| *** sawblade6 has joined #openstack-ansible | 00:04 | |
| *** sawblade6 has quit IRC | 00:11 | |
| *** dmsimard has joined #openstack-ansible | 00:19 | |
| *** sawblade6 has joined #openstack-ansible | 00:34 | |
| *** sawblade6 has quit IRC | 00:39 | |
| *** sawblade6 has joined #openstack-ansible | 00:39 | |
| *** cshen has joined #openstack-ansible | 00:43 | |
| *** sawblade6 has quit IRC | 00:44 | |
| *** cshen has quit IRC | 00:48 | |
| *** ianychoi has quit IRC | 00:49 | |
| *** ianychoi has joined #openstack-ansible | 00:50 | |
| *** mmercer has quit IRC | 00:55 | |
| openstackgerrit | Merged openstack/openstack-ansible-os_congress stable/queens: Define cache_timeout variable https://review.openstack.org/604093 | 01:13 |
|---|---|---|
| *** vnogin has joined #openstack-ansible | 01:14 | |
| *** sawblade6 has joined #openstack-ansible | 01:15 | |
| *** vnogin has quit IRC | 01:18 | |
| *** sawblade6 has quit IRC | 01:20 | |
| openstackgerrit | Merged openstack/openstack-ansible-os_congress stable/rocky: Define cache_timeout variable https://review.openstack.org/604092 | 01:25 |
| openstackgerrit | Merged openstack/openstack-ansible-os_congress stable/rocky: Remove galera_client meta-dependency https://review.openstack.org/603791 | 01:25 |
| openstackgerrit | Merged openstack/openstack-ansible-os_keystone master: SUSE: Switch distro jobs as voting https://review.openstack.org/604085 | 01:31 |
| openstackgerrit | Merged openstack/openstack-ansible-os_tacker stable/rocky: Clean up role tests https://review.openstack.org/603881 | 01:45 |
| *** spatel has joined #openstack-ansible | 02:02 | |
| spatel | cloudnull: what freenode client are you using ? | 02:02 |
| cloudnull | I use "The Lounge" | 02:03 |
| spatel | I am using OSX and my client sucks | 02:03 |
| openstackgerrit | Merged openstack/openstack-ansible-os_cinder stable/rocky: Clean up role tests https://review.openstack.org/603868 | 02:03 |
| cloudnull | I used to use textual a long time ago (when I was on a mac) | 02:03 |
| spatel | I am using LimeChat | 02:03 |
| cloudnull | textual was good. | 02:03 |
| spatel | let me give it a shot | 02:03 |
| cloudnull | the lounge runs on a VM its web app | 02:04 |
| cloudnull | I have it running on my openstack cloud :) | 02:04 |
| spatel | hmmm!!! so what are the advantage of it? | 02:05 |
| cloudnull | mobile, desktop, tablet, etc. its always connected. | 02:05 |
| * cloudnull doesnt miss messages. | 02:05 | |
| cloudnull | https://thelounge.chat/ | 02:05 |
| spatel | that is what my problem is... | 02:05 |
| cloudnull | https://github.com/cloudnull/Ansible-Playbook-TheLounge | 02:06 |
| spatel | if my client disconnect i lost everything.. | 02:06 |
| cloudnull | textual is good for a local client | 02:06 |
| spatel | i want to stay connected | 02:06 |
| cloudnull | it logs and will pick up where you left off. but doesnt give you something that's always connected | 02:07 |
| spatel | does it give you offline mesg? | 02:07 |
| cloudnull | my setup is as follows: ZNC connecting to my networks, TheLounge connecting to ZNC | 02:07 |
| cloudnull | yes when I go to sleep (I do in fact sleep) the ZNC setup continues to recieve messages | 02:08 |
| spatel | then i am going to use that | 02:08 |
| cloudnull | you can do a lot of this without thelounge | 02:09 |
| cloudnull | all you really need is ZNC | 02:09 |
| cloudnull | ZNC keeps you connected, then you have your client connet to ZNC | 02:09 |
| cloudnull | in my case the client is "TheLounge" | 02:10 |
| spatel | is ZNC like middleware ? | 02:10 |
| cloudnull | https://wiki.znc.in/ZNC | 02:10 |
| cloudnull | ZNC is an IRC bouncer | 02:10 |
| spatel | oh! | 02:10 |
| spatel | let me read and set it up | 02:10 |
| *** errr has quit IRC | 02:12 | |
| openstackgerrit | Merged openstack/openstack-ansible-os_trove stable/rocky: Remove galera_client meta-dependency https://review.openstack.org/603763 | 02:30 |
| *** errr has joined #openstack-ansible | 02:37 | |
| *** cshen has joined #openstack-ansible | 02:44 | |
| *** dxiri has joined #openstack-ansible | 02:46 | |
| *** cshen has quit IRC | 02:49 | |
| *** Bhujay has joined #openstack-ansible | 03:01 | |
| *** sawblade6 has joined #openstack-ansible | 03:01 | |
| *** sawblade6 has quit IRC | 03:06 | |
| openstackgerrit | Merged openstack/openstack-ansible-os_cinder stable/rocky: Remove galera_client meta-dependency https://review.openstack.org/603714 | 03:26 |
| *** jmccrory has quit IRC | 03:32 | |
| openstackgerrit | Merged openstack/openstack-ansible-os_nova stable/pike: Remove galera_client meta-dependency https://review.openstack.org/603713 | 03:33 |
| *** jmccrory has joined #openstack-ansible | 03:37 | |
| openstackgerrit | Merged openstack/openstack-ansible-os_nova master: Drop SELinux support for CentOS 7 https://review.openstack.org/603860 | 03:38 |
| *** sdake has quit IRC | 03:38 | |
| *** mino_ has joined #openstack-ansible | 03:40 | |
| *** sawblade6 has joined #openstack-ansible | 03:42 | |
| *** sdake has joined #openstack-ansible | 03:43 | |
| openstackgerrit | Merged openstack/openstack-ansible-os_octavia stable/queens: Remove galera_client meta-dependency https://review.openstack.org/603788 | 03:51 |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/openstack-ansible master: Imported Translations from Zanata https://review.openstack.org/604255 | 03:51 |
| *** sawblade6 has quit IRC | 03:52 | |
| *** aedc has joined #openstack-ansible | 03:56 | |
| *** aevoo has quit IRC | 03:56 | |
| *** spatel has quit IRC | 03:58 | |
| *** Bhujay has quit IRC | 03:58 | |
| openstackgerrit | Merged openstack/openstack-ansible master: Minimal(ist) network config for nspawn gating https://review.openstack.org/593230 | 04:23 |
| *** udesale has joined #openstack-ansible | 04:29 | |
| *** jhesketh_ has joined #openstack-ansible | 04:32 | |
| *** d34dh0r53 has quit IRC | 04:34 | |
| *** d34dh0r53 has joined #openstack-ansible | 04:34 | |
| *** dxiri has quit IRC | 04:35 | |
| *** jhesketh has quit IRC | 04:36 | |
| *** sawblade6 has joined #openstack-ansible | 04:40 | |
| *** cshen has joined #openstack-ansible | 04:43 | |
| *** sawblade6 has quit IRC | 04:44 | |
| *** cshen has quit IRC | 04:47 | |
| *** vnogin has joined #openstack-ansible | 05:14 | |
| *** Bhujay has joined #openstack-ansible | 05:18 | |
| *** vnogin has quit IRC | 05:19 | |
| prometheanfire | lol, ansible just tagged vX.Y.Z | 05:23 |
| *** jhesketh_ is now known as jhesketh | 05:25 | |
| *** sawblade6 has joined #openstack-ansible | 05:27 | |
| *** sawblade6 has quit IRC | 05:36 | |
| *** shyamb has joined #openstack-ansible | 05:51 | |
| *** sawblade6 has joined #openstack-ansible | 05:55 | |
| *** sawblade6 has quit IRC | 05:59 | |
| *** shyamb has quit IRC | 06:01 | |
| *** shyamb has joined #openstack-ansible | 06:02 | |
| openstackgerrit | Andreas Jaeger proposed openstack/openstack-ansible-os_octavia master: switch documentation job to new PTI https://review.openstack.org/594685 | 06:07 |
| *** gkadam has joined #openstack-ansible | 06:13 | |
| *** shyamb has quit IRC | 06:13 | |
| *** gkadam has quit IRC | 06:13 | |
| *** shyamb has joined #openstack-ansible | 06:14 | |
| *** shyamb has quit IRC | 06:21 | |
| *** shyamb has joined #openstack-ansible | 06:22 | |
| *** chkumar|off is now known as chkumar|ruck | 06:23 | |
| openstackgerrit | Andreas Jaeger proposed openstack/openstack-ansible-os_octavia master: switch documentation job to new PTI https://review.openstack.org/594685 | 06:26 |
| *** cshen has joined #openstack-ansible | 06:27 | |
| *** sawblade6 has joined #openstack-ansible | 06:35 | |
| *** shyamb has quit IRC | 06:38 | |
| *** sawblade6 has quit IRC | 06:46 | |
| *** shyamb has joined #openstack-ansible | 06:58 | |
| *** shyamb has quit IRC | 07:13 | |
| *** shyamb has joined #openstack-ansible | 07:13 | |
| *** sep has joined #openstack-ansible | 07:16 | |
| *** sawblade6 has joined #openstack-ansible | 07:17 | |
| *** gkadam has joined #openstack-ansible | 07:19 | |
| *** shyamb has quit IRC | 07:20 | |
| *** shyamb has joined #openstack-ansible | 07:20 | |
| mbuil_ | is anybody else experiencing an error in the task "os_neutron : Grant access to the DB for the service" in the distro_install_ubuntu_xenial gate? | 07:20 |
| *** sawblade6 has quit IRC | 07:22 | |
| *** shardy has joined #openstack-ansible | 07:23 | |
| *** phasespace has joined #openstack-ansible | 07:28 | |
| *** fghaas has joined #openstack-ansible | 07:30 | |
| openstackgerrit | Markos Chandras (hwoarang) proposed openstack/openstack-ansible-rabbitmq_server master: RedHat: Use the distribution package https://review.openstack.org/604124 | 07:31 |
| *** hamzaachi has joined #openstack-ansible | 07:31 | |
| openstackgerrit | Markos Chandras (hwoarang) proposed openstack/openstack-ansible-os_neutron master: SUSE: Add support for openSUSE Leap 15 https://review.openstack.org/604286 | 07:32 |
| *** shyamb has quit IRC | 07:39 | |
| *** tosky has joined #openstack-ansible | 07:40 | |
| *** ppetit has joined #openstack-ansible | 07:49 | |
| arxcruz | hey guys, where the jobs for os_tempest are described? I check the zuul.d directory, and there's only a template... | 07:49 |
| arxcruz | nevermind, got it | 07:51 |
| openstackgerrit | Markos Chandras (hwoarang) proposed openstack/openstack-ansible-rabbitmq_server master: RedHat: Use the distribution package https://review.openstack.org/604124 | 07:55 |
| openstackgerrit | Merged openstack/openstack-ansible-os_designate stable/rocky: Remove galera_client meta-dependency https://review.openstack.org/603731 | 08:05 |
| openstackgerrit | Andreas Jaeger proposed openstack/openstack-ansible-os_octavia master: switch documentation job to new PTI https://review.openstack.org/594685 | 08:06 |
| openstackgerrit | Merged openstack/openstack-ansible-os_glance stable/queens: Remove galera_client meta-dependency https://review.openstack.org/603726 | 08:12 |
| *** shyamb has joined #openstack-ansible | 08:13 | |
| *** vnogin has joined #openstack-ansible | 08:15 | |
| *** DanyC has joined #openstack-ansible | 08:15 | |
| *** vnogin has quit IRC | 08:19 | |
| openstackgerrit | Markos Chandras (hwoarang) proposed openstack/openstack-ansible-rabbitmq_server master: RedHat: Use the distribution package https://review.openstack.org/604124 | 08:23 |
| openstackgerrit | Arx Cruz proposed openstack/openstack-ansible-tests master: Add stackviz directory to the collect logs https://review.openstack.org/604292 | 08:24 |
| *** cshen has quit IRC | 08:24 | |
| *** smoke87 has quit IRC | 08:26 | |
| hwoarang | jrosser: hello. ubuntu still deploys the rocky repo on master branches. is there a stein or master that we should use instead? | 08:34 |
| hwoarang | i am a bit at a loss on where to find this info. http://ubuntu-cloud.archive.canonical.com/ubuntu/dists/ only has rocky | 08:35 |
| openstackgerrit | Markos Chandras (hwoarang) proposed openstack/openstack-ansible-os_neutron master: zuul: Switch Xenial jobs with Bionic https://review.openstack.org/604297 | 08:38 |
| hwoarang | mbuil_: ^ | 08:38 |
| hwoarang | mbuil_: if you are seeing distro+xenial probs on rocky, then we should backport | 08:39 |
| *** shyamb has quit IRC | 08:41 | |
| openstackgerrit | Arx Cruz proposed openstack/openstack-ansible-os_tempest master: WIP - Enable stackviz support https://review.openstack.org/603100 | 08:43 |
| mbuil_ | hwoarang: ok :) | 08:44 |
| odyssey4me | hwoarang: ubuntu won't push anything to UCA for Stein until M1 | 08:44 |
| odyssey4me | Apparently there is a PPA for nightly builds. | 08:44 |
| hwoarang | aha | 08:44 |
| hwoarang | ok i am guessing we should be removing xenial on this cycle right? | 08:45 |
| *** Emine has joined #openstack-ansible | 08:45 | |
| * hwoarang checks the ptg etherpad | 08:46 | |
| evrardjp | yeah that would make sense | 08:47 |
| evrardjp | jamespage told me the ppa was regularily "fixed" | 08:47 |
| hwoarang | centos and suse are on nightly builds so we can be brave and switch ubuntu too ;p | 08:48 |
| *** evrardjp has quit IRC | 08:51 | |
| hwoarang | how do you feel about https://review.openstack.org/#/c/604124/ ? maybe help with reliability of centos7 jobs + added benefit of RDO supported package | 08:56 |
| *** evrardjp has joined #openstack-ansible | 08:56 | |
| odyssey4me | hwoarang: yep, xenial should be dropped from master - all distro jobs for rocky should be bionic only | 09:00 |
| odyssey4me | hwoarang: you've done the same for suse haven't you? | 09:02 |
| hwoarang | odyssey4me: what do you mean "the same" ? | 09:04 |
| odyssey4me | hwoarang: yep, opensuse is also using the 'distro' source | 09:04 |
| hwoarang | yeah and leap 42 and 15 are supported in stein as far as i know | 09:04 |
| hwoarang | ah you prob mean the rabbitmq_server patch | 09:06 |
| hwoarang | yeah. i am just not sure what happens with the package locks on centos7. maybe upgrades wont play nice | 09:06 |
| *** sawblade6 has joined #openstack-ansible | 09:08 | |
| *** shyamb has joined #openstack-ansible | 09:08 | |
| *** Emine has quit IRC | 09:09 | |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible stable/rocky: Minimal(ist) network config for nspawn gating https://review.openstack.org/604308 | 09:09 |
| *** sawblade6 has quit IRC | 09:12 | |
| odyssey4me | hwoarang: maybe we should add a distro install job template in the tests repo, then just use the template in the repositories which support it? | 09:13 |
| odyssey4me | I suppose the support is a bit inconsistent right now, right? | 09:13 |
| hwoarang | hmm | 09:18 |
| hwoarang | core services should be fine i suppose. not sure how centos is doing to be honest | 09:18 |
| *** Emine has joined #openstack-ansible | 09:21 | |
| *** sawblade6 has joined #openstack-ansible | 09:28 | |
| *** cshen has joined #openstack-ansible | 09:32 | |
| *** sawblade6 has quit IRC | 09:33 | |
| *** cshen has quit IRC | 09:38 | |
| *** shyamb has quit IRC | 09:40 | |
| *** fghaas has quit IRC | 09:45 | |
| *** sawblade6 has joined #openstack-ansible | 09:51 | |
| odyssey4me | jrosser evrardjp mind reviewing https://review.openstack.org/603677 to get it into the rocky release? | 09:53 |
| *** cshen has joined #openstack-ansible | 09:54 | |
| *** sawblade6 has quit IRC | 09:55 | |
| odyssey4me | hwoarang: think this is transient? http://logs.openstack.org/56/603756/1/check/openstack-ansible-functional-distro_install-opensuse-423/a58cd40/job-output.txt.gz#_2018-09-21_00_50_43_964159 from https://review.openstack.org/#/c/603756/ | 09:58 |
| hwoarang | oh this again? | 09:59 |
| hwoarang | let me check | 09:59 |
| evrardjp | voted | 09:59 |
| evrardjp | not what you like but voted anyway : p | 09:59 |
| odyssey4me | evrardjp: it is needed because those are the current requirements. mbuil_ is working on fixing the requirements upstream, but until then, we're stuck with bad requirements. | 10:00 |
| hwoarang | odyssey4me: thank you i am going to bug the cloud people again | 10:00 |
| odyssey4me | freezing on an old version is not something we can perpetuate forever | 10:01 |
| openstackgerrit | Merged openstack/openstack-ansible-lxc_hosts master: Drop SELinux support for CentOS 7 https://review.openstack.org/604190 | 10:01 |
| evrardjp | odyssey4me: I agree. But this is a stable branch. I am not fine in deciding "now we'll install horizon and ceilometer in neutron venv" | 10:01 |
| evrardjp | it least a better explanation and more complete | 10:02 |
| evrardjp | like I said in my comment | 10:02 |
| evrardjp | if we are ready to bring that in, we should 1) unfreeze the shas for networking-odl/networking-bgpvpn 2) make sure horizon is also in all 3) put a release note | 10:03 |
| *** shyamb has joined #openstack-ansible | 10:03 | |
| evrardjp | I said that far earlier that it had to be done and I was told it was planned -- Nobody did it, and now we are in workarounds time, and those workarounds aren't great. | 10:04 |
| odyssey4me | right, ok | 10:06 |
| evrardjp | I strongly believe we should fix upstream instead of implementing workarounds in OSA, that's just my point :) | 10:06 |
| evrardjp | until upstream is fixed, what's the problem with freezing? | 10:06 |
| odyssey4me | evrardjp: yep, I'll just wait for mbuil_ to finish sorting that out - mbuil_ if you need a hand, let me know | 10:06 |
| odyssey4me | evrardjp: the problem with freezing is that we're frozen on pre-release stuff, but meh, they're not our primary use-cases so I guess we just wait | 10:07 |
| openstackgerrit | Merged openstack/openstack-ansible-os_almanach stable/ocata: Do not log passwords https://review.openstack.org/603951 | 10:07 |
| evrardjp | odyssey4me: remember that adding horizon in neutron venvs is fun: it's gonna bring all the django in... For neutron agents that seem very overkill (yes it's not for all cases, but still -- what a bloat) | 10:07 |
| evrardjp | odyssey4me: well I said it many times, and pinged many times: this should be fixed -- told multiple people too. I am not sure why people didn't care, but I didn't get the chance to work on it, so I am very sorry. | 10:08 |
| evrardjp | if people believe this is critical and should be done, please vote positively. I am just giving my opinion :) | 10:09 |
| mbuil_ | odyssey4me, evrardjp: to be honest, for my project in OPNFV, I could not care less about networking-bgpvpn. I sent an email last week to the guys using that project explaining them the situation and asking them for help. I got no answer, which is not surprise as most of the people from that project left. So, I wonder if anybody is using networking-bgpvpn... therefore, I suggest that we keep networking-bgpvpn frozen until somebody | 10:11 |
| mbuil_ | fixes the horizon dependency | 10:11 |
| hwoarang | but wasn't there a discussion in the ML about ceilometer and bgpvpn? | 10:13 |
| hwoarang | prometheanfire was complaining about the bad requirements in these projects with ceilometer or something. i can dig up the link again but mbuil_ I think I sent it to you. was that for a different issue | 10:14 |
| hwoarang | ? | 10:14 |
| *** fghaas has joined #openstack-ansible | 10:14 | |
| mbuil_ | hwoarang: ceilometer dependency problem was in networking-odl project. That one we need for all OPNFV scenarios that require ODL (e.g. bgpvpn and sfc). I am taking care of that one with the ceilometer workarounds and this patch ==> https://review.openstack.org/#/c/602622/ | 10:15 |
| hwoarang | i see | 10:15 |
| mbuil_ | hwoarang: networking-bgpvpn has a similar problem with, in this case, a horizon dependency... | 10:15 |
| openstackgerrit | Merged openstack/openstack-ansible-apt_package_pinning master: add the project source code repository https://review.openstack.org/600612 | 10:17 |
| *** fghaas has quit IRC | 10:18 | |
| *** gokhani has quit IRC | 10:19 | |
| openstackgerrit | Merged openstack/openstack-ansible-openstack_hosts stable/rocky: Reduce number of installed packages for CentOS https://review.openstack.org/603926 | 10:21 |
| openstackgerrit | Merged openstack/openstack-ansible-os_sahara stable/rocky: Remove galera_client meta-dependency https://review.openstack.org/603644 | 10:29 |
| openstackgerrit | Merged openstack/openstack-ansible-os_swift stable/queens: Clean up role tests https://review.openstack.org/603898 | 10:30 |
| openstackgerrit | Merged openstack/openstack-ansible-os_swift stable/ocata: Clean up role tests https://review.openstack.org/603904 | 10:30 |
| openstackgerrit | Merged openstack/openstack-ansible-os_swift stable/ocata: Updated from OpenStack Ansible Tests https://review.openstack.org/602317 | 10:30 |
| openstackgerrit | Merged openstack/openstack-ansible-os_swift stable/ocata: import zuul job settings from project-config https://review.openstack.org/594760 | 10:30 |
| openstackgerrit | Merged openstack/openstack-ansible-os_swift stable/pike: Clean up role tests https://review.openstack.org/603902 | 10:30 |
| *** shyamb has quit IRC | 10:32 | |
| *** shyamb has joined #openstack-ansible | 10:33 | |
| *** mino_ has quit IRC | 10:34 | |
| *** ppetit has quit IRC | 10:37 | |
| *** ppetit has joined #openstack-ansible | 10:37 | |
| *** shyamb has quit IRC | 10:42 | |
| *** Emine has quit IRC | 10:44 | |
| *** cshen has quit IRC | 10:46 | |
| *** sum12 has joined #openstack-ansible | 10:52 | |
| openstackgerrit | Merged openstack/openstack-ansible-os_neutron stable/queens: Clean up role tests https://review.openstack.org/603861 | 10:55 |
| *** fghaas has joined #openstack-ansible | 11:01 | |
| openstackgerrit | Merged openstack/openstack-ansible-os_neutron stable/queens: Drop SELinux support for CentOS 7 https://review.openstack.org/603472 | 11:03 |
| openstackgerrit | Merged openstack/openstack-ansible-os_neutron stable/queens: Remove galera_client meta-dependency https://review.openstack.org/603721 | 11:03 |
| *** udesale has quit IRC | 11:05 | |
| *** cshen has joined #openstack-ansible | 11:13 | |
| openstackgerrit | Arx Cruz proposed openstack/openstack-ansible-os_tempest master: WIP - Enable stackviz support https://review.openstack.org/603100 | 11:13 |
| *** udesale has joined #openstack-ansible | 11:13 | |
| *** ppetit has quit IRC | 11:15 | |
| odyssey4me | mbuil_: your patch looks good to me, and hopefully that can get resolved ASAP - could you perhaps submit a similar one for networking-bgpvpn given that you've now worked it out? I know it's not something you directly care about, but it'd be helpful for us | 11:16 |
| evrardjp | mbuil_: that makes sense that you care on your focus area. In that case, may I suggest that, if we decide to bring ceilometer in the neutron venvs, we unfreeze odl branch matching (networking_odl_git_track_branch), and document your other upstream work in the commit message | 11:17 |
| odyssey4me | I do hope that we can also get that patch ported back to rocky. | 11:17 |
| jamesdenton | mornin' folks | 11:17 |
| evrardjp | in the meantime I am fine with a split approach bgpvpn being different than odl | 11:18 |
| evrardjp | morning jamesdenton | 11:18 |
| evrardjp | if we agree on the approach I will re-issue a release anyway today | 11:18 |
| evrardjp | I also have less issues on bringing ceilometer in neutron than horizon in neutron venvs :p | 11:19 |
| *** shyamb has joined #openstack-ansible | 11:22 | |
| odyssey4me | yep, that's true | 11:25 |
| *** spatel has joined #openstack-ansible | 11:26 | |
| *** spatel has quit IRC | 11:31 | |
| *** ppetit has joined #openstack-ansible | 11:32 | |
| *** udesale has quit IRC | 11:34 | |
| *** udesale has joined #openstack-ansible | 11:34 | |
| *** hamzaachi has quit IRC | 11:34 | |
| *** sawblade6 has joined #openstack-ansible | 11:36 | |
| *** sawblade6 has quit IRC | 11:44 | |
| *** chkumar|ruck is now known as chkumar|off | 11:46 | |
| mbuil_ | odyssey4me: I'll create a patch for neutron-bgpvpn which "optionalizes" horizon, no problem :) | 11:48 |
| kaiokmo | mornings | 11:48 |
| kaiokmo | evrardjp: I left a comment for you on https://review.openstack.org/#/c/588584/ | 11:55 |
| *** aedc has quit IRC | 11:58 | |
| *** aedc_ has joined #openstack-ansible | 11:58 | |
| evrardjp | kaiokmo: thanks -- I think I was unclear: I don't really mind seing zypper instead of package (as it is actually faster). My concern was that we have variables undefined by distro -- which sets a precedent. And if someone defines that variable for another distro, the deploy would simply fail | 12:05 |
| mbuil_ | evrardjp: let me check if I understand correctly, you want me to add the unfreeze of networking-odl into this patch (https://review.openstack.org/#/c/603677/) and explain in the commit message the current status of the "optionalization" of ceilometer in n-odl and horizon in n-bgpvpn? | 12:05 |
| evrardjp | mbuil_: not anything with bgpvpn | 12:05 |
| kaiokmo | evrardjp: ah, I see. | 12:06 |
| evrardjp | mbuil_: the goal of this patch is to have latest networking-odl: so please make sure you have latest networking-odl by doing two things: allowing ceilometer to build at all cases (that's what you did), but also making sure it's properly bump-able (see the track branch). On the commit message, explaining why you had to do 'ceilometer_git_project_group: all' seems important, explicitly telling | 12:07 |
| evrardjp | that line need to happen until an upstream fix of n-odl removes that mandatory dependency | 12:07 |
| evrardjp | which is kinda your commit message already, but needs clarification : ) | 12:08 |
| evrardjp | I will propose something for you :) | 12:09 |
| evrardjp | mbuil_: I suggested a commit message : ) | 12:14 |
| evrardjp | kaiokmo: you can keep the module as this is right now, but just add those conditions, and I am good :) | 12:15 |
| evrardjp | explicit is better than implicit :D | 12:15 |
| kaiokmo | evrardjp: will do. thanks | 12:15 |
| evrardjp | I thank you : ) | 12:15 |
| openstackgerrit | Andy Smith proposed openstack/openstack-ansible-os_keystone master: Update messaging configuration https://review.openstack.org/604340 | 12:18 |
| odyssey4me | thank you very much mbuil_ - we appreciate the effort you've made | 12:20 |
| odyssey4me | maybe we can have evrardjp/mnaser try to get the teams involved to review them quickly ;) | 12:20 |
| openstackgerrit | Merged openstack/openstack-ansible stable/rocky: Ensure package cache is updated for nspawn containers https://review.openstack.org/604106 | 12:21 |
| odyssey4me | I have a suspicion that we might have hit some rate limits or something like that with packagecloud yesterday due to the volume of jobs trying to grab the rabbitmq packages. | 12:22 |
| mbuil_ | odyssey4me: you are welcome. Before I thought you were expecting me to create the whole workaround to make horizon work but I think it does make sense ;) | 12:23 |
| odyssey4me | mbuil_: oh no, it's far better for everyone in openstack if the requirements as far upstream as possible are correctly expressed | 12:23 |
| odyssey4me | as far as I know, the requirements team already had issues with these same packages - but the teams involved seem either not to care, or are clueless how to solve it | 12:24 |
| *** udesale has quit IRC | 12:26 | |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-os_gnocchi master: Define cache_timeout variable https://review.openstack.org/604346 | 12:27 |
| odyssey4me | cores - I'd appreciate a prompt review of https://review.openstack.org/604346 - it's a very simple patch | 12:28 |
| *** shyamb has quit IRC | 12:29 | |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-os_cinder stable/ocata: Remove tempest testing and clean up role tests https://review.openstack.org/603876 | 12:30 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-os_cinder stable/ocata: import zuul job settings from project-config https://review.openstack.org/594743 | 12:31 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-os_cinder stable/ocata: Updated from OpenStack Ansible Tests https://review.openstack.org/602302 | 12:31 |
| *** gkadam has quit IRC | 12:34 | |
| openstackgerrit | Andy Smith proposed openstack/openstack-ansible-os_heat master: Update messaging notification configuration https://review.openstack.org/604348 | 12:35 |
| mbuil_ | odyssey4me: can I upload a patch set to your cherry pick: https://review.openstack.org/#/c/603677/ with evrardjp's comments (unfreezing n-odl)? | 12:37 |
| evrardjp | mbuil_: Nike method | 12:38 |
| evrardjp | mbuil_: "Just do it" | 12:38 |
| evrardjp | hahah | 12:38 |
| odyssey4me | mbuil_: well, it's probably best to push a patch up to master first - then for rocky we combine them | 12:38 |
| mbuil_ | evrardjp: I first ask, I am a gentleman | 12:40 |
| evrardjp | hehe | 12:40 |
| evrardjp | valid reasoning :) odyssey4me is a "do it" kind of person fyi :p | 12:41 |
| openstackgerrit | Merged openstack/openstack-ansible-os_aodh stable/rocky: Remove galera_client meta-dependency https://review.openstack.org/603717 | 12:41 |
| odyssey4me | yeah, I don't have issues with people changing my patch as long as it doesn't change it materially (ie do something in a different way) - if that's needed, then an alternative patch is better. | 12:42 |
| odyssey4me | But spelling corrections, extra comments, formatting changes, etc are always welcome. | 12:43 |
| openstackgerrit | Manuel Buil proposed openstack/openstack-ansible master: Unfreeze networking-odl https://review.openstack.org/602243 | 12:47 |
| mbuil_ | odyssey4me ^ | 12:48 |
| openstackgerrit | Markos Chandras (hwoarang) proposed openstack/openstack-ansible-os_neutron master: SUSE: Add support for openSUSE Leap 15 https://review.openstack.org/604286 | 12:48 |
| *** hamzaachi has joined #openstack-ansible | 12:49 | |
| openstackgerrit | Markos Chandras (hwoarang) proposed openstack/openstack-ansible-os_nova master: SUSE: Add support for openSUSE Leap 15 https://review.openstack.org/604080 | 12:50 |
| *** aedc_ is now known as aedv | 12:51 | |
| *** aedv is now known as aedc | 12:51 | |
| hwoarang | odyssey4me: re http://logs.openstack.org/97/604297/1/check/openstack-ansible-functional-distro_install-ubuntu-bionic/02cb46d/job-output.txt.gz#_2018-09-21_09_11_30_068148 from https://review.openstack.org/#/c/604297/ should I assume that there is a temp issue with the ubuntu repos or do i need to dig deeper? | 12:53 |
| openstackgerrit | Merged openstack/openstack-ansible-os_cinder stable/rocky: Remove python-keystoneclient package install https://review.openstack.org/603409 | 12:54 |
| odyssey4me | hwoarang: it looks like there's been a half update to UCA, because the beta dep is still in http://ubuntu-cloud.archive.canonical.com/ubuntu/dists/bionic-updates/rocky/main/binary-amd64/Packages | 12:55 |
| odyssey4me | perhaps tinwood or jamespage can help | 12:56 |
| odyssey4me | lemme ask in the charms channel | 12:57 |
| openstackgerrit | Manuel Buil proposed openstack/openstack-ansible master: Unfreeze networking-odl https://review.openstack.org/602243 | 12:57 |
| odyssey4me | hwoarang: it looks like the package name changed - 'neutron-linuxbridge-agent' shows that it replaces 'neutron-plugin-linuxbridge-agent' | 12:59 |
| hwoarang | hmm | 12:59 |
| hwoarang | is that official or by accident? :p | 13:00 |
| hwoarang | before they had both | 13:00 |
| odyssey4me | yeah, it looks like they've consolidated packages - all the beta 'neutron-plugin-*' packages are present in 'Replaces:' for higher level packages. | 13:03 |
| odyssey4me | neutron-plugin-ml2 is still valid | 13:03 |
| odyssey4me | so it's all the 'neutron-plugin-*-agent' packages which have consolidated | 13:04 |
| hwoarang | ok les try | 13:07 |
| openstackgerrit | Markos Chandras (hwoarang) proposed openstack/openstack-ansible-os_neutron master: zuul: Switch Xenial jobs to Bionic https://review.openstack.org/604297 | 13:07 |
| odyssey4me | I guess this work was initially done with xenial/queens - where all the package names were different. | 13:08 |
| openstackgerrit | Markos Chandras (hwoarang) proposed openstack/openstack-ansible master: zuul: Restore the openSUSE deployment jobs https://review.openstack.org/599425 | 13:18 |
| mnaser | morning | 13:19 |
| mnaser | odyssey4me: im lacking context, get who involved in what? :> | 13:20 |
| *** spatel_ has joined #openstack-ansible | 13:21 | |
| openstackgerrit | Arx Cruz proposed openstack/openstack-ansible-os_tempest master: WIP - Enable stackviz support https://review.openstack.org/603100 | 13:21 |
| spatel_ | jamesdenton: morning | 13:22 |
| jamesdenton | hello spatel_ | 13:22 |
| jamesdenton | saw you had some better results? | 13:23 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-os_heat master: Move heat domain setup into service setup tasks https://review.openstack.org/604362 | 13:23 |
| spatel_ | now i am able to pump "nuttcp -T 30 -i -u -w4m -R 7G 10.10.10.10" with zero packetloss | 13:23 |
| spatel_ | 3540.0918 MB / 30.00 sec = 989.8372 Mbps 52 %TX 42 %RX 0 / 3625054 drop/pkt 0.00 %loss | 13:23 |
| *** clone99 has joined #openstack-ansible | 13:23 | |
| spatel_ | This is what is changed, upgrade kernel to 4.18 (disable specter/meltdown patch) | 13:23 |
| spatel_ | I have noticed specter/meltdown disable boot your performance 10% | 13:24 |
| *** clone99 has quit IRC | 13:24 | |
| spatel_ | boost* | 13:24 |
| *** SimAloo has joined #openstack-ansible | 13:24 | |
| spatel_ | jamesdenton: check this out http://paste.openstack.org/show/730537/ | 13:25 |
| odyssey4me | mnaser: some of the neutron driver/plugin projects did some bad things in their requirements - eg: networking-odl added ceilometer as a requirement, and networking-bgpvpn added horizon as a requirement. mbuil_ is working on switching those from hard requirements to optional requirements and we may need some help pushing the teams to review the patches, otherwise our SHA pins for those drivers stay pre-re | 13:25 |
| odyssey4me | lease | 13:25 |
| jamesdenton | so one thing - That middle number, 989.8372 Mbps, to me implies it's only pushing < 1G. Compared to mine where I saw 6999 Mbps or so. You might want to verify with iperf or something that you can push ~10G with TCP. | 13:25 |
| spatel_ | in iperf i am getting ~1.2G max :( | 13:26 |
| mnaser | odyssey4me: i can try pinging people once we have reviews | 13:26 |
| spatel_ | that is very odd | 13:26 |
| spatel_ | does VF limit your data ? | 13:26 |
| jamesdenton | spatel_ that seems highly suspect. That's between computes also? or just VMs | 13:26 |
| openstackgerrit | Merged openstack/openstack-ansible-os_neutron stable/rocky: Remove galera_client meta-dependency https://review.openstack.org/603720 | 13:26 |
| spatel_ | between compute i get 4 to 5Gbps on iperf | 13:27 |
| odyssey4me | mnaser evrardjp has done a great job on the one that's up to pick up on a problem with the current patch - you can track https://review.openstack.org/602622 FYI | 13:27 |
| spatel_ | but vm-to-vm only 1.2Gbps | 13:27 |
| jamesdenton | spatel_ even that isn't great. I am in the process of bringing up a node w/ that QLogic NIC. | 13:27 |
| spatel_ | perfect!! | 13:27 |
| spatel_ | look like its something hardware related issue.. | 13:27 |
| spatel_ | do you have any special kernel/tcp tuning | 13:28 |
| mnaser | odyssey4me: evrardjp i know the ODL people well | 13:28 |
| mnaser | i can push this over to get it reviewed if we don't have a -1 :P | 13:28 |
| odyssey4me | hwoarang: any news for http://logs.openstack.org/56/603756/1/check/openstack-ansible-functional-distro_install-opensuse-423/2da1856/logs/ara-report/result/92e3380b-4279-4811-bcac-dc1e7244fc43/ ? | 13:29 |
| spatel_ | i think INTEL is always best.. if your QLogic match with my result then i will be happy :) | 13:29 |
| hwoarang | odyssey4me: yeah it's fixed just waiting for mirrors to sync | 13:29 |
| odyssey4me | ok good, thanks! | 13:29 |
| evrardjp | spatel_: my intel says otherwise. That was a bad joke, sorry. | 13:29 |
| jamesdenton | spatel_ Well, yes, on those other nodes I have CPU affinity, DPDK, hugepages, etc. But host-to-host I get ~9G, so none of that is taken into consideration. That's on Intel X520/X540 though. This node will be configured for SR-IOV so i can compare. Had to move the PCI card to get SR-IOV working. | 13:30 |
| jamesdenton | evrardjp tells jokes, too. what can he NOT do? | 13:30 |
| evrardjp | jamesdenton: it looks like we have the same NICs: p | 13:30 |
| evrardjp | jamesdenton: hahaha | 13:30 |
| hwoarang | odyssey4me: i see the infra mirrors have the package. could you run a 'recheck' ? | 13:30 |
| evrardjp | jamesdenton: do things well? :p | 13:30 |
| spatel_ | When you say CPU affinity does that mean (NUMA) ? | 13:31 |
| jamesdenton | <--- same here | 13:31 |
| spatel_ | jamesdenton: i have meeting i will be back in 30 min... :) | 13:31 |
| jamesdenton | spatel_ just happen to be pinned to CPUs in the same NUMA node in this case, yes | 13:31 |
| jamesdenton | found some bugs with the SR-IOV playbooks, so i'll have to log those. "Throw em on the pile" | 13:32 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-os_monasca master: Delegate the database schema import to monasca_db_setup_host https://review.openstack.org/604364 | 13:36 |
| odyssey4me | hwoarang: thanks will do that now | 13:36 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-os_monasca master: Remove galera_client meta-dependency https://review.openstack.org/602905 | 13:37 |
| openstackgerrit | Markos Chandras (hwoarang) proposed openstack/openstack-ansible master: [DO NOT MERGE] SUSE: Add support for openSUSE Leap 15 https://review.openstack.org/604366 | 13:37 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-os_cinder stable/queens: Remove galera_client meta-dependency https://review.openstack.org/603715 | 13:41 |
| openstackgerrit | Markos Chandras (hwoarang) proposed openstack/openstack-ansible master: zuul: Restore the openSUSE deployment jobs https://review.openstack.org/599425 | 13:44 |
| *** hamzaachi has quit IRC | 13:45 | |
| *** spatel_ has quit IRC | 13:47 | |
| openstackgerrit | Markos Chandras (hwoarang) proposed openstack/openstack-ansible master: SUSE: Add support for openSUSE Leap 15 https://review.openstack.org/604366 | 13:47 |
| openstackgerrit | Kaio Kassiano Moura Oliveira proposed openstack/openstack-ansible-os_monasca master: Add support for openSUSE https://review.openstack.org/588584 | 13:49 |
| *** goldenfri has joined #openstack-ansible | 13:59 | |
| openstackgerrit | Merged openstack/openstack-ansible-rabbitmq_server master: RedHat: Use the distribution package https://review.openstack.org/604124 | 14:05 |
| odyssey4me | mnaser hwoarang how do you feel about porting https://review.openstack.org/604124 back to rocky? | 14:07 |
| *** openstackgerrit has quit IRC | 14:07 | |
| *** phasespace has quit IRC | 14:08 | |
| odyssey4me | I would personally support it. :) | 14:08 |
| *** openstackgerrit has joined #openstack-ansible | 14:12 | |
| openstackgerrit | Merged openstack/openstack-ansible-os_cinder stable/ocata: Remove tempest testing and clean up role tests https://review.openstack.org/603876 | 14:12 |
| kaiokmo | hwoarang: thank you for the comments | 14:12 |
| *** aedc_ has joined #openstack-ansible | 14:12 | |
| *** aedc has quit IRC | 14:13 | |
| *** radeks has joined #openstack-ansible | 14:13 | |
| openstackgerrit | Merged openstack/openstack-ansible-os_neutron stable/ocata: Clean up role tests https://review.openstack.org/603865 | 14:17 |
| openstackgerrit | Merged openstack/openstack-ansible-os_gnocchi master: Define cache_timeout variable https://review.openstack.org/604346 | 14:19 |
| prometheanfire | mbuil_: hwoarang I get grumpy when I think of projects(plugins) using service projects (non lib, non cycle with intermediary) as libraries | 14:21 |
| hwoarang | makes sense | 14:22 |
| hwoarang | odyssey4me: i dont mind (re rabbitmq) | 14:22 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-rabbitmq_server stable/rocky: RedHat: Use the distribution package https://review.openstack.org/604374 | 14:23 |
| *** lbragstad is now known as elbragstad | 14:24 | |
| *** goldenfri has quit IRC | 14:25 | |
| openstackgerrit | Merged openstack/openstack-ansible-os_nova stable/rocky: Remove galera_client meta-dependency https://review.openstack.org/603710 | 14:27 |
| openstackgerrit | Kaio Kassiano Moura Oliveira proposed openstack/openstack-ansible-os_monasca master: Add support for openSUSE 42.x https://review.openstack.org/588584 | 14:28 |
| openstackgerrit | Merged openstack/openstack-ansible-os_cinder stable/ocata: import zuul job settings from project-config https://review.openstack.org/594743 | 14:34 |
| openstackgerrit | Merged openstack/openstack-ansible-os_cinder stable/ocata: Updated from OpenStack Ansible Tests https://review.openstack.org/602302 | 14:34 |
| *** cshen has quit IRC | 14:38 | |
| *** Bhujay has quit IRC | 14:38 | |
| ansmith | odyssey4me: hello, thanks for the comments re. https://review.openstack.org/#/c/604131 | 14:39 |
| ansmith | odyssey4me: including/excluding the configuration does not inhibit a notifier | 14:40 |
| mnaser | odyssey4me: im worried about upgrade path | 14:40 |
| mnaser | we need to see what that looks like | 14:40 |
| mnaser | we went from 3.7 to 3.6 | 14:40 |
| hwoarang | and i think the locks are not managed properly when you switch from external repo to distro | 14:40 |
| hwoarang | maybe... | 14:41 |
| ansmith | odyssey4me: the patch is looking to ensure the correct messaging bus gets used in the hybrid case as if the notification transport is not in the config, the default transport will get used | 14:41 |
| *** Bhujay has joined #openstack-ansible | 14:41 | |
| ansmith | odyssey4me: https://github.com/openstack/oslo.messaging/blob/ede61c360d8e35266ed472a37a8e5f8cde7b970d/oslo_messaging/transport.py#L104 | 14:41 |
| openstackgerrit | Merged openstack/ansible-role-python_venv_build stable/rocky: Add option for default pip package install https://review.openstack.org/603929 | 14:42 |
| *** gkadam has joined #openstack-ansible | 14:43 | |
| *** spatel has joined #openstack-ansible | 14:52 | |
| spatel | jamesdenton: I am back, also my client got disconnected | 14:52 |
| *** vnogin has joined #openstack-ansible | 14:53 | |
| *** jonher has quit IRC | 14:53 | |
| jamesdenton | spatel cool. I got SR-IOV setup on this node and have just been trying various testing scenarios. | 14:54 |
| spatel | cool! | 14:54 |
| spatel | waiting for your result | 14:54 |
| spatel | This is what my memory look like on host node: http://paste.openstack.org/show/730545/ | 14:56 |
| odyssey4me | ansmith: oh, so you're saying it's always enabled by default in the code - but if there's no config it uses the rpc settings... so basically enabling/disabling notifications is not actually an option | 14:57 |
| spatel | look like i don't have hugepage configured | 14:57 |
| ansmith | odyssey4me: yes, that is the case...when you have single rabbit backend, no difference in behavior, when you have separate backends notifications end up with rpc intermediary | 14:58 |
| *** vnogin has quit IRC | 14:58 | |
| ansmith | odyssey4me: it shows up as tempest failures, etc. | 14:58 |
| *** aedc_ has quit IRC | 14:58 | |
| *** radeks has quit IRC | 14:58 | |
| *** aedc_ has joined #openstack-ansible | 14:58 | |
| jamesdenton | spatel compiling my results now. They key to better performance for UDP, I'm finding, is making sure jumbo frames are enabled all the way down | 14:58 |
| *** radeks has joined #openstack-ansible | 14:58 | |
| jamesdenton | as far as this testing is concerned, anyway | 14:59 |
| *** dxiri has joined #openstack-ansible | 14:59 | |
| odyssey4me | ansmith: I see, ok - could you update the commit message to explain that so that other reviewers don't get confused. If the code always enables it anyway, then us having a toggle has been a false sense of security so it makes sense to always configure it. | 14:59 |
| spatel | Ok, but in real world UDP jumbofram won't work, our application generate RTP traffic which is 120bytes UDP packets stream | 15:00 |
| ansmith | odyssey4me: yes, will do...I should have provided the details in the first place :) | 15:00 |
| *** Leo_m has joined #openstack-ansible | 15:00 | |
| odyssey4me | ansmith: tyvm, nice to see some more old cruft get removed :) | 15:01 |
| *** radeks has quit IRC | 15:01 | |
| *** radeks has joined #openstack-ansible | 15:01 | |
| odyssey4me | ansmith: please don't forget to port back the driver implementations which didn't make it into rocky in time to ensure that the feature is there for experimental use | 15:02 |
| odyssey4me | I know master and rocky's pip install implementations are quite different now, so they won't be clean picks - apologies for that | 15:02 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-os_gnocchi stable/rocky: Define cache_timeout variable https://review.openstack.org/604379 | 15:03 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-os_gnocchi stable/queens: Define cache_timeout variable https://review.openstack.org/604380 | 15:05 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-os_gnocchi stable/pike: Define cache_timeout variable https://review.openstack.org/604381 | 15:05 |
| *** Leo_m has quit IRC | 15:06 | |
| cjloader | mnaser: ping | 15:07 |
| *** gkadam has quit IRC | 15:10 | |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-os_gnocchi stable/rocky: Remove galera_client meta-dependency https://review.openstack.org/603745 | 15:10 |
| *** Leo_m has joined #openstack-ansible | 15:13 | |
| *** rpittau has quit IRC | 15:21 | |
| *** fghaas has quit IRC | 15:22 | |
| jamesdenton | spatel here are my original results. http://paste.openstack.org/show/730548/. Problem is, i've tweaked a few too many things and am no longer getting those results | 15:28 |
| spatel | hmm | 15:30 |
| spatel | reading.. | 15:30 |
| spatel | your result are much better than mine | 15:31 |
| spatel | are you running vm to vm right? | 15:31 |
| spatel | did you try Qlogic to Qlogic? | 15:32 |
| spatel | i am surprised why i am not getting 4 to 5Gbps throughput ? | 15:33 |
| spatel | I don't have hugepage enabled i think i should try with that | 15:33 |
| spatel | jamesdenton: ^^ | 15:33 |
| spatel | let me download ubuntu image and try | 15:34 |
| jamesdenton | i only have 1 qlogic NIC. i am doing some testing from the SR-IOV VM to the baremetal node hosting the VM now. | 15:37 |
| spatel | what hugepage size you have on host? | 15:37 |
| jamesdenton | From my VM using SR-IOV w/ Qlogic, i can hit a baremetal node running X520 at 9.32 Gbits/sec. I can hit a VM on the same host at 8.20 Gbits/sec. | 15:38 |
| jamesdenton | 1G, only for the DPDK VMs though | 15:38 |
| jamesdenton | and i have jumbo frames enabled everywhere | 15:39 |
| openstackgerrit | Manuel Buil proposed openstack/openstack-ansible master: Unfreeze networking-odl https://review.openstack.org/602243 | 15:39 |
| spatel | Can you try without jumbo frame with 1500 MTU | 15:39 |
| jamesdenton | one sec | 15:39 |
| jamesdenton | spatel 7.26 Gbits/sec between VMs | 15:40 |
| spatel | hmm! still much better | 15:40 |
| spatel | what is limiting my throughput ? | 15:41 |
| spatel | I have 10G switch and both compute node running on same switch | 15:41 |
| *** aedc_ has quit IRC | 15:41 | |
| *** aedc_ has joined #openstack-ansible | 15:41 | |
| jamesdenton | if you are still only getting ~5Gbps between compute nodes, you may need to dig into that first. I dropped MTU to 1500 on my computes and still get 8.5Gbps with iPerf between computes - qlogic to intel | 15:42 |
| spatel | what hardware you are using for compute node? | 15:42 |
| jamesdenton | these are Dl360e G8s w/ E5-2450L (1.8Ghz) | 15:42 |
| spatel | I have 2.9GHz CPU DL 360p G8 | 15:43 |
| spatel | 32 core | 15:43 |
| jamesdenton | this is single threaded, but still, plenty of horsepower | 15:43 |
| spatel | only difference is you have ubuntu and i have CentOS7 | 15:43 |
| jamesdenton | that is a difference. | 15:43 |
| spatel | Let me upgrade kernel to 4.x and both compute node and run test so i can see difference | 15:44 |
| spatel | Does iptables make difference? | 15:45 |
| jamesdenton | i'm not using it | 15:45 |
| spatel | hmm! | 15:45 |
| jamesdenton | it doesn't come into play for sr-iov anyway | 15:45 |
| spatel | when you running compute-to-compute are you using br-XXXX interface to run test? | 15:45 |
| jamesdenton | no | 15:45 |
| spatel | I am running between br-mgmt on both compute node | 15:46 |
| spatel | i wonder if bridge slowing down stuff | 15:46 |
| spatel | let me try configure IP on physical node and see | 15:46 |
| spatel | Let me try all kind of combination and compare result | 15:47 |
| *** Bhujay has quit IRC | 15:48 | |
| *** radeks has quit IRC | 15:50 | |
| *** radeks has joined #openstack-ansible | 15:50 | |
| cjloader | is there a planned release upcoming for rocky? | 15:51 |
| spatel | jamesdenton: [ 3] 0.0-10.0 sec 10.9 GBytes 9.39 Gbits/sec | 15:52 |
| spatel | this is non-bridge result | 15:52 |
| *** skiedude has joined #openstack-ansible | 15:52 | |
| jamesdenton | spatel alright. | 15:53 |
| spatel | I am getting 10G | 15:53 |
| spatel | now going to configure bridge and will see again | 15:53 |
| spatel | hold your horse back :) | 15:53 |
| cjloader | can I get some reviews on https://review.openstack.org/#/c/556586/? | 15:59 |
| cjloader | I updated the spec | 15:59 |
| cloudnull | anyone around want to give this a nudge https://review.openstack.org/#/c/601467/ | 16:14 |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible master: Removed conditional in test setup https://review.openstack.org/575748 | 16:15 |
| cloudnull | mnaser guilhermesp did we work out what all is needed to make the "smart sources" patches go? | 16:16 |
| cloudnull | i remember there was an issue with the paste configs with rdo . | 16:16 |
| *** aedc_ has quit IRC | 16:16 | |
| cloudnull | i dont remember where that landed | 16:16 |
| * cloudnull was circling back on those prs | 16:16 | |
| odyssey4me | hmm, cloudnull I think hwoarang was also recently doing some apparmor things for haproxy re: https://review.openstack.org/#/c/601467/2/tasks/haproxy_post_install.yml | 16:17 |
| *** DanyC has quit IRC | 16:18 | |
| logan- | ya https://review.openstack.org/#/c/603078/ | 16:18 |
| cloudnull | that's something nicolasbock and I ran into at the PTG - if there's a better way already in flight i'll abandon that | 16:19 |
| openstackgerrit | German Eichberger proposed openstack/openstack-ansible-os_octavia master: Fixes too low security group rules quota https://review.openstack.org/600826 | 16:19 |
| cloudnull | ah ha! | 16:19 |
| *** jonher has joined #openstack-ansible | 16:19 | |
| hwoarang | cloudnull: is the haproxy patch you mentioned still needed? | 16:19 |
| hwoarang | i am guessing not | 16:20 |
| cloudnull | nope abandoned | 16:20 |
| hwoarang | i am sorry i did not notice yours before i work on mine | 16:20 |
| *** aedc has joined #openstack-ansible | 16:20 | |
| spatel | cloudnull: ^^ what hugepage size is good 2M or 1G ? | 16:20 |
| spatel | on 32G memory box | 16:20 |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible-haproxy_server stable/rocky: Disable HAProxy apparmor profile if present https://review.openstack.org/604409 | 16:22 |
| cloudnull | nicolasbock ^ | 16:22 |
| *** aedc has quit IRC | 16:22 | |
| cloudnull | hwoarang all good | 16:22 |
| cloudnull | your patches was better anyways :) | 16:22 |
| *** aedc has joined #openstack-ansible | 16:23 | |
| odyssey4me | cloudnull: that apparmor fix is suse only - does that meet your needs? | 16:24 |
| odyssey4me | if not, perhaps a patch to master will be needed to broaden it | 16:24 |
| hwoarang | yeah only suse ships an haproxy profile | 16:26 |
| cloudnull | spatel 1G is plenty | 16:26 |
| cloudnull | 2M is generally default | 16:26 |
| openstackgerrit | Cam J. Loader proposed openstack/openstack-ansible-specs master: create Designate back end provider spec https://review.openstack.org/556586 | 16:31 |
| openstackgerrit | Merged openstack/openstack-ansible-ops master: Add playbook to automatically refresh fields in kibana https://review.openstack.org/603517 | 16:31 |
| openstackgerrit | Merged openstack/openstack-ansible-ops master: Add more optionality when customizing node roles https://review.openstack.org/603526 | 16:31 |
| spatel | cloudnull: thanks!!! so smaller is better or bigger? | 16:34 |
| spatel | or depend on workload | 16:34 |
| guilhermesp | cloudnull: ah I think was the neutron api-paste.ini issue a while ago? do you have thse patches? | 16:35 |
| guilhermesp | I think the fix was already merged | 16:35 |
| guilhermesp | i think was this https://github.com/openstack/openstack-ansible-os_neutron/commit/1dd266d35a44ff946a9f50ec63f7ee9be865884a ? | 16:36 |
| guilhermesp | I remember that for me was a really ugly workaround, but we had no choise at that time. Can be redone in the future if needed, but for I think it is working | 16:37 |
| cloudnull | cool! I'll make sure to try and replicate the pattern in our other patches just to make sure its consistent | 16:38 |
| cloudnull | I just wanted to make sure be for I rebase all these things | 16:39 |
| guilhermesp | I think you can rebase right now... I'm just searching the PRs to give you that confirmation | 16:39 |
| guilhermesp | but as the commit is already upstream, I think we're good with that | 16:39 |
| cloudnull | spatel depends generally depends on the workloads - also in order to enable hugepages in the vms you might need to add meta-data to your flavors/images. | 16:40 |
| cloudnull | excellent ! | 16:40 |
| cloudnull | guilhermesp ^ | 16:40 |
| spatel | yes i am going to add in flavor | 16:40 |
| *** ming has joined #openstack-ansible | 16:40 | |
| guilhermesp | cloudnull: https://review.openstack.org/#/c/595796/ https://review.openstack.org/#/c/599468/ | 16:41 |
| *** DanyC has joined #openstack-ansible | 16:41 | |
| cloudnull | nice! tyvm | 16:41 |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible-tests master: Convert test network setup to use the networkd role https://review.openstack.org/592596 | 16:43 |
| *** radeks has quit IRC | 16:44 | |
| openstackgerrit | Markos Chandras (hwoarang) proposed openstack/openstack-ansible-lxc_hosts master: tasks: apparmor: Disable the HAproxy profile on all hosts. https://review.openstack.org/604420 | 16:45 |
| *** SimAloo has quit IRC | 16:45 | |
| hwoarang | cloudnull: incidentally haproxy needs one more fix ^^^ | 16:45 |
| hwoarang | this is getting painful | 16:45 |
| openstackgerrit | Markos Chandras (hwoarang) proposed openstack/openstack-ansible-os_nova master: SUSE: Add support for openSUSE Leap 15 https://review.openstack.org/604080 | 16:46 |
| *** DanyC has quit IRC | 16:46 | |
| *** Leo_m has quit IRC | 16:47 | |
| *** Leo_m_ has joined #openstack-ansible | 16:48 | |
| cloudnull | hwoarang lgtm | 16:48 |
| cloudnull | apparmor, like selinux, is a pain. | 16:49 |
| hwoarang | one day we will just stop the service and be done with it ;p | 16:50 |
| hwoarang | ok it's late here. time to go. have a great weekend everybody | 16:50 |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible-os_keystone master: Cleanup files and templates using smart sources https://review.openstack.org/588960 | 16:50 |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible-os_glance master: Cleanup files and templates using smart sources https://review.openstack.org/588959 | 16:52 |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible-os_heat master: Cleanup files and templates using smart sources https://review.openstack.org/588954 | 16:52 |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible-os_nova master: Cleanup files and templates using smart sources https://review.openstack.org/588951 | 16:53 |
| *** SimAloo has joined #openstack-ansible | 16:58 | |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible-os_cinder master: Cleanup files and templates using smart sources https://review.openstack.org/588953 | 16:58 |
| cloudnull | hwoarang have a great weekend | 16:59 |
| *** gyee has joined #openstack-ansible | 17:01 | |
| *** gkadam has joined #openstack-ansible | 17:03 | |
| *** nicolasbock_ has joined #openstack-ansible | 17:03 | |
| cloudnull | if folks have a moment to give https://review.openstack.org/#/q/topic:smart-sources+(status:open) a review I'd appreciate it | 17:04 |
| *** spsurya has quit IRC | 17:05 | |
| *** ming has quit IRC | 17:10 | |
| *** hamzaachi has joined #openstack-ansible | 17:12 | |
| spotz | cloudnull: I have it open was just waiting to see if it got past the gates:) | 17:18 |
| odyssey4me | cloudnull guilhermesp here are the patches that followed the smart sources in os_neutron https://review.openstack.org/591463 & https://review.openstack.org/595796 & https://review.openstack.org/599468 | 17:18 |
| *** SimAloo has quit IRC | 17:20 | |
| cloudnull | odyssey4me spotz thanks! i need to confirm if that paste change is needed in other rdo package installed services. | 17:22 |
| *** dxiri has quit IRC | 17:26 | |
| *** dxiri has joined #openstack-ansible | 17:27 | |
| *** SimAloo has joined #openstack-ansible | 17:36 | |
| *** cshen has joined #openstack-ansible | 17:42 | |
| *** ppetit has quit IRC | 17:46 | |
| openstackgerrit | Merged openstack/openstack-ansible-galera_server master: Allow configuration of startup timeout https://review.openstack.org/603847 | 17:56 |
| jamesdenton | If i have a variable defined in user_variables.yml, is there a way to override that on a per-host basis? Or does user_variables trump everything? If so, and I need to tweak per-host, would defining in openstack_user_config for each host be the way to go? | 17:57 |
| *** mmercer has joined #openstack-ansible | 18:01 | |
| openstackgerrit | Merged openstack/openstack-ansible-os_gnocchi stable/queens: Define cache_timeout variable https://review.openstack.org/604380 | 18:02 |
| *** noonedeadpunk has quit IRC | 18:03 | |
| *** clone99 has joined #openstack-ansible | 18:03 | |
| openstackgerrit | Merged openstack/openstack-ansible-os_gnocchi stable/rocky: Define cache_timeout variable https://review.openstack.org/604379 | 18:04 |
| openstackgerrit | Merged openstack/openstack-ansible-os_gnocchi stable/rocky: Remove galera_client meta-dependency https://review.openstack.org/603745 | 18:04 |
| jamesdenton | n/m i think i answered my own question. | 18:05 |
| *** SimAloo has quit IRC | 18:06 | |
| cloudnull | jamesdenton i was getting lunch. did you make it go ? | 18:08 |
| jamesdenton | yes - i just moved my var to openstack_user_config and defined it accordingly on each host | 18:09 |
| openstackgerrit | Cam J. Loader proposed openstack/openstack-ansible-specs master: create Designate back end provider spec https://review.openstack.org/556586 | 18:10 |
| *** jonher_ has joined #openstack-ansible | 18:11 | |
| *** nicolasbock_ has quit IRC | 18:12 | |
| *** jonher has quit IRC | 18:13 | |
| *** jonher_ is now known as jonher | 18:13 | |
| openstackgerrit | Cam J. Loader proposed openstack/openstack-ansible-specs master: create Designate back end provider spec https://review.openstack.org/556586 | 18:19 |
| cloudnull | jamesdenton ++ thats how i would have done it | 18:23 |
| jamesdenton | i'm learning. thanks :D | 18:27 |
| *** cshen has quit IRC | 18:33 | |
| mgariepy | logan-, are you there? | 18:34 |
| mgariepy | is there a way to have quota on calico networks ? | 18:35 |
| *** cshen has joined #openstack-ansible | 18:36 | |
| mgariepy | logan-, nm. thanks | 18:42 |
| logan- | i am now | 18:45 |
| cjloader | cloudnull: mind taking a peek at https://review.openstack.org/#/c/556586/? | 18:46 |
| cloudnull | cjloader sure | 18:48 |
| openstackgerrit | Nicolas Bock proposed openstack/openstack-ansible-openstack_hosts master: Do not uninstall grub2 https://review.openstack.org/602482 | 18:49 |
| openstackgerrit | Merged openstack/openstack-ansible-lxc_hosts master: tasks: apparmor: Disable the HAproxy profile on all hosts. https://review.openstack.org/604420 | 18:53 |
| nicolasbock | Thanks cloudnull and hwoarang ! | 18:59 |
| nicolasbock | I'll give that a try | 19:00 |
| jamesdenton | cloudnull when developing across platforms (e.g. centos, ubuntu, suse), do you tend to let the gating find the inconsistencies or do you run VMs of each to test? | 19:02 |
| *** mandeer has joined #openstack-ansible | 19:02 | |
| *** dave-mccowan has quit IRC | 19:03 | |
| cloudnull | i tend to let the gating find the issues | 19:03 |
| cloudnull | if it cant then i spin up vms of whatever i need | 19:03 |
| jamesdenton | gotcha | 19:03 |
| odyssey4me | I personally build on one until it works, then push the patch up... ^ what he said | 19:03 |
| jamesdenton | oh hey odyssey4me. yeah i like that plamn | 19:03 |
| jamesdenton | *plan | 19:03 |
| openstackgerrit | James Denton proposed openstack/openstack-ansible-os_neutron master: Adds DPDK support for Open vSwitch https://review.openstack.org/590810 | 19:04 |
| jamesdenton | Is there a way to push patchsets and avoid gating? But to avoid losing the work? | 19:06 |
| odyssey4me | jamesdenton: don't be shy - zuul doesn't mind doing the work | 19:06 |
| mandeer | howdy fellas. I got through an openstack-ansible deploy on a centos 7 test env, and a few more questions. | 19:06 |
| jamesdenton | just looking out for my fellow stackers, trying to save resources on likely failures :D | 19:07 |
| mandeer | in horizon, when i go to "orchestration" pages, i get an error saying it can't load the stack | 19:07 |
| odyssey4me | jamesdenton: if I'm working on something that takes a while, I usually push the review up and just mark it -w so that everyone knows it's not ready yet | 19:07 |
| mandeer | and it logs : <Message>The request signature does not conform to AWS standards</Message><Code>IncompleteSignature</Code><Type>Sender</Type> | 19:07 |
| mandeer | is that expected behavior? | 19:07 |
| odyssey4me | sometimes I get paranoid and push any changes up every hour or so | 19:07 |
| jamesdenton | how do you mark it -w? like in a comment? or somewhere else? | 19:07 |
| cloudnull | mandeer that is not expected behavior | 19:08 |
| cloudnull | is that on master? | 19:08 |
| odyssey4me | hit 'reply', then '-1' in the workflow row | 19:08 |
| jamesdenton | oh right on. thanks! | 19:08 |
| odyssey4me | mandeer: that's an odd message, what exactly was it that logged that? | 19:09 |
| mandeer | i specified 18.0.0.0rc2, i think per the rocky docs | 19:09 |
| mandeer | it's coming up /openstack/log/infra_horizon_container-51b92ba3/horizon/horizon-error.log | 19:10 |
| mandeer | or rather that's where it's logging to | 19:10 |
| mandeer | the full line is [Fri Sep 21 19:03:51.671124 2018] [:error] [pid 2438] Recoverable error: ERROR: <ErrorResponse><Error><Message>The request signature does not conform to AWS standards</Message><Code>IncompleteSignature</Code><Type>Sender</Type></Error></ErrorResponse> | 19:10 |
| cloudnull | your seeing that error in the UI or just noise in the logs ? | 19:10 |
| mandeer | the error in the UI is "unable to retrieve stack list" | 19:11 |
| mandeer | or "unable to retrieve stack resource types" or "unable to retrieve template versions" respectively for each page | 19:12 |
| mandeer | under Orchest > Template Gen, i get the following error in the UI | 19:12 |
| mandeer | Unable to retrieve resources Template versions: <ErrorResponse><Error><Message>The request signature does not conform to AWS standards</Message><Code>IncompleteSignature</Code><Type>Sender</Type></Error></ErrorResponse>. Qos policies: The resource could not be found. Neutron server returns request_ids: ['req-afbd9611-1b6c-464c-a69d-84c789be6406']. | 19:13 |
| mandeer | is it a networking issue between nodes? | 19:13 |
| cloudnull | while we test heat, ive not personally used horizon with it . | 19:14 |
| cloudnull | maybe something is not quite right in rocky and horizon atm | 19:14 |
| mandeer | i had some network issues during the setup and deployment process, i may blow it all out and start over on the off chance that elements of my net config got stuck | 19:14 |
| *** dave-mccowan has joined #openstack-ansible | 19:14 | |
| mandeer | would you say i'd have a better shot with queens at the moment? | 19:14 |
| cloudnull | that could be | 19:14 |
| jamesdenton | maybe it wants the neutron 'qos' extension loaded? | 19:15 |
| cloudnull | that could be ^ | 19:15 |
| mandeer | ok, i'll look into that. | 19:16 |
| jamesdenton | You could try adding the overrides here: http://paste.openstack.org/show/730563/ | 19:16 |
| mandeer | thank you both | 19:16 |
| cloudnull | totally could be something we need to enable by default when heat is present | 19:16 |
| *** shardy has quit IRC | 19:16 | |
| cloudnull | which would be something new | 19:17 |
| jamesdenton | could be | 19:17 |
| *** shardy has joined #openstack-ansible | 19:17 | |
| cloudnull | poking around it would seem some of the heat templates have been updated to use QOS and LBaaS by default | 19:20 |
| jamesdenton | weeeeeeee | 19:23 |
| openstackgerrit | Nicolas Bock proposed openstack/openstack-ansible-openstack_hosts master: Do not uninstall grub2 https://review.openstack.org/602482 | 19:27 |
| openstackgerrit | Cam J. Loader proposed openstack/openstack-ansible-specs master: create Designate back end provider spec https://review.openstack.org/556586 | 19:29 |
| *** joabdearaujo has joined #openstack-ansible | 19:30 | |
| *** skiedude has quit IRC | 19:32 | |
| *** gkadam has quit IRC | 19:35 | |
| *** radeks has joined #openstack-ansible | 19:45 | |
| *** radeks_ has joined #openstack-ansible | 19:46 | |
| openstackgerrit | Merged openstack/openstack-ansible-haproxy_server stable/rocky: Disable HAProxy apparmor profile if present https://review.openstack.org/604409 | 19:48 |
| *** radeks has quit IRC | 19:50 | |
| guilhermesp | cloudnull: did you manage to rebase the smart-sources? | 19:54 |
| *** aedc has quit IRC | 20:21 | |
| *** aedc has joined #openstack-ansible | 20:21 | |
| spotz | cloudnull: The review I've bene waiting to go through the gates just finally went through( | 20:28 |
| *** DanyC has joined #openstack-ansible | 20:30 | |
| openstackgerrit | Andy Smith proposed openstack/openstack-ansible-os_nova master: Update messaging notification configuration https://review.openstack.org/604107 | 20:35 |
| *** phasespace has joined #openstack-ansible | 20:37 | |
| *** radeks_ has quit IRC | 20:38 | |
| openstackgerrit | Andy Smith proposed openstack/openstack-ansible-os_neutron master: Update messaging notification configuration https://review.openstack.org/604117 | 20:44 |
| *** clone99 has quit IRC | 20:45 | |
| *** cshen has quit IRC | 20:47 | |
| prometheanfire | johnsom: you still using that gentoo musl image for testing? | 20:49 |
| johnsom | prometheanfire Was I ever? | 20:49 |
| prometheanfire | johnsom: I thought octavia was | 20:50 |
| johnsom | prometheanfire We never supported gentoo. I think rm_work was considering it, but ran into trouble and never finished working on it | 20:50 |
| prometheanfire | maybe that's it | 20:51 |
| mnaser | so are we dropping support to all distros and doing gentoo only | 20:51 |
| *** spatel has quit IRC | 20:51 | |
| johnsom | Ha, if prometheanfire had his way.... | 20:51 |
| prometheanfire | mnaser: yes | 20:51 |
| mnaser | OSA appliance | 20:52 |
| prometheanfire | ya, that's typically what people use it for (alliance type installs) | 20:52 |
| * prometheanfire is still working though infra ansible making gentoo happy | 20:52 | |
| mnaser | i'm more of an alliance type installs, not a big horde type installs tbh | 20:53 |
| mnaser | i wonder if there's a distro that just has bleeding edge everything | 20:53 |
| mnaser | latest kernel, latest qemu, latest libvirt, etc | 20:54 |
| mnaser | it'd be a super cool platform tbh. | 20:54 |
| johnsom | I just want one that doesn't feel the need to have EVERY kernel module included.... | 20:54 |
| mnaser | yes | 20:54 |
| mnaser | that too | 20:54 |
| mnaser | something literally designed for hypervisors only | 20:55 |
| spotz | We're going to gentoo? | 20:55 |
| mnaser | cloudnull: wild question, why ubuntu over debian | 21:00 |
| mnaser | i know this was a decision likely taken really long time ago | 21:00 |
| mnaser | clearlinux could be cool | 21:02 |
| spotz | Kolla is doing oracle linux I thought a strange choice | 21:04 |
| *** cshen has joined #openstack-ansible | 21:05 | |
| openstackgerrit | Markos Chandras (hwoarang) proposed openstack/openstack-ansible-lxc_hosts master: tasks: lxc_apparmor: Fix apparmor profile name for HAproxy https://review.openstack.org/604473 | 21:05 |
| openstackgerrit | Markos Chandras (hwoarang) proposed openstack/openstack-ansible-os_nova master: SUSE: Add support for openSUSE Leap 15 https://review.openstack.org/604080 | 21:06 |
| logan- | seems weird to have that in lxc_hosts hwoarang? i thought it was in os_neutron and haproxy_server | 21:06 |
| logan- | (re: 604473) | 21:06 |
| *** DanyC has quit IRC | 21:06 | |
| *** mmercer has quit IRC | 21:07 | |
| logan- | oh, I see the one we merged for neutron earlier was in lxc_hosts | 21:07 |
| logan- | that seems wrong to me | 21:07 |
| hwoarang | logan-: so if you see here http://logs.openstack.org/80/604080/3/check/openstack-ansible-functional-distro_install-opensuse-150/1a75868/logs/suse-zypper-list-installed-host.txt.gz something is pulling haproxy on the host but it is not haproxy_server role. maybe a depenency of something else. so we end up having haproxy and its apparmor profile on all lxc_host nodes | 21:11 |
| *** DanyC has joined #openstack-ansible | 21:11 | |
| hwoarang | so we need to manage it there too | 21:11 |
| logan- | oh wow | 21:11 |
| logan- | what a mess, we're pulling haproxy in on every host? | 21:11 |
| hwoarang | seems like haproxy is needed by neutron | 21:11 |
| hwoarang | at least... | 21:11 |
| logan- | oh, yes it is | 21:12 |
| hwoarang | so maybe other services with metadata agents need it too | 21:12 |
| hwoarang | to intercept the metadata traffic | 21:12 |
| hwoarang | but you can't run apparmor on a container (afaik). and even if you do, the host needs to have a matching configuration because even if you allow haproxy on the container, the host will block it | 21:12 |
| hwoarang | it just feels easier to do that on lxc_host if it happens to have haproxy there to make sure that any future containers will be fine | 21:13 |
| hwoarang | similar to the dnsmasq which is not really a lxc_host issue but you want that in case you have neutron containers on that host | 21:14 |
| mnaser | neutron agent needs haproxy | 21:14 |
| mnaser | to replace neutron metadata proxy | 21:14 |
| hwoarang | it almost feels like it might just be easier to disable apparmor on all hosts and deal with it | 21:14 |
| *** shardy has quit IRC | 21:15 | |
| *** shardy has joined #openstack-ansible | 21:15 | |
| logan- | is there any way we can get the suse packaging issue fixed? i can't find any apparmor profile in vbernat, debian, ubuntu, etc packaging.. I think it is some suse specific thing | 21:15 |
| logan- | also i think we should just address the neutron issue in the neutron role. I think that is the only place haproxy is used besides haproxy_server? | 21:16 |
| logan- | the thing we have now will break again once nspawn_hosts is the default | 21:17 |
| hwoarang | logan-: it's the same reason we disable the dnsmasq and ping profiles in lxc_host | 21:17 |
| hwoarang | you can't fix it just in the container. the real host needs to have it fixed too. because haproxy is a process in a container but the real host sees it too | 21:17 |
| hwoarang | and blocks it. same with ping and dnsmasq | 21:17 |
| logan- | so delegate_to: "{{ physical_host }}" if is_metal is not true? | 21:18 |
| hwoarang | that could work yes. | 21:18 |
| logan- | didn't we move all of the neutron agents to metal anyway? | 21:18 |
| hwoarang | can't remember. but i can look into moving the apparmor stuff from lxc_hosts to os_neutron on monday if that helps | 21:19 |
| logan- | it is easy to support both though, but the appropriate place is os_neutron since this is not lxc specific, it is neutron specific and will break with other container techs, so yeah i think we need to move it | 21:20 |
| logan- | also this current fix won't work on greenfield even with 604473 because lxc_hosts runs long before haproxy is installed by os_neutron here https://git.openstack.org/cgit/openstack/openstack-ansible-os_neutron/tree/vars/ubuntu.yml#n113 .. the stat.exists test will always fail, so we need to make sure this is inserted in os_neutron after the distro package installs take place | 21:24 |
| logan- | i don't think there's any need to cater to container vs not container, since if neutron agents are running containerized and install haproxy, the profile will be a noop w/ apparmor being disabled in containers | 21:26 |
| cloudnull | mnaser no reason in particular | 21:36 |
| cloudnull | a very long time ago ported OSA to debian https://github.com/cloudnull/os-ansible-deploymenT/tree/debian-compat | 21:36 |
| cloudnull | however there just wasn't a lot of interest in using debian instead of ubuntu so I dropped it | 21:37 |
| *** ivve has quit IRC | 21:38 | |
| cloudnull | today the work to get debian supported in osa would be fairly minimal (i suspect) given all of the multi-os things we have today vs when I did the debian compat work all that long ago | 21:38 |
| cloudnull | hwoarang if we need to disable apparmor to make things go I'd support that | 21:46 |
| cloudnull | though itd be sad to see it go (similar to selinux) | 21:46 |
| *** DanyC has quit IRC | 21:56 | |
| *** hamzaachi has quit IRC | 21:59 | |
| odyssey4me | cloudnull: I think the difference with selinux was that mhayden said it was all container focused, and didn't work for us - vs apparmor which is more agnostic, and does - largely | 22:01 |
| cloudnull | ++ | 22:04 |
| *** dave-mccowan has quit IRC | 22:04 | |
| openstackgerrit | Merged openstack/openstack-ansible-os_cinder stable/queens: Remove galera_client meta-dependency https://review.openstack.org/603715 | 22:11 |
| *** aedc has quit IRC | 22:11 | |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-repo_build stable/rocky: Remove galera_client meta-dependency https://review.openstack.org/604489 | 22:13 |
| openstackgerrit | Merged openstack/openstack-ansible-os_horizon stable/rocky: Remove galera_client meta-dependency https://review.openstack.org/603756 | 22:19 |
| *** cshen has quit IRC | 22:33 | |
| *** cshen has joined #openstack-ansible | 22:33 | |
| *** cshen has quit IRC | 22:38 | |
| *** Leo_m has joined #openstack-ansible | 22:45 | |
| *** Leo_m_ has quit IRC | 22:47 | |
| *** KeithMnemonic has quit IRC | 22:58 | |
| *** nsmeds has joined #openstack-ansible | 22:58 | |
| *** tosky has quit IRC | 22:58 | |
| *** KeithMnemonic has joined #openstack-ansible | 22:59 | |
| *** tosky has joined #openstack-ansible | 22:59 | |
| nsmeds | question about LXC vs nspawn - with LXC containers I can simply `lxc-attach` into running container | 22:59 |
| nsmeds | however, with nspawn attempts to `machinectl login` require user/pass | 22:59 |
| nsmeds | anyway to bypass this and just get a shell in running container? | 22:59 |
| odyssey4me | nsmeds: machinectl shell bash IIRC | 23:00 |
| nsmeds | oh, well that was easy | 23:01 |
| nsmeds | much appreciated | 23:01 |
| odyssey4me | nsmeds: :) | 23:01 |
| *** tosky has quit IRC | 23:13 | |
| *** spatel has joined #openstack-ansible | 23:13 | |
| *** spatel has quit IRC | 23:18 | |
| *** mschuppert has joined #openstack-ansible | 23:36 | |
| *** FrankZhang has quit IRC | 23:36 | |
| *** FrankZhang has joined #openstack-ansible | 23:37 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!