15:01:01 #startmeeting ironic 15:01:02 Meeting started Mon Mar 18 15:01:01 2019 UTC and is due to finish in 60 minutes. The chair is TheJulia. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:01:02 o/ 15:01:03 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:01:05 o/ 15:01:06 The meeting name has been set to 'ironic' 15:01:09 o/ 15:01:13 o/ 15:01:15 o/ 15:01:33 o/ 15:01:39 o/ 15:01:48 \o 15:02:16 o/ 15:02:19 o/ 15:02:28 Our agenda is a little short this week on the wiki, given this is essentially release week for ironic, hopefully this meeting will be short. :) 15:02:36 \o 15:02:48 \o 15:02:48 #link https://wiki.openstack.org/wiki/Meetings/Ironic 15:03:00 o/ 15:03:06 #topic Announcements / Reminders 15:03:50 As previously mentioned, we're working to cut our Stein stable branches early this week. If there are any blockers that anyone is aware of, please raise that visibility as soon as possible. 15:04:06 #info Ironic to cut stable/stein this week. 15:04:54 There is a call for help to address issues with Bionic. We'll circle back to this, but we need reviews if there are any explicit remaining items for it. 15:05:54 right now, IIUC, the multinode grenade job, the UEFI job and the postgres/PXE job do not pass 15:06:06 correct 15:06:07 #link https://review.openstack.org/#/c/642015/ 15:06:08 patch 642015 - ironic - Move all testing to Bionic - 2 patch sets 15:06:36 #info the PTG discussion topic etherpad is still available. Some forum sessions were submitted and approved. We do not have an explicit ops feedback session this time. Ops focused contributors are welcome to engage with the ironic community during our PTG meetings. 15:06:38 we have te problem with sushy with py3 in bionic too 15:06:51 https://review.openstack.org/#/c/591828/ 15:06:52 patch 591828 - sushy - Change sushy devstack job to python3 - 4 patch sets 15:07:03 I have a question on the bionic thing, but can come back in open discussion if folks would rather 15:07:18 #link https://etherpad.openstack.org/p/DEN-train-ironic-brainstorming 15:07:35 Does anyone have anything they would like to announce or remind us of? 15:09:04 I'm taking silence as a no. If there is no objection, we'll jump directly to Discussion/Open Discussion with first Bionic and then any remaining py3 needs. 15:09:38 * iurygregory that was fast XD 15:09:52 :) 15:10:16 #topic Discussion 15:10:21 Bionic! 15:10:25 jroll: you had a question 15:10:26 yeah, so 15:10:42 I was under the impression that march 15, a patch would land that would force legacy jobs to bionic 15:10:59 but I assume our gate isn't broken, so maybe I missed something? or maybe we're doing something unexpected? 15:11:01 I think it did happen 15:11:11 but the legacy jobs and the UEFI job hardcode xenial 15:11:23 the postgres job went down, but it's non-voting 15:11:26 oh duh, ok 15:11:32 XD 15:11:33 the unit tests had problems, but they got quickly fixed 15:11:52 There were a couple larger community wide problems, but those were also fixed very quickly 15:12:00 yeah some jobs we have the node set to xenial, most are ok to switch 15:12:01 cool, so we need to fix these things ASAP (because stein release), but we aren't broken 15:12:08 yeah, so we're running bionic where we don't set xenial explicitly 15:12:24 awesome. thanks for getting me back into the loop I fell out of :) 15:12:34 jroll: last think I heard was that fixing the postgres job requires a new feature in neutron.. 15:12:37 * thing 15:12:39 and since we moved most of things to zuulv3 we dont have legacy jobs and by default they are bionic 15:12:52 iurygregory: grenades are legacy 15:12:56 this is still pending though https://review.openstack.org/642315 15:12:57 patch 642315 - ironic - Migrate ironic-grenade-dsvm-multinode-multitenant ... - 2 patch sets 15:13:04 or at least this ^^ one 15:13:05 dtantsur: ugh. I'm not as concerned about fixing that one, but we should 15:13:16 yeah grenades cant be moved to zuulv3 at least according to documentation 15:13:22 jroll: it's the only job covering PXE (as opposed to iPXE) 15:13:36 for postgres, we should at least revert to xenial https://review.openstack.org/642430 15:13:37 patch 642430 - ironic - Pinning postgres job to xenial - 8 patch sets 15:13:44 iurygregory: we can try moving them to bionic even without moving them to zuul v3 15:14:07 yeap 15:14:15 then 2 patches are needed, one to fix the missing package https://review.openstack.org/642507 15:14:15 patch 642507 - ironic - Update dist filter for devstack ubuntu - 1 patch set 15:14:16 dtantsur: right, which is why I'm not saying "just drop it", it just seems less important than uefi/multinode 15:14:17 Is it a reasonable test that someone would install rocky on bionic, and then upgrade to stein? 15:14:40 I wouldn't be surprised if somebody did 15:14:46 just to keep the two upgrades distinct 15:14:49 but I don't think we supported bionic back then 15:15:01 so the more reasonable path is rocky -> stein -> bionic 15:15:12 (because stein was tested with xenial at least to some extent) 15:15:12 what dtantsur said, so I'm actually less worried about grenade 15:15:36 but, jroll, your making a good point, I just don't think we have the resources to cover every case either 15:15:50 fair nuff 15:16:19 I feel like there is going to be a larger grenade discussion, or maybe not... at the ptg 15:16:36 feels kinda late :P 15:16:41 Ilya Etingof proposed openstack/ironic master: Explicitly start `vbmcd` supervisor https://review.openstack.org/637971 15:16:46 Anyway, well yeah. 15:16:54 jroll: maintainership is the other issue 15:17:38 it seems that infra will have to keep xenial floating for stein lifetime, at least for grenades 15:17:38 so if we try and migrate grenade stuffs this next cycle, that just leaves us the uefi job right? 15:17:48 TheJulia, dtantsur, the patch on neutron side to fix postgres (and not only that) on bionic is just a PoC for now 15:17:49 TheJulia: and PXE/postgres 15:18:10 does anyone have a link explaining why the pxe job needs a neutron feature? 15:18:14 rpittau: do you ahve a ink handy? 15:18:17 that's very confusing 15:18:26 https://review.openstack.org/643570 15:18:27 patch 643570 - neutron - WiP/PoC - Router Namespace - Connection Tracker he... - 1 patch set 15:18:35 oh jeeze that 15:18:55 there may be a way to do that with some hack until it's merged 15:19:19 should we raise this as a transition blocker? 15:19:20 The last thing I saw was a desire to discuss at the PTG... 15:19:30 ... short term it feels like we could turn on nf_conntrack_helper 15:19:46 I think Derek was looking into it, but he seems offline 15:20:00 if we get blocked on the neutron patch, we certainly miss stein with it 15:20:02 jroll: only set on module load.... 15:20:03 jroll I think that requires a reboot 15:20:15 the module may be able to be unloaded and reloaded, fwiw 15:20:22 worst case, put it in the base devstack image 15:20:43 I suspect that is going to be refused 15:20:49 or never reviewed 15:20:53 we can create the rule with a script, get the route namespace and add it, in some way..... 15:22:16 Lets do this. Raise it as a scenario blocker, document it in a release note. Try to fix it during the next cycle. It doesn't sound like postgres, but PXE that is taking the scenario 15:22:17 is the UEFI problem the same or something else? 15:22:36 TheJulia: we can keep testing it on xenial, rpittau has a patch for it 15:22:41 dtantsur: I _think_ it is files moving around... 15:22:41 s/can/must/? 15:22:58 xenial works fine with the pxelinux pkg installed 15:23:21 pxelinux for libvirt's usage? 15:23:32 dtantsur: must could work, we need to document it as a known issue 15:24:25 ideally, we should update our docs with how to avoid the conntrack problem on bionic 15:24:34 since it's really just a new installation procedure for PXE boot 15:24:54 dtantsur: yeah, we don't need docs at this moment, they can get backported too 15:25:03 release note is slightly more important 15:25:20 probably. although it's not a bug we can ever fix. 15:25:27 (if I get it right?) 15:25:49 that's not even a bug, it's "security enhancement" xD 15:25:55 \o/ 15:26:47 so uefi, yeah is libvirt fun 15:27:05 the actual loader has a very limited text output 15:27:18 heh, joy 15:27:27 * dtantsur knows the top priority for this week now :) 15:28:15 I think that means the scenario was broken by the libvirt packages and likely can be fixed, but it doesn't mean the scenario is just completely broken. That being said, I think we can proceed to release with no additional note for uefi 15:28:35 Can I get a volunteer to write a small known issue release note regarding the security enhancement in bionic? 15:28:48 Julia Kreger proposed openstack/ironic master: Train: Bump semver for 12.2.0 to be master https://review.openstack.org/644246 15:29:15 TheJulia, I can take care of that 15:29:20 Julia Kreger proposed openstack/ironic master: Train: Bump semver to 12.2.0 for master https://review.openstack.org/644246 15:29:24 rpittau: thanks! 15:29:36 Are we in agreement, and is there anything else pressing to discuss today? 15:29:48 TheJulia: priorities for the week, if we have any? 15:29:58 dtantsur: none really, just get the release out I think 15:30:15 * dtantsur imagines punching a bear out of the door 15:30:20 lol 15:30:34 so can we get some artwork for this?!? 15:30:35 lol 15:30:43 #topic Open Discussion 15:31:52 cri.. 15:32:03 .. 15:32:06 ... 15:32:08 ..ckets? 15:32:16 love me some crickets 15:33:41 \o/ 15:33:48 In that case, thanks everyone! Time to go make more coffee! 15:34:03 time for more coffee and more meetings yay 15:34:36 oh yeah I have time for proper coffee \o/ 15:34:42 thanks TheJulia :) 15:35:05 tks 15:35:06 Thanks everyone! 15:35:10 #endmeeting