillal has quit IRC | 10:22 | |
| evrardjp | git-harry: but it's not that hard with our lxc system, just destroying every container is easy | 10:22 |
| skamithi12 | mattt, evrardjp: thx. i don't like virtualbox, but would love a Vagrantfile using virtualbox. a colleague at work is working on developing a vagrant plugin where you give a graphviz file and it spits out a vagrantfile in virtualbox format or libvirt format. | 10:22 |
| skamithi12 | not sure how soon his work will be opensourced | 10:22 |
| evrardjp | skamithi12: same, don't like virtualbox but I have been bitten by vagrant-libvirt in the past... so having some ppl generating the good tools for it is really great | 10:24 |
| git-harry | evrardjp: yes, a lot of the configuration is done in containers and that helps but there is still a significant numbers of changes that are made to the hosts. I don't use the teardown.sh script myself so in that sense I'm not impacted. I was wondering if anyone else used it because if not it would be easier just to remove. | 10:24 |
| openstackgerrit | Hugh Saunders proposed stackforge/os-ansible-deployment: Remove double register in swift handlers https://review.openstack.org/221237 | 10:24 |
| evrardjp | so to answer your question git-harry, Until I have my testlab hosted in openstack, I'll continue to use teardown.sh ;) | 10:25 |
| evrardjp | then I'll probably won't care anymore | 10:26 |
| mancdaz | odyssey4me yes, that's that same bug, only for cinder | 10:28 |
| skamithi12 | evrardjp: the latest master branch of libvirt-vagrant has stable libvirt isolated network support, so building isolated bridges, like I do in vagrant-osad project is rock solid. I'm adding libvirt/QEMU udp unicast tunnel support to vagrant-libvirt this week, for stable point to point VM connections in case someone wants to go to the next level and add switches like cumulus linux/junos. i'll have a blog post about it soon | 10:33 |
| skamithi12 | n linuxsimba.com/vagrant.html. libvirt just got the patch last week thx to another colleague of mine. http://libvirt.org/git/?p=libvirt.git;a=commit;h=5c668a78d85b0d71e6ac8e23f2c605058b44df65. I have a ppa for this if you run Ubuntu 14.04. | 10:33 |
| evrardjp | nice | 10:34 |
| evrardjp | working for cumulus? interesting :D | 10:35 |
| skamithi12 | evrardjp: yes i do. do alot of network testing in virtual env. networking is growing up..becoming more like regular devops. | 10:37 |
| *** shoutm has joined #openstack-ansible | 10:39 | |
| evrardjp | :) | 10:39 |
| *** gparaskevas has joined #openstack-ansible | 10:40 | |
| evrardjp | there will be plenty of work to do relative to osad... whether it's integrating and testing some components or simply have additional playbooks outside osad with roles to install additional software | 10:40 |
| evrardjp | a bright future :) | 10:40 |
| openstackgerrit | Darren Birkett proposed stackforge/os-ansible-deployment: include cinder_all group vars file in cinder-common playbook https://review.openstack.org/221242 | 10:43 |
| mancdaz | andymccr odyssey4me https://review.openstack.org/#/c/221242/1 | 10:44 |
| mancdaz | odyssey4me andymccr just in case this becomes something that needs to be fixed for upgrades | 10:44 |
| mancdaz | even though in most deployment scenarios, cinder-volumes will be on it's own host | 10:44 |
| skamithi12 | evrardjp: i'm still learning openstack, is there a irc channels where openstack operators hang out. want to learn how how ppl manage openstack and then see how i can help improve openstack network troubleshooting. | 10:45 |
| openstackgerrit | Merged stackforge/os-ansible-deployment: Updated juno for new dev work - 3 Sep 2015 https://review.openstack.org/220151 | 10:46 |
| skamithi12 | odyssey4me: signed up on reviews.openstack.org | 10:47 |
| skamithi12 | read the coding guidelines and template.rst. anything else i need to review? | 10:48 |
| *** Mudpuppy has joined #openstack-ansible | 10:49 | |
| *** Mudpuppy has quit IRC | 10:54 | |
| *** jlvillal has joined #openstack-ansible | 11:03 | |
| odyssey4me | mancdaz ah yes, that's it :) | 11:08 |
| *** javeriak has joined #openstack-ansible | 11:10 | |
| odyssey4me | skamithi12 if you have that ready then you can clone the specs repo, checkout a branch to add your spec and git review it so that we can all take a look at your proposed changes | 11:10 |
| odyssey4me | skamithi12 operators typically hang out in #openstack | 11:11 |
| *** javeriak has quit IRC | 11:13 | |
| *** javeriak has joined #openstack-ansible | 11:24 | |
| openstackgerrit | Merged stackforge/os-ansible-deployment: Set client versions deployed to use global requirements https://review.openstack.org/220049 | 11:24 |
| openstackgerrit | Merged stackforge/os-ansible-deployment: Update master for liberty-3 testing https://review.openstack.org/220504 | 11:24 |
| *** skamithi12 has quit IRC | 11:37 | |
| git-harry | hughsaunders: https://review.openstack.org/#/c/218354/ | 11:42 |
| openstackgerrit | Darren Birkett proposed stackforge/os-ansible-deployment: include cinder_all group vars file in cinder-common playbook https://review.openstack.org/221242 | 11:44 |
| *** cloudtrainme has quit IRC | 12:03 | |
| *** javeriak has quit IRC | 12:04 | |
| *** javeriak has joined #openstack-ansible | 12:07 | |
| *** markvoelker has joined #openstack-ansible | 12:19 | |
| *** markvoelker has quit IRC | 12:20 | |
| *** markvoelker has joined #openstack-ansible | 12:20 | |
| *** woodard has joined #openstack-ansible | 12:25 | |
| *** tlian has joined #openstack-ansible | 12:32 | |
| *** tlian has quit IRC | 12:32 | |
| openstackgerrit | Darren Birkett proposed stackforge/os-ansible-deployment: include cinder_all group vars file in cinder-common playbook https://review.openstack.org/221242 | 12:32 |
| openstackgerrit | Darren Birkett proposed stackforge/os-ansible-deployment: include cinder_all group vars file in cinder-* playbooks https://review.openstack.org/221242 | 12:33 |
| *** tlian has joined #openstack-ansible | 12:37 | |
| *** shoutm has quit IRC | 12:39 | |
| openstackgerrit | Jesse Pretorius proposed stackforge/os-ansible-deployment: Updated MariaDB to the new release version (10.0) https://review.openstack.org/178259 | 12:46 |
| odyssey4me | cloudnull it appears that keys.gnupg.net does not serve keys properly via port 80, so I've switched the key server to ubuntu's which appears to work in https://review.openstack.org/178259 | 12:48 |
| openstackgerrit | Jesse Pretorius proposed stackforge/os-ansible-deployment: Updated MariaDB to the new release version (10.0) https://review.openstack.org/178259 | 12:51 |
| mgariepy | good morning everyone | 12:51 |
| openstackgerrit | Jesse Pretorius proposed stackforge/os-ansible-deployment: Updated MariaDB to the new release version (10.0) https://review.openstack.org/178259 | 12:52 |
| *** Mudpuppy has joined #openstack-ansible | 12:58 | |
| *** Mudpuppy has quit IRC | 12:59 | |
| *** spotz_zzz is now known as spotz | 12:59 | |
| *** Mudpuppy has joined #openstack-ansible | 13:00 | |
| *** KLevenstein has joined #openstack-ansible | 13:05 | |
| hughsaunders | if using a flat network for public, and not aio, br-vlan-veth <--> eth12 (or whatever the container inteface is called) needs to be created manually on compute hosts? | 13:07 |
| odyssey4me | good morning mgariepy :) | 13:07 |
| mgariepy | how was your weekend ? | 13:08 |
| odyssey4me | mgariepy pretty good - kept away from work and relaxed :) | 13:10 |
| *** woodard has quit IRC | 13:11 | |
| mgariepy | you do that ? i though you were working 24/7 ;p | 13:11 |
| hughsaunders | mgariepy: this is progress for odyssey4me :) | 13:15 |
| *** woodard has joined #openstack-ansible | 13:15 | |
| mgariepy | hehe ;) | 13:16 |
| odyssey4me | mgariepy I go through obsessive phases. :p | 13:19 |
| mgariepy | following the moon ? | 13:20 |
| hughsaunders | odyssey4me is waning | 13:20 |
| odyssey4me | hahahaha | 13:20 |
| *** pradk has joined #openstack-ansible | 13:22 | |
| odyssey4me | mancdaz build success! https://review.openstack.org/221242 andymccr are you happy with the final implementation? | 13:27 |
| andymccr | odyssey4me: that one worked for me! | 13:30 |
| openstackgerrit | Jesse Pretorius proposed stackforge/os-ansible-deployment: Updated MariaDB to the new release version (10.0) https://review.openstack.org/178259 | 13:31 |
| odyssey4me | andymccr great :) now I have to wonder whether the issue also counts for kilo and above? | 13:32 |
| andymccr | odyssey4me: nope shouldnt count for kilo | 13:32 |
| andymccr | we would've run into it for certain if it did | 13:32 |
| andymccr | because cinder is now on metal :) | 13:32 |
| andymccr | and an aio would definitely hit that problem | 13:32 |
| andymccr | we changed teh structure of our vars though | 13:33 |
| andymccr | *the | 13:33 |
| odyssey4me | cool - will mark as invalid as I suspected that the vars structure change makes that a non-issue | 13:33 |
| *** kerwin_bai has joined #openstack-ansible | 13:33 | |
| openstackgerrit | Merged stackforge/os-ansible-deployment: Used named veth pairs that match container https://review.openstack.org/219457 | 13:34 |
| openstackgerrit | Merged stackforge/os-ansible-deployment: Add galera_address to all playbooks installing galera_client https://review.openstack.org/207939 | 13:34 |
| openstackgerrit | Jesse Pretorius proposed stackforge/os-ansible-deployment: Add galera_address to all playbooks installing galera_client https://review.openstack.org/221301 | 13:37 |
| *** skamithi13 has joined #openstack-ansible | 13:45 | |
| *** javeriak has quit IRC | 13:48 | |
| *** mrstanwell has joined #openstack-ansible | 13:57 | |
| *** javeriak has joined #openstack-ansible | 13:57 | |
| *** galstrom_zzz is now known as galstrom | 13:58 | |
| odyssey4me | miguelgrinberg when you're in, if you have a gap - can you take a look at https://review.openstack.org/218184 and try to figure out why it's not working properly, then revise the patch to make it work | 13:59 |
| odyssey4me | miguelgrinberg I could be lazy and just make it use the openstack client for everything, but I think we need to fix that - feel free to make the patch your own | 14:00 |
| javeriak | guys, need a little help, one of my infra's continously errors out on ansible ssh, after the Flush net cache task | 14:06 |
| *** cloudtrainme has joined #openstack-ansible | 14:07 | |
| *** galstrom is now known as galstrom_zzz | 14:11 | |
| *** sigmavirus24_awa is now known as sigmavirus24 | 14:11 | |
| *** cloudtrainme has quit IRC | 14:11 | |
| cloudnull | morning | 14:14 |
| odyssey4me | javeriak are you deploying from a tag, or branch? | 14:16 |
| odyssey4me | what was your last commit in the log? | 14:16 |
| javeriak | odyssey4me, HEAD kilo | 14:16 |
| odyssey4me | javeriak and your last commit? | 14:17 |
| javeriak | "Merge "Additional retries for ssh wait check" into kilo" | 14:17 |
| odyssey4me | javeriak are you working with old-ish hardware or something? | 14:18 |
| odyssey4me | maybe highly contended virtual environment? | 14:18 |
| javeriak | by oldish do you mean if the OS is new, then nope, i cleaned up a juno deployment before this on it | 14:19 |
| javeriak | odyssey4me, no the hardware is server grade, not having a problem with of my other machine, just the one, and even with that just the memcached container | 14:19 |
| odyssey4me | if so, you may wish to try setting ssh_delay to a higher value than 5 in user_variables.yml - otherwise you may need to validate that your ssh connection to the servers have 'OpenSSH' in the response | 14:20 |
| *** woodard has quit IRC | 14:20 | |
| odyssey4me | javeriak so you're only seeing the issue with a single container? | 14:20 |
| javeriak | odyssey4me, well yes, it fails out on memcached infra3 | 14:21 |
| javeriak | i dont a problem ssh'ing into it otherwise, ping also works | 14:21 |
| *** cloudtrainme has joined #openstack-ansible | 14:21 | |
| odyssey4me | javeriak that's odd - can you confirm that the container does not have a conflicting ip address with something else? | 14:22 |
| *** woodard has joined #openstack-ansible | 14:22 | |
| javeriak | yep, that was my first thought, no conflicts | 14:22 |
| odyssey4me | mhayden now you see why I was reticent from the start to include an alternatively licensed file (re: https://review.openstack.org/216849) - it's complicated | 14:23 |
| odyssey4me | mhayden would it be possible for us to simply pull the file down within the gate check script instead of carrying it? | 14:23 |
| mhayden | odyssey4me: bleh | 14:23 |
| mhayden | odyssey4me: could i ping openstack legal about it and figure out the best path? | 14:24 |
| odyssey4me | mhayden you could, although perhaps a discussion in #openstack-infra would give you a pointer too | 14:25 |
| odyssey4me | this is not something that openstack is unfamiliar with | 14:25 |
| mhayden | hah okay | 14:25 |
| cloudnull | ++ or rope in van from OpenStack Dev ? | 14:26 |
| odyssey4me | sure, if there is a more particular person that we know about - go there | 14:26 |
| odyssey4me | the point is that this is something the community has dealt with many times over | 14:27 |
| odyssey4me | you could even ML the question :) | 14:27 |
| odyssey4me | I'm sure that'd generate a nice little /. style flame war as a tangent :) | 14:27 |
| * odyssey4me buys popcorn in preparation | 14:28 | |
| odyssey4me | cloudnull did you see my note about gnupgp earlier? | 14:28 |
| odyssey4me | javeriak that is very, very odd - something must be interfering on that host | 14:29 |
| odyssey4me | maybe an iptables leftover rule or something? | 14:29 |
| cloudnull | I did. I fired off a retry last night thinking that gnupg was the issue but went to bed before confirming the issue. Thanks for updating that patch. | 14:29 |
| odyssey4me | maybe you can rekick the host and rebuild all the containers on it? | 14:29 |
| odyssey4me | cloudnull yeah, I tested a few things to confirm that the issue was the protocol/port with that host - I did confirm that the ubuntu and the normal secondary we use both serve the key, so we're all good | 14:30 |
| javeriak | odyssey4me, thats possible, but wouldn't I see an issue in the setup-hosts playbook, that runs fine, just re-ran it to confirm and no issues | 14:30 |
| javeriak | yea I'll tear it down if all else fails | 14:31 |
| palendae | odyssey4me: Could have sworn that the FAQ I linked addressed the MIT license in OpenStack projects | 14:31 |
| odyssey4me | javeriak note that the tear down script is best effort - it's not perfect | 14:32 |
| odyssey4me | palendae yes, but mattt raised a good question from the FAQ | 14:33 |
| palendae | Ah | 14:33 |
| mattt | odyssey4me mhayden palendae : in my comment i wasn't suggesting to update our top level license, was just throwing it out there to avoid us getting into a legal battle :) | 14:40 |
| odyssey4me | mattt sure, so we need a definitive direction from someone who knows stuff | 14:41 |
| *** javeriak_ has joined #openstack-ansible | 14:42 | |
| *** javeriak has quit IRC | 14:44 | |
| *** javeriak has joined #openstack-ansible | 14:45 | |
| *** phalmos has joined #openstack-ansible | 14:46 | |
| *** javeriak_ has quit IRC | 14:46 | |
| mhayden | sheesh, mattt :P | 14:48 |
| mhayden | j/k | 14:48 |
| *** phalmos_ has joined #openstack-ansible | 14:49 | |
| *** phalmos has quit IRC | 14:51 | |
| openstackgerrit | Merged stackforge/os-ansible-deployment: Fix keystone shibboleth template to detect ssl usage correctly https://review.openstack.org/220480 | 14:53 |
| *** k_stev has joined #openstack-ansible | 14:53 | |
| palendae | odyssey4me: Yeah, that's a fair statement. None of us are lawyers | 14:54 |
| palendae | Turing should have GPL'ed his whole machine and everything would be easy! | 14:54 |
| openstackgerrit | Merged stackforge/os-ansible-deployment: Remove double register in swift handlers https://review.openstack.org/221237 | 14:54 |
| openstackgerrit | Merged stackforge/os-ansible-deployment: include cinder_all group vars file in cinder-* playbooks https://review.openstack.org/221242 | 14:54 |
| openstackgerrit | Merged stackforge/os-ansible-deployment: Switch Nova/Tempest to use/test Cinder API v2 https://review.openstack.org/220081 | 14:54 |
| *** CheKoLyN has joined #openstack-ansible | 14:54 | |
| mhayden | odyssey4me / mattt: http://lists.openstack.org/pipermail/legal-discuss/2015-September/000397.html | 15:00 |
| *** javeriak_ has joined #openstack-ansible | 15:01 | |
| openstackgerrit | Jesse Pretorius proposed stackforge/os-ansible-deployment: Fix keystone shibboleth template to detect ssl usage correctly https://review.openstack.org/221347 | 15:02 |
| mhayden | odyssey4me / mattt: and there you have it -> http://lists.openstack.org/pipermail/legal-discuss/2015-September/000398.html | 15:02 |
| * mhayden gets to work | 15:02 | |
| *** javeriak_ has quit IRC | 15:03 | |
| palendae | Linderg'd! | 15:03 |
| *** javeriak_ has joined #openstack-ansible | 15:03 | |
| *** javeriak has quit IRC | 15:03 | |
| *** KLevenstein has quit IRC | 15:04 | |
| mhayden | DUN GOT VAN'D | 15:04 |
| odyssey4me | mhayden orsm :) | 15:04 |
| mhayden | orsm? :) | 15:05 |
| * mhayden googled and wished he hadn't | 15:05 | |
| odyssey4me | mhayden orsm = awesome | 15:05 |
| mhayden | oh, gotcha | 15:05 |
| mhayden | :) | 15:05 |
| palendae | You can google Lindberg'd and be safe | 15:05 |
| palendae | Don't google pythong though | 15:05 |
| odyssey4me | lol, now I'm curious | 15:06 |
| *** javeriak_ has quit IRC | 15:06 | |
| openstackgerrit | Major Hayden proposed stackforge/os-ansible-deployment: Add profiling for Ansible tasks https://review.openstack.org/216849 | 15:09 |
| mhayden | odyssey4me: bumped with license fun ^^ | 15:10 |
| *** KLevenstein has joined #openstack-ansible | 15:23 | |
| *** skamithi13 has quit IRC | 15:24 | |
| *** skamithi14 has joined #openstack-ansible | 15:24 | |
| openstackgerrit | Jean-Philippe Evrard proposed stackforge/os-ansible-deployment: Changed certificate order for pem generation with CA files https://review.openstack.org/221361 | 15:24 |
| evrardjp | this ^is important to me, because I introduced a regression in certificate generation for haproxy | 15:25 |
| evrardjp | it didn't appear in my previous testing | 15:25 |
| *** cloudtrainme has quit IRC | 15:26 | |
| *** cloudtrainme has joined #openstack-ansible | 15:28 | |
| evrardjp | I've found another bug, but I'm maybe the only to have that: does someone have haproxy_ssl: True with nova_spice_html5proxy_base_proto: https? | 15:36 |
| evrardjp | if someone has, there are chances that you have protocol mismatch issues, but also haproxy misconfiguration... so I don't know if I should fix haproxy misconfiguration | 15:36 |
| *** devlaps has joined #openstack-ansible | 15:43 | |
| odyssey4me | evrardjp the spice/haproxy mismatch sounds legit - only the keystone config currently adapts the backend haproxy config if keystone is serving via https | 15:43 |
| evrardjp | I'll solve it then | 15:43 |
| odyssey4me | evrardjp as I recall with novnc there was a setting in nova.conf which was used for publishing the URL for the console - but not used for anything else... maybe there's an easier override for spice? | 15:44 |
| odyssey4me | ie spice is still served via http but https is done at haproxy | 15:45 |
| *** KLevenstein is now known as KLev-dentist | 15:45 | |
| evrardjp | there is a variable | 15:46 |
| evrardjp | there are a few variables that changes what's shown on novnc | 15:46 |
| evrardjp | I'll post a fix, you'll see | 15:46 |
| evrardjp | :) | 15:46 |
| openstackgerrit | Jesse Pretorius proposed stackforge/os-ansible-deployment: Update tempest configuration https://review.openstack.org/220559 | 15:48 |
| *** alop has joined #openstack-ansible | 15:48 | |
| openstackgerrit | Jesse Pretorius proposed stackforge/os-ansible-deployment: Add galera_address to all playbooks installing galera_client https://review.openstack.org/221301 | 15:49 |
| *** kerwin_bai has quit IRC | 15:52 | |
| evrardjp | is there a way to define a part of a dict conditionnally (in variables)? Or should define it the variable in my dict anyway and change the checks from is defined to is true in the usage part (like a template)? | 15:55 |
| odyssey4me | evrardjp I'm not aware of a way of conditionally including a piece of dict. | 16:01 |
| evrardjp | ok thanks | 16:01 |
| evrardjp | that's what I thought too | 16:01 |
| odyssey4me | except if the dict is something like the backend haproxy config | 16:01 |
| odyssey4me | like what's done here: https://github.com/stackforge/os-ansible-deployment/blob/master/playbooks/haproxy-install.yml#L46 | 16:02 |
| evrardjp | because the check is already done in the template, that's what you mean, right? | 16:02 |
| odyssey4me | the value for the backend options happens to be a dict | 16:02 |
| odyssey4me | so it includes either one value or the other based on a condition | 16:02 |
| evrardjp | yeah, it's not really what I meant ;) | 16:02 |
| odyssey4me | ok, just thought I'd throw it out there | 16:03 |
| evrardjp | in your example I meant defining or not the variable haproxy_backend_options | 16:03 |
| odyssey4me | it was a hack | 16:03 |
| *** shoutm has joined #openstack-ansible | 16:03 | |
| evrardjp | it's a good hack I sometimes use ;) | 16:03 |
| *** gparaskevas has quit IRC | 16:03 | |
| odyssey4me | yeah, I haven't found a way to do that - you can't really append to a dict without using a module | 16:04 |
| openstackgerrit | Jean-Philippe Evrard proposed stackforge/os-ansible-deployment: Fix of haproxy ssl misconfiguration with nova_console https://review.openstack.org/221386 | 16:04 |
| odyssey4me | sorry, append's the wrong word for a dict I guess - perhaps the right word for a list | 16:04 |
| openstackgerrit | Jean-Philippe Evrard proposed stackforge/os-ansible-deployment: Fix of haproxy ssl misconfiguration with nova_console https://review.openstack.org/221386 | 16:04 |
| evrardjp | I'm learning, so I thought it was maybe possible with some power user skillz ! | 16:05 |
| evrardjp | but it isn't | 16:05 |
| evrardjp | there are workarounds in any case | 16:06 |
| *** sdake_ has joined #openstack-ansible | 16:06 | |
| *** jwagner_away is now known as jwagner | 16:07 | |
| *** phalmos_ has quit IRC | 16:08 | |
| cloudnull | sorru for the late meeting ping for bug triage today: cloudnull, mattt, andymccr, d34dh0r53, hughsaunders, b3rnard0, palendae, Sam-I-Am, odyssey4me, serverascode, rromans, mancdaz, dolphm, _shaps_, BjoernT, claco, echiu, dstanek, jwagner, ayoung, prometheanfire, evrardjp, arbrandes, mhayden | 16:09 |
| d34dh0r53 | o\ | 16:09 |
| prometheanfire | hi | 16:09 |
| hughsaunders | hey | 16:09 |
| evrardjp | hi | 16:09 |
| cloudnull | hey hey | 16:10 |
| cloudnull | :) | 16:10 |
| Sam-I-Am | moo | 16:10 |
| dstanek | hey | 16:10 |
| palendae | Here-ish | 16:10 |
| cloudnull | its like a monday, yesterday was one of the rare US bank holidays :) | 16:10 |
| *** waded_ has left #openstack-ansible | 16:10 | |
| prometheanfire | that what we are calling it? | 16:11 |
| Sam-I-Am | it was not-computer day | 16:11 |
| Sam-I-Am | actually not-computer weekend | 16:11 |
| odyssey4me | o/ | 16:11 |
| hughsaunders | \o/ bank holiday | 16:12 |
| cloudnull | (╯°□°)╯︵ ┻━┻>>bank holiday | 16:12 |
| cloudnull | ok https://bugs.launchpad.net/openstack-ansible/+bug/1493411 | 16:12 |
| openstack | Launchpad bug 1493411 in openstack-ansible " Switch Nova/Tempest to use/test Cinder API v2" [Undecided,New] | 16:12 |
| *** cloudtrainme has quit IRC | 16:13 | |
| cloudnull | doc bug that needs to be sorted. | 16:13 |
| *** shoutm has quit IRC | 16:13 | |
| cloudnull | issue origination https://review.openstack.org/#/c/214045/ | 16:13 |
| cloudnull | Sam-I-Am: you have bandwith to try and bang that out? --cc odyssey4me | 16:14 |
| Sam-I-Am | nope | 16:14 |
| Sam-I-Am | that would be more of an rpc release notes thing | 16:14 |
| odyssey4me | cloudnull the standard for docs bugs at this stage is to set them to low and add the rpcdocs group | 16:14 |
| Sam-I-Am | i dont think we have a place in upstream osad for things like that | 16:15 |
| odyssey4me | ie confirmed, low, rpcdocs | 16:15 |
| odyssey4me | they then get picked up whenever someone has time, and that list is also given to the osic team to pick up | 16:15 |
| cloudnull | we should touch on that, maybe on the thursday meeting . as it seems that our "doc" related issues are growing. | 16:16 |
| odyssey4me | agreed | 16:16 |
| Sam-I-Am | cloudnull: there's no agreement on what in-tree docs should provide | 16:16 |
| cloudnull | ok Ill get to adding that to the agenda | 16:16 |
| cloudnull | next https://bugs.launchpad.net/openstack-ansible/+bug/1492080 | 16:17 |
| openstack | Launchpad bug 1492080 in openstack-ansible "radvd not installed in neutron_agents_container" [Undecided,New] | 16:17 |
| Sam-I-Am | do we support v6? | 16:17 |
| Sam-I-Am | i dont recall testing it | 16:17 |
| prometheanfire | iirc it is a req of neutron | 16:17 |
| prometheanfire | even if we don't support it | 16:17 |
| prometheanfire | or used to be | 16:18 |
| cloudnull | evrardjp: mrstanwell: rackertom: getting that into liberty-x for "official" ipv6 support would be interesting. | 16:18 |
| evrardjp | it should be supported, v6 is the future | 16:18 |
| cloudnull | ^ ++ | 16:18 |
| evrardjp | I'll work on it anyway | 16:18 |
| odyssey4me | ++ | 16:18 |
| mrstanwell | technically, v6 is the past... | 16:18 |
| Sam-I-Am | someday we need to test v6 | 16:18 |
| hughsaunders | v6 has been the future for a while | 16:18 |
| mrstanwell | we just haven't got there yet. | 16:18 |
| cloudnull | hahaha | 16:18 |
| evrardjp | mrstanwell: true | 16:18 |
| Sam-I-Am | patches welcome :) | 16:18 |
| evrardjp | I don't see a problem in anycase to introduce radvd | 16:19 |
| cloudnull | I'd like to ipv6 support in the near future. its been mostly functional since kilo so i think its time we take a look at it espially with interested parties involved in making it go. | 16:19 |
| prometheanfire | looks like neutron needs it | 16:19 |
| prometheanfire | https://blueprints.launchpad.net/neutron/+spec/neutron-ipv6-radvd-ra https://blueprints.launchpad.net/neutron/+spec/ipv6-two-attributes | 16:19 |
| evrardjp | cloudnull: it's really not that simple for tenants | 16:19 |
| odyssey4me | my only concern would be whether radvd needs to run as a service, or whether it's started within the customer networks and managed that way | 16:19 |
| mrstanwell | I installed radvd by hand in the container, and slaac started working. that's low-hanging fruit. | 16:19 |
| odyssey4me | I haven't looked into the implementation requirements, but I'd be concerned with data leaks between tenants. | 16:20 |
| cloudnull | *especially | 16:20 |
| mrstanwell | I haven't gotten back to it yet to see what other evils lurk, however. | 16:20 |
| Sam-I-Am | it probably does what it does with every other service... launch the daemon with a generated config file somewhere in /var/lib/neutron | 16:20 |
| odyssey4me | Sam-I-Am I would hope so | 16:20 |
| prometheanfire | odyssey4me: flat network problems, not really ipv6 specific | 16:20 |
| mrstanwell | sam: yes, that's what it does | 16:20 |
| evrardjp | I've got tenant IPv6 working without installing rdavd IIRC | 16:20 |
| evrardjp | that's why I asked if it was really necessary | 16:21 |
| prometheanfire | I have v6 working with external radvd | 16:21 |
| evrardjp | same | 16:21 |
| evrardjp | so I don't see a problem to introduce radvd, but not as a mandatory item | 16:21 |
| cloudnull | prometheanfire can you join the party on that issue? | 16:21 |
| prometheanfire | sure | 16:21 |
| Sam-I-Am | if we're going to support v6, we can't be choosy on how people do it | 16:22 |
| mrstanwell | neutron tries to start it, whether it's there or not... | 16:22 |
| odyssey4me | it'd be nice to see this come in with liberty | 16:22 |
| cloudnull | ++ | 16:22 |
| prometheanfire | wonder what'd happen if it's already running as a service... | 16:22 |
| odyssey4me | anyone volunteer to pick it up? | 16:22 |
| Sam-I-Am | prometheanfire: probably nothing as long as the config files dont conflict | 16:22 |
| evrardjp | prometheanfire: or running somewhere else... it needs to be carefully studied... I can't do it right now, sorry | 16:23 |
| prometheanfire | ya, also, could be listening on other interfaces | 16:23 |
| Sam-I-Am | however, two radvd services on the same network... nope. | 16:23 |
| Sam-I-Am | unless they're advertising the same thing | 16:23 |
| evrardjp | it will work, but it sounds bad | 16:23 |
| prometheanfire | indeed | 16:23 |
| prometheanfire | that's a general ops problem though, we can't control outside resources | 16:23 |
| evrardjp | maybe there should be spec for it | 16:24 |
| evrardjp | explaining what we see as a possible issue | 16:24 |
| odyssey4me | I'd agree to an overall ipv6 spec. | 16:24 |
| prometheanfire | ya, that could be useful | 16:25 |
| odyssey4me | But if Neutron is trying to start this and fails (because it's not there), then this is a bug. | 16:25 |
| prometheanfire | would be nice if tempest did v6 testing | 16:25 |
| Sam-I-Am | i havent seen neutron fail without this | 16:25 |
| evrardjp | Sam-I-Am: +1 | 16:25 |
| prometheanfire | either have I | 16:25 |
| odyssey4me | I'd suggest doing an initial simple solution for the bug - then doing the more careful inspection as part of the spec. | 16:25 |
| Sam-I-Am | it might with v6 things configured and no radvd | 16:25 |
| mrstanwell | Neutron definitely tries to start it. That's what tipped me off. It was in the sudo log. | 16:25 |
| prometheanfire | and I set v6 addresses in neutron | 16:25 |
| Sam-I-Am | but as-is, osad does not gail here. | 16:25 |
| Sam-I-Am | fail | 16:25 |
| cloudnull | would somone want to draft the initial spec ? | 16:26 |
| evrardjp | odyssey4me: I agree | 16:26 |
| evrardjp | maybe mrstanwell, because he saw the bug? :D | 16:26 |
| prometheanfire | sure, guess I can do it | 16:26 |
| mrstanwell | ;-) I'm happy to help. Not terribly experienced with either osad or ostack.... | 16:26 |
| prometheanfire | need to start working on the standalone swift testing spec sometime though | 16:26 |
| odyssey4me | prometheanfire you may wish to review https://review.openstack.org/221009 as it's a dependency for that | 16:27 |
| cloudnull | prometheanfire | mrstanwell: you guys have some cycles to put into a quick spec? would be nice to have some different oppinions on it so that we get the best solution going forward . | 16:27 |
| odyssey4me | +1 | 16:27 |
| prometheanfire | cloudnull: ya, many people on one lab makes time | 16:28 |
| *** phalmos has joined #openstack-ansible | 16:28 | |
| prometheanfire | odyssey4me: if that's only dependant then fine, but that's a distraction from v6 atm :P | 16:28 |
| cloudnull | mrstanwell: experience in osad or ostack is not required :) | 16:28 |
| cloudnull | but we can help in the spec draft | 16:29 |
| cloudnull | ok marked issue as wishlist / confirmed. | 16:29 |
| prometheanfire | ya, added to todo list | 16:29 |
| mrstanwell | not clear on the scope of the spec. is this new support, really? | 16:29 |
| odyssey4me | mrstanwell the spec would be for ipv6 support end to end | 16:30 |
| *** mnestheu1 has joined #openstack-ansible | 16:30 | |
| prometheanfire | not really, more of a verification and definition of what is supported | 16:30 |
| odyssey4me | and you can define the scope in the spec | 16:30 |
| prometheanfire | odyssey4me: openstack services listening on v6? | 16:30 |
| odyssey4me | prometheanfire heh, no idea - scope it in the spec according to what you think is practical and useful | 16:30 |
| *** mnestheu1 is now known as scarlisle | 16:31 | |
| prometheanfire | ya, gonna start with guests, since that's easy :P | 16:31 |
| cloudnull | prometheanfire: i'd say no. I'd start with tenant support for v6 and go from there. | 16:31 |
| evrardjp | I think 2 specs (or more) would be great: One with the analysis of radvd and another to complete with other ipv6 analysis | 16:31 |
| odyssey4me | ++ cloudnull | 16:31 |
| evrardjp | cloudnull: ++ | 16:31 |
| prometheanfire | would need ipv6 ansible inventory, probably | 16:31 |
| cloudnull | evrardjp: that'd be great IMO if we have cycles to get that done. | 16:31 |
| mrstanwell | tenant support is one thing. ostack service v6 support in osad is a whole 'nother ballgame. | 16:31 |
| prometheanfire | anyway, next? | 16:31 |
| prometheanfire | mrstanwell: exactly | 16:32 |
| miguelgrinberg | odyssey4me: hey, happy to look at that heat problem. I'll let you know what I find. | 16:32 |
| odyssey4me | thanks miguelgrinberg | 16:32 |
| *** javeriak has joined #openstack-ansible | 16:32 | |
| cloudnull | cool! prometheanfire evrardjp mrstanwell | typie typie make the v6 go :) | 16:32 |
| prometheanfire | after lunch kthnxbai | 16:33 |
| mrstanwell | hey, cut us some slaac. }-) | 16:33 |
| cloudnull | next https://bugs.launchpad.net/openstack-ansible/+bug/1490591 | 16:33 |
| openstack | Launchpad bug 1490591 in openstack-ansible trunk "Update 'Add a compute host' operations guide section for kIlo" [Low,New] | 16:33 |
| evrardjp | we'll have to define a timeline to see if that meets my time availability ;) | 16:33 |
| cloudnull | mrstanwell: lol | 16:33 |
| evrardjp | mrstanwell: :D | 16:33 |
| cloudnull | doc issue. palendae you have some time to fixerate that ? | 16:34 |
| palendae | cloudnull: Maybe, still poking upgrade issues | 16:34 |
| Sam-I-Am | cloudnull: this probably also applies to rpc | 16:35 |
| Sam-I-Am | too many things to track | 16:35 |
| palendae | Sam-I-Am: It does | 16:35 |
| Sam-I-Am | DocImpact is for osad bugs, not for rpc bugs | 16:36 |
| palendae | Sam-I-Am: rpc is rpc's concern | 16:36 |
| palendae | This is upstream triage | 16:36 |
| Sam-I-Am | so just tagging something DocImpact does not necessarily mean it will also be analyzed for rpc | 16:36 |
| Sam-I-Am | palendae: yeah, i know. but this is extremely hard to follow. | 16:36 |
| palendae | Sam-I-Am: I understand. Just saying it's not necessarily an openstack-ansible problem to fix RPC's docs | 16:36 |
| palendae | I realize having our feet in both camps is hard | 16:37 |
| odyssey4me | all the openstack-ansible docs need an overhaul | 16:37 |
| cloudnull | more on that thursday :) | 16:37 |
| odyssey4me | I'd suggest that we look at it as a 'feature' as we do work on the upgrade mechanism for liberty | 16:37 |
| cloudnull | thats a good idea | 16:38 |
| odyssey4me | three tracks - one for the configs, one for the docs and one for the upgrade mechanism | 16:38 |
| palendae | Sure...right now the problem is bodies | 16:38 |
| palendae | Only Sam-I-Am and I bother with actually committing docs | 16:38 |
| palendae | And we're both tied up | 16:38 |
| cloudnull | we just need more hours in the day . | 16:38 |
| Sam-I-Am | can we script that? | 16:39 |
| cloudnull | ok triaged confirmed low, added to list for meeting thrusday | 16:39 |
| cloudnull | next: https://bugs.launchpad.net/openstack-ansible/+bug/1489451 | 16:39 |
| openstack | Launchpad bug 1489451 in openstack-ansible "Not able to connect memcached_container while running setup-infrastructure.yml " [Undecided,New] | 16:39 |
| odyssey4me | palendae evrardjp and I have too, but yes we could do with more people paying attention to working on docs more consistently | 16:39 |
| Sam-I-Am | downloadmorehours.com ? | 16:39 |
| palendae | Anyway, if I get the time, yes, I can tackle that. If someone gets to it before me, even better | 16:39 |
| odyssey4me | this one was raised earlier | 16:39 |
| odyssey4me | it seemed to have been one container on one host in the infra | 16:40 |
| *** Guest60469 is now known as mgagne | 16:40 | |
| *** mgagne has joined #openstack-ansible | 16:40 | |
| odyssey4me | but it may relate to the forks value when using real hardware | 16:40 |
| odyssey4me | ie https://bugs.launchpad.net/openstack-ansible/+bug/1479812 | 16:40 |
| openstack | Launchpad bug 1479812 in openstack-ansible trunk "ssh connection failures during deployments" [Low,New] - Assigned to Jesse Pretorius (jesse-pretorius) | 16:40 |
| cloudnull | this issue seems like an issue with the image on the disk and ties to a larger effort to remove the requirement to download a cached image in favor of creating an image within the environment which can be updated and managed strictly through osad.-cc mhayden odyssey4me andymccr. | 16:40 |
| odyssey4me | there's been discussion there and there's an outstanding query of bgmccollum_ | 16:40 |
| odyssey4me | that may be, and yes given a little time andymccr, mhayden and I have discussed perhaps including an image build process into openstack-ansible instead of using the download mechanism | 16:42 |
| palendae | What kind of image? | 16:43 |
| palendae | Containers? | 16:43 |
| odyssey4me | initially just the base lxc image we're downloading | 16:43 |
| palendae | Would that interfere with the ability to split out all services into their own physical nodes (if a deployer wanted)? | 16:43 |
| odyssey4me | but perhaps expanding that to a broader process, which would need a blueprint/spec | 16:43 |
| palendae | I seem to remember that being a thing we wanted to keep compatiblke | 16:43 |
| odyssey4me | palendae it should not - it's an implementation detail handled in the playbooks currently | 16:44 |
| odyssey4me | it's only used in the container create role | 16:44 |
| palendae | Ok | 16:44 |
| palendae | And the environment, since you'd have to modify that to get the bare metal properties...but continue :) | 16:44 |
| odyssey4me | well, the project should not try to cater for every possible install method with the same playbooks IMO - but that's another discussion | 16:45 |
| evrardjp | cloudnull: if you are talking of template distribution/caching improvement process, you can count me in. In terms of schedule, starting with the next summit | 16:45 |
| cloudnull | nice ! | 16:45 |
| odyssey4me | evrardjp nice - I think that perhaps this is a worthy topic for discussion at the summit | 16:46 |
| evrardjp | this needs to be solved, at least for security awareness | 16:46 |
| odyssey4me | we can work on it during the Mitaka timeframe | 16:46 |
| cloudnull | yes i think thats a major improvement we should be able to tackle in the M cycle. | 16:46 |
| cloudnull | and potentially backport to Liberty if at all possible. but that may be too far forward looking. | 16:47 |
| palendae | Liberty needs an upgrade path too | 16:47 |
| palendae | An implementation of one | 16:47 |
| odyssey4me | yeah, we make the backport call once we have it working | 16:47 |
| cloudnull | ++ | 16:48 |
| odyssey4me | palendae agreed, your spec needs reviews | 16:48 |
| palendae | odyssey4me: It does and I need to finish it | 16:48 |
| odyssey4me | +1 | 16:48 |
| palendae | I lost some work because of the latest patchset conflict | 16:48 |
| palendae | Biggest thing I think I lost was the ML discussion links for the epoch things | 16:49 |
| cloudnull | last issue on deck https://bugs.launchpad.net/openstack-ansible/+bug/1489406 | 16:49 |
| openstack | Launchpad bug 1489406 in openstack-ansible "Horizon configuration to support Keystone V3 multiple domains" [Undecided,New] | 16:49 |
| cloudnull | is Cristian Calin (cristi-calin) here ? | 16:50 |
| odyssey4me | there's a patch included there, which I asked the author to try and submit for review | 16:50 |
| cloudnull | ^ that | 16:50 |
| openstackgerrit | Jean-Philippe Evrard proposed stackforge/os-ansible-deployment: Updated the configuration to current dynamic inventory standards https://review.openstack.org/221402 | 16:50 |
| odyssey4me | unfortunately the author went offline and I have not seen him on since | 16:50 |
| cloudnull | if not we can make it go, but would not want to snake the commit from them | 16:51 |
| odyssey4me | so I think one of us needs to just submit on his behalf | 16:51 |
| odyssey4me | just have him as the author | 16:51 |
| odyssey4me | I'm happy to do that if no-one else picks it up | 16:52 |
| odyssey4me | it's pretty key for Liberty as Keystone v2 is most likely to be deprecated for Liberty | 16:53 |
| openstackgerrit | Jean-Philippe Evrard proposed stackforge/os-ansible-deployment-specs: Add spec for major update of haproxy role https://review.openstack.org/214089 | 16:53 |
| cloudnull | I responded to the issue we'll see what happens and if nothing we'll pickup the changeset | 16:53 |
| cloudnull | ok so thats it | 16:54 |
| cloudnull | anything anyone want to raise ? | 16:54 |
| evrardjp | none | 16:55 |
| evrardjp | keystone v2.0 will be deprecated in Liberty, really? | 16:56 |
| cloudnull | thanks everyone ! | 16:56 |
| *** woodard has quit IRC | 16:56 | |
| evrardjp | thanks to you too | 16:56 |
| cloudnull | dolphm: lbragstad: -cc ^^ | 16:56 |
| odyssey4me | evrardjp assuming they reach their target, yes | 16:56 |
| evrardjp | interesting | 16:56 |
| * cloudnull being cross functional | 16:56 | |
| cloudnull | :D | 16:56 |
| dolphm | evrardjp: the v2 client (especially the CLI) is facing imminent deprecation, but the API is not quite there yet | 16:57 |
| odyssey4me | KEystone v3 has been out for several cycles | 16:57 |
| evrardjp | anyway I'll need to test the federation stuff with multiple IdP and a DS later... so I guess I'll be for v3 too | 16:57 |
| *** woodard has joined #openstack-ansible | 16:57 | |
| evrardjp | odyssey4me: indeed, but not in the clients and some of the consumers of keystone | 16:57 |
| odyssey4me | evrardjp yeah, there has been some effort in gate checks and the ML to change that | 16:58 |
| prometheanfire | mentioning this I guess https://etherpad.openstack.org/p/openstack-ansible-mitaka-summit | 16:58 |
| prometheanfire | not that etherpad is working | 16:58 |
| odyssey4me | the encouragement is generally for the projects to support the use of the openstack client instead, but that's taking some time | 16:58 |
| odyssey4me | prometheanfire yeah I noticed that earlier and thought it might be transitional as -infra is doing aquite a bit of work on etherpad at the moment | 16:59 |
| prometheanfire | neat | 16:59 |
| evrardjp | we should put the names of those who have written the text.... | 16:59 |
| evrardjp | I didn't write anything! | 17:00 |
| prometheanfire | ? | 17:00 |
| prometheanfire | I have my name in the little dropdown | 17:00 |
| evrardjp | yes, but it's far harder to know what you changed | 17:01 |
| odyssey4me | me too | 17:01 |
| evrardjp | with only colors | 17:01 |
| * prometheanfire shrugs | 17:01 | |
| odyssey4me | yeah, etherpad sucks quite a bit actually | 17:01 |
| *** skamithi12 has joined #openstack-ansible | 17:02 | |
| *** harlowja has joined #openstack-ansible | 17:03 | |
| *** javeriak has quit IRC | 17:03 | |
| evrardjp | I'm off for today, see you tomorrow! | 17:07 |
| *** woodard has quit IRC | 17:10 | |
| *** jwagner is now known as jwagner_away | 17:16 | |
| *** jwagner_away is now known as jwagner | 17:17 | |
| *** jwagner is now known as jwagner_away | 17:18 | |
| *** sdake_ has quit IRC | 17:19 | |
| *** jwagner_away is now known as jwagner | 17:19 | |
| *** woodard has joined #openstack-ansible | 17:21 | |
| *** skamithi12 has quit IRC | 17:22 | |
| *** abitha has joined #openstack-ansible | 17:30 | |
| *** abitha has quit IRC | 17:31 | |
| *** abitha has joined #openstack-ansible | 17:31 | |
| *** KLev-dentist is now known as KLevenstein | 17:41 | |
| *** sdake has joined #openstack-ansible | 17:44 | |
| mhayden | odyssey4me: sorry i missed the discussion above -- here's what i got from some quick benchmarking -> https://bugs.launchpad.net/openstack-ansible/+bug/1489169 | 17:46 |
| openstack | Launchpad bug 1489169 in openstack-ansible "Speed up deployment by upgrading cached container image" [Wishlist,Confirmed] - Assigned to Major Hayden (rackerhacker) | 17:46 |
| *** CheKoLyN has quit IRC | 17:51 | |
| *** javeriak has joined #openstack-ansible | 17:57 | |
| *** cloudtrainme has joined #openstack-ansible | 18:00 | |
| *** cloudtrainme has quit IRC | 18:01 | |
| *** pabelanger has joined #openstack-ansible | 18:03 | |
| pabelanger | ohai | 18:03 |
| *** javeriak has quit IRC | 18:04 | |
| *** javeriak has joined #openstack-ansible | 18:05 | |
| *** CheKoLyN has joined #openstack-ansible | 18:05 | |
| *** CheKoLyN has quit IRC | 18:22 | |
| cloudnull | o/ pabelanger | 18:23 |
| pabelanger | cloudnull: good to read you have a plan for ansible roles under big tent. Looking forward to this | 18:25 |
| cloudnull | its a wip so far but its getting there | 18:25 |
| *** sdake_ has joined #openstack-ansible | 18:25 | |
| pabelanger | For the most part, the roles I am working on are -infra related. Nodepool, zuul, jenkins, etc. But as long as we have a general ansible team, I don't see a problem | 18:25 |
| cloudnull | thats awesome! | 18:27 |
| cloudnull | if we can help any please let us know. | 18:28 |
| *** sdake has quit IRC | 18:29 | |
| *** openstackgerrit has quit IRC | 18:31 | |
| *** openstackgerrit has joined #openstack-ansible | 18:31 | |
| *** cloudtrainme has joined #openstack-ansible | 18:34 | |
| *** CheKoLyN has joined #openstack-ansible | 18:39 | |
| *** sdake_ is now known as sdake | 18:40 | |
| *** javeriak_ has joined #openstack-ansible | 18:40 | |
| *** cloudtrainme has quit IRC | 18:40 | |
| *** cloudtrainme has joined #openstack-ansible | 18:41 | |
| *** javeriak has quit IRC | 18:43 | |
| *** javeriak has joined #openstack-ansible | 18:46 | |
| *** sdake_ has joined #openstack-ansible | 18:46 | |
| *** javeriak_ has quit IRC | 18:46 | |
| *** sdake has quit IRC | 18:49 | |
| *** abitha has quit IRC | 18:56 | |
| *** KLevenstein_ has joined #openstack-ansible | 18:59 | |
| *** KLevenstein has quit IRC | 18:59 | |
| *** KLevenstein_ is now known as KLevenstein | 18:59 | |
| *** abitha has joined #openstack-ansible | 19:01 | |
| *** cloudtrainme has quit IRC | 19:02 | |
| *** cloudtrainme has joined #openstack-ansible | 19:03 | |
| *** cloudtrainme has quit IRC | 19:04 | |
| *** pabelanger has quit IRC | 19:06 | |
| *** cloudtrainme has joined #openstack-ansible | 19:15 | |
| scarlisle | Greetings all. Do the playbooks or the lxc-template setting file descriptor limits for containers? | 19:15 |
| scarlisle | er | 19:16 |
| scarlisle | eh, you know what I mean | 19:16 |
| cloudnull | scarlisle: you want to set the ulimit within the container ? | 19:22 |
| *** Mudpuppy_ has joined #openstack-ansible | 19:22 | |
| cloudnull | or are you looking at modifying the inotify limits on the container host ? | 19:23 |
| cloudnull | or all of the above ? | 19:23 |
| scarlisle | cloudnull trying to figure this out because even though the playbooks are setting rabbit's file descriptor limit at 4096, its only getting 3996 | 19:23 |
| cloudnull | Juno? Kilo? master? -- version? | 19:26 |
| * cloudnull Assuming Juno | 19:26 | |
| cloudnull | https://github.com/stackforge/os-ansible-deployment/blob/juno/rpc_deployment/roles/rabbit_common/templates/rabbitmq-server should set it to 4096 | 19:26 |
| *** Mudpuppy has quit IRC | 19:26 | |
| *** arbrandes has joined #openstack-ansible | 19:26 | |
| *** Mudpuppy_ has quit IRC | 19:27 | |
| scarlisle | yep, Juno | 19:28 |
| scarlisle | noticing it in 10.1.6 and 10.1.11 at least | 19:29 |
| *** bgmccollum_ is now known as bgmccollum | 19:29 | |
| scarlisle | and yeah, that file is there | 19:29 |
| scarlisle | well, present inside the rabbit containers | 19:29 |
| *** Mudpuppy has joined #openstack-ansible | 19:29 | |
| *** javeriak has quit IRC | 19:30 | |
| *** k_stev has quit IRC | 19:31 | |
| *** k_stev has joined #openstack-ansible | 19:35 | |
| *** arbrandes1 has joined #openstack-ansible | 19:35 | |
| *** arbrandes has quit IRC | 19:36 | |
| *** k_stev has quit IRC | 19:39 | |
| cloudnull | scarlisle: how are you determining that its set to 3996 in the rabbit contianer? | 19:47 |
| * cloudnull trying to recreate | 19:47 | |
| *** sdake_ is now known as sdake | 19:49 | |
| scarlisle | cloudnull rabbitmqctl status | grep -A5 file_descriptors | 19:51 |
| scarlisle | root@<redacted>:~# rabbitmqctl status | grep -A5 file_descriptor | 19:52 |
| scarlisle | {file_descriptors, | 19:52 |
| scarlisle | [{total_limit,3996}, | 19:52 |
| scarlisle | {total_used,251}, | 19:52 |
| scarlisle | {sockets_limit,3594}, | 19:52 |
| scarlisle | {sockets_used,249}]}, | 19:52 |
| scarlisle | {processes,[{limit,1048576},{used,4605}]}, | 19:52 |
| *** Mudpuppy has quit IRC | 19:56 | |
| *** jmckind has joined #openstack-ansible | 19:57 | |
| stevelle | this sounds like something I might have fixed for kilo/master | 19:58 |
| *** Mudpuppy has joined #openstack-ansible | 19:59 | |
| cloudnull | scarlisle: it looks like based on the setting there are some reserved file descriptors on process start up. | 20:02 |
| cloudnull | IE i set it to 102400 and the running value is 102300 | 20:02 |
| mattt | mhayden: nice, that was straight-forward enough :) | 20:02 |
| cloudnull | scarlisle: and setting somthing off like 4224 results in the same 3996 running value | 20:03 |
| scarlisle | interesting | 20:07 |
| scarlisle | so if we need to bump an environment beyond 4096 we have to bump it much higher in order to get something close to 4096? Or am I misunderstanding? | 20:08 |
| cloudnull | yes its seems to go in steps. | 20:09 |
| cloudnull | Depending on the host, I'd assume that you can set it to 8196 which should be more than enough. | 20:11 |
| cloudnull | and in my basic testing that seems to work | 20:11 |
| cloudnull | scarlisle: the resultant of setting it to 8196 is 7284 | 20:19 |
| scarlisle | weird | 20:31 |
| scarlisle | I just tried it | 20:31 |
| scarlisle | set ulimit to 8196 | 20:31 |
| scarlisle | and now my rabbit is at 8096 | 20:31 |
| *** Bjoern_ has joined #openstack-ansible | 20:31 | |
| *** Bjoern_ is now known as BjoernT | 20:31 | |
| *** k_stev has joined #openstack-ansible | 20:31 | |
| scarlisle | was hoping there was a correlation | 20:31 |
| scarlisle | cloudnull what do you get when you run cat /proc/sys/fs/file-nr? | 20:32 |
| cloudnull | scarlisle: http://paste.openstack.org/show/450752/ | 20:34 |
| cloudnull | the value on the host is inheritted within the container | 20:34 |
| *** k_stev has quit IRC | 20:36 | |
| scarlisle | yep | 20:38 |
| scarlisle | so the value in the left column is basically spread across the host as well as its containers, right? | 20:40 |
| scarlisle | shared, rather | 20:40 |
| cloudnull | yes | 20:41 |
| *** tlian2 has joined #openstack-ansible | 20:55 | |
| *** tlian has quit IRC | 20:57 | |
| *** woodard has quit IRC | 21:06 | |
| *** sdake_ has joined #openstack-ansible | 21:16 | |
| *** sdake has quit IRC | 21:20 | |
| *** arbrandes1 has quit IRC | 21:28 | |
| *** cloudtrainme has quit IRC | 21:36 | |
| *** devlaps has quit IRC | 21:38 | |
| *** cloudtrainme has joined #openstack-ansible | 21:43 | |
| scarlisle | so we have to manually change that 4096 | 21:44 |
| scarlisle | ran a python child bomb against rabbit with an fd of 15000 | 21:44 |
| scarlisle | things scaled rather nicely, actually | 21:44 |
| *** cloudtrainme has quit IRC | 21:46 | |
| *** darrenc_ is now known as darrenc | 21:47 | |
| openstackgerrit | Matthew Thode proposed stackforge/os-ansible-deployment-specs: Add spec for IPv6 support for tenants https://review.openstack.org/221516 | 21:51 |
| prometheanfire | mrstanwell: ^ | 21:51 |
| *** cloudtrainme has joined #openstack-ansible | 21:52 | |
| *** sdake_ is now known as sdake | 21:55 | |
| *** cloudtra_ has joined #openstack-ansible | 21:57 | |
| *** jmckind has quit IRC | 22:00 | |
| *** cloudtrainme has quit IRC | 22:01 | |
| *** sdake_ has joined #openstack-ansible | 22:02 | |
| *** sdake has quit IRC | 22:05 | |
| cloudnull | tyvm prometheanfire | 22:10 |
| prometheanfire | cloudnull: ja | 22:10 |
| prometheanfire | not really sure on how to do ipv6 testing, but that's for the future | 22:10 |
| *** jwagner is now known as jwagner_away | 22:11 | |
| *** spotz is now known as spotz_zzz | 22:16 | |
| mrstanwell | prometheanfire: that makes sense, thanks. | 22:17 |
| prometheanfire | :D | 22:17 |
| *** Mudpuppy_ has joined #openstack-ansible | 22:19 | |
| *** Mudpuppy has quit IRC | 22:22 | |
| *** pradk has quit IRC | 22:22 | |
| *** Mudpuppy_ has quit IRC | 22:23 | |
| *** skamithi13 has joined #openstack-ansible | 22:26 | |
| Sam-I-Am | prometheanfire: s/tenants/projects :) | 22:27 |
| prometheanfire | Sam-I-Am: meh | 22:28 |
| openstackgerrit | Matthew Thode proposed stackforge/os-ansible-deployment-specs: Add spec for IPv6 support for tenants https://review.openstack.org/221516 | 22:29 |
| prometheanfire | Sam-I-Am: better? | 22:29 |
| *** sigmavirus24 is now known as sigmavirus24_awa | 22:30 | |
| *** skamithi14 has quit IRC | 22:30 | |
| Sam-I-Am | still says tenants? | 22:30 |
| prometheanfire | where? | 22:31 |
| prometheanfire | the tag will remain I think | 22:31 |
| prometheanfire | also, title | 22:31 |
| prometheanfire | raise an issue and if acked I'll modify | 22:31 |
| Sam-I-Am | so... theres a commit message and a filename | 22:32 |
| Sam-I-Am | i suppose you could change both unless i'm missing something | 22:32 |
| prometheanfire | bah, guess I'll do it | 22:32 |
| prometheanfire | just to make you go away | 22:32 |
| openstackgerrit | Matthew Thode proposed stackforge/os-ansible-deployment-specs: Add spec for IPv6 support for projects https://review.openstack.org/221516 | 22:34 |
| prometheanfire | Sam-I-Am: now leave me alone | 22:34 |
| Sam-I-Am | prometheanfire: no | 22:35 |
| Sam-I-Am | but thanks | 22:35 |
| *** cloudtra_ has quit IRC | 22:45 | |
| *** ggillies has joined #openstack-ansible | 22:48 | |
| *** jwagner_away is now known as jwagner | 22:51 | |
| *** jwagner is now known as jwagner_away | 22:56 | |
| *** CheKoLyN has quit IRC | 23:14 | |
| *** scarlisle has quit IRC | 23:38 | |
| *** KLevenstein has quit IRC | 23:42 | |
| coolj | now come up an equivalent word to multitenancy based on the word project | 23:43 |
| *** phalmos has quit IRC | 23:47 | |
| *** shoutm has joined #openstack-ansible | 23:49 | |
| Sam-I-Am | coolj: multiprojectcy | 23:55 |
| Sam-I-Am | duhh | 23:56 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!