Friday, 2017-06-16

*** thorst has quit IRC00:47
*** esberglu has quit IRC01:12
*** jwcroppe has joined #openstack-powervm02:11
*** thorst has joined #openstack-powervm04:38
*** thorst has quit IRC04:46
*** BorD__ has quit IRC05:00
*** tjakobs has joined #openstack-powervm05:13
*** thorst has joined #openstack-powervm06:01
*** thorst has quit IRC06:05
*** tjakobs has quit IRC06:26
openstackgerritVu Cong Tuan proposed openstack/ceilometer-powervm master: Fix html_last_updated_fmt for Python3
*** thorst has joined #openstack-powervm08:02
*** thorst has quit IRC08:06
*** k0da has joined #openstack-powervm09:10
*** thorst has joined #openstack-powervm10:03
*** thorst has quit IRC10:07
*** thorst has joined #openstack-powervm10:55
*** thorst has quit IRC10:58
*** thorst has joined #openstack-powervm10:58
*** thorst has quit IRC11:02
*** smatzek has joined #openstack-powervm11:51
*** edmondsw has joined #openstack-powervm12:01
*** BorD_ has joined #openstack-powervm12:01
*** thorst has joined #openstack-powervm12:21
*** kylek3h has joined #openstack-powervm12:52
*** dwayne has joined #openstack-powervm12:59
efriededmondsw Can I get a fresh look at 5453?13:02
edmondswefried sure, one sec13:02
efriedNo real hurry, just been out there a couple days.13:03
*** jpasqualetto has joined #openstack-powervm13:05
*** jwcroppe has quit IRC13:07
*** jwcroppe has joined #openstack-powervm13:08
*** esberglu has joined #openstack-powervm13:08
*** jwcroppe has quit IRC13:12
efriedesberglu looks like it failed two tests; but the results & logs aren't there.13:26
*** tjakobs has joined #openstack-powervm13:26
esbergluefried: Yeah I saw that. It says logs/ doesn't exist13:26
esbergluWe define the logs path only as logs/*.gz in the jenkins job config13:26
esbergluI tried changing it to the full path to see if that's what is needed13:26
efriedOther weird CI failures abound since last night.  I'll leave you to it.13:27
*** jwcroppe has joined #openstack-powervm13:28
*** mdrabe has joined #openstack-powervm13:29
*** dwayne has quit IRC13:31
esbergluefried: See 5469 for a couple quick fixes13:39
esbergluI also put up a patch that should allow us to patch to any openstack repo 546813:40
esbergluGonna try that out once I get the log publishing fixed13:40
*** jwcroppe has quit IRC13:55
*** jwcroppe has joined #openstack-powervm13:55
efriedesberglu Regarding 5469, edmondsw had some questions as to whether this test was really supposed to be related to security_compliance13:59
*** jwcroppe has quit IRC14:00
efriedesberglu But if so, I would like to get that tempest bug opened and include the bug number in these comments and commit msg.14:01
efriedesberglu See eavesdrop from last night:
esbergluefried: Yeah that's on my list of things to explore today. My thinking was that since right now it says it's disabled we should at least actually have it disabled14:02
efriedWell, it's already disabled.14:02
efriedOh, does having the name instead of the UUID not actually work anymore?14:02
esbergluefried: I'm pretty sure we need id now14:03
edmondswesberglu then I think that proves we didn't need to disable this, doesn't it?14:03
efriedHah, it would if you could point to a passing run since last night.  Can you?14:04
edmondswlol... no14:04
edmondswefried well actually...14:04
*** jwcroppe has joined #openstack-powervm14:04
esbergluWell they are all gonna fail because that test is failing14:04
efriedRight, that's my point.14:05
esbergluefried: That and 1 other. But idk what the other one is since we can't look at the logs... So that's step 114:06
edmondswefried I thought I'd seen a run pass, but I can't find it14:06
esbergluedmondsw: Yeah I think a newton CI check passed14:06
efriedObjection, your honor: hearsay.14:06
esbergluBecause failing to delete a network in newton must return something different and doesn't kill the whole run14:07
*** tjakobs has quit IRC14:08
esbergluefried: Not sure where we landed on 5469. Do we want to leave that test disabled until I have a chance to take a further look?14:14
esbergluI don't have enough information to open a bug at this point14:14
efriedesberglu Okay, go for it.14:15
edmondswesberglu yeah, leave it skipped for now, but once we get the CI passing let's come back to that14:21
esbergluedmondsw: Yep14:21
openstackgerritMatthew Edmonds proposed openstack/nova-powervm master: Remove volume_group_vios_name
edmondswefried are you saying there ^ that nova will raise an exception if a conf setting isn't defined in code? I don't think that can be right... it should just get ignored14:24
efriededmondsw I may be thinking of referring to a conf option that isn't defined.14:25
efriedEasy enough to try...14:25
edmondswefried yeah, that would raise14:25
efriededmondsw Yuh, mahbad.14:28
efriedSeems slightly dangerous - a one-character typo in a conf option will get you silently using the default, which may take a long time to notice depending on what option we're talking about.14:30
efriededmondsw Having gotten no bites on perhaps you should send something out to the ML.14:36
efriedFor something like this, I think describing the problem in words will get you more attention than "yet another review someone's asking me to look at'14:37
edmondswefried sure14:44
edmondswesberglu should 5468 be split into a change that adds the new capability vs. a change that remove the old -n?14:49
edmondswesberglu then you could add the new capability, remove -n in neo-os-ci, remove the -n support14:50
edmondswas 3 separate patches in that order14:50
esbergluedmondsw: Yeah good idea14:50
*** dwayne has joined #openstack-powervm14:55
esbergluefried: edmondsw: No luck specifying the full /opt/stack/logs/ path14:57
*** tjakobs has joined #openstack-powervm15:03
edmondswesberglu are the logs just not there?15:10
esbergluedmondsw: I ran a manual run up through the journalctl output15:11
esbergluThe logs are in /opt/stack/logs/ and all get zipped. You can see it happen in console.html as well15:11
esbergluI think it has to do with the jenkins "workspace"15:11
esbergluNot sure what changed though. Our config didn't15:12
*** mdrabe has quit IRC15:18
edmondswefried should I wait for the CI to pass before +W on ?15:21
*** mdrabe has joined #openstack-powervm15:30
efriededmondsw I'm not sure we're allowed to merge *-powervm without our CI passing.15:42
efriedIn any case, it's bad form.  So yeah, let's hold off.15:42
efriedesberglu Sounds like it could be a permissions thing.15:42
esbergluefried: edmondsw: Patching logic is tested and working16:03
efriedesberglu Okay, I have a couple of nits on there, could be done in a followon.16:05
esbergluefried: Nah I can do them. No real hurry to get this in, we aren't patching anything right now anyways16:05
*** k0da has quit IRC16:06
efriedesberglu Comments up in PS516:07
esbergluefried: Okay. Will address them later. Back to jenkins16:07
esbergluefried: After this I just need to add support to patch devstack and we can live patch essentially everything16:09
esbergluAlso gonna move the list of pypowervm patches external16:09
esbergluefried: Okay so logs/*.gz is what we want for the source (actually we could probably be more descriptive)16:12
esbergluThis is using antglob syntax so that should match /opt/stack/logs/*.gz16:12
esbergluJust have to find out why it can't find the logs directory16:13
esberglulogs/*.gz matches with any number of directories in front of logs16:14
esbergluI might try outputting journalctl to logs instead of /opt/stack/logs/ while I try to find out why it can't see /opt/stack/logs16:16
esbergluThat should also match the regex, but maybe jenkins has access there16:16
esbergluefried: How do you look at how rst renders in gerrit? Or is that only for openstack gerrit16:20
edmondswesberglu if you modify 5468 to make "openstack/" be part of the repo names, you could then use repo names like 'openstack-dev/devstack' and share the same patching logic, no?16:23
edmondswefried are you going to rebase ?16:25
esbergluedmondsw: Sure could16:25
efriededmondsw I *think* you can say tox -e docs and it'll build the docs/build dir, whose contents you can then open up in a browser.16:28
*** dwayne has quit IRC16:28
edmondswesberglu I think that was for you16:28
efriedoh, yeah, sorry.16:28
efriededmondsw Yes, I'll rebase it this afternoon.16:28
esbergluefried: thx16:28
esbergluI'm going to try to add a devstack silent job to our CI16:29
esbergluWhich should help us identify when it's devstack breaking us vs. something else16:29
esbergluefried: edmondsw: One more fix. 547016:58
esbergluMissed a line in the in-tree local.conf16:58
edmondswesberglu remove the leading /17:00
esbergluedmondsw: Done17:01
edmondswI thought you'd already fixed that in the OOT one, but it looks like it also has a leading /17:02
edmondswfix it as well?17:02
edmondswor am I missing something? esberglu17:02
esbergluedmondsw: I must not have pushed properly. Because I know I went and fixed that. And was pretty sure that I had that line in the it conf17:04
esbergluedmondsw: Done17:05
edmondswesberglu +217:06
esbergluedmondsw: Okay everything should be passing now minus 1 unidentified test for master OOT and log publishing17:07
*** dwayne has joined #openstack-powervm17:16
*** kylek3h has quit IRC17:19
*** esberglu has quit IRC17:21
*** esberglu has joined #openstack-powervm17:22
*** esberglu has quit IRC17:26
*** esberglu has joined #openstack-powervm17:34
*** esberglu has quit IRC17:39
*** tjakobs has quit IRC17:41
*** esberglu has joined #openstack-powervm17:47
*** esberglu has quit IRC18:19
*** esberglu has joined #openstack-powervm18:19
edmondswefried your comment in about not needing to rebuild the adapter doesn't really help us unless I put back the _initialized var you had me remove18:20
edmondswthat what you want?18:20
*** burgerk has joined #openstack-powervm18:22
edmondswI don't think it would really help, though18:22
edmondswI mean, I don't think it would simplify any logic18:23
*** tjakobs has joined #openstack-powervm18:31
efriededmondsw Why wouldn't `adapter is None` equate to `not _initialized`?19:18
efriedAnd the logic in question would be a big try: block which starts with `if adapter is None: create Adapter`19:20
edmondswefried I don't understand why you ask... I already made exactly that change19:21
*** smatzek has quit IRC19:21
edmondswyour first statement anyway... the second doesn't work19:21
efriedOkay, let's do it like this: I'm gonna write it up pseudocode style and then we can walk through it and you can tell me why it doesn't work.19:23
efriedAssuming me writing it up doesn't make that clear to me :)19:23
edmondswgood luck :)19:23
efriededmondsw It's the stuff like _built_adapter and _initialized_at_entry that makes me go ooky19:25
efriedPretty sure the not-needing-to-build-the-adapter-more-than-once thing allows us to get rid of _built_adapter, anyway.19:26
edmondswI don't see how19:26
efriedOkay, okay, let me play with it.19:26
efriedYou may very well be right.19:26
efriedBut I'll go down fighting.19:26
edmondswwouldn't have it any other way19:26
efriededmondsw So actually, I'm having a new thought here that I wouldn't mind brainstorming a bit.19:28
efriedWe want to do two completely different things that potentially disable/enable the compute host.19:29
efriedOne is initial setup, which includes adapter creation, vios check, scrub, disk driver.19:29
efriedThe other is refreshing the host_wrapper.19:29
efriedSo I wouldn't mind being able to say...19:30
efriedwith update_compute_host_status(...):19:30
efried    self.host_wrapper = System.get()[0]19:31
efriedkind of thing19:31
edmondswefried I think you lost me at "The other is..."19:32
edmondswyou seem to be implying we only do adapter creation, vios check, scrub, and disk driver in init_host... and we can't do that19:32
efriedInitial setup we have to do once.  Though arguably we want to do the VIOS check any time we come back to life, but let's put that aside for the moment.19:32
edmondsweven adapter creation cannot only be done once19:33
edmondswit has to be done until it succeeds once, which isn't the same thing19:33
efriedWe only need to do it once successfully - yeah.19:33
efriedokay, so let me play with *that* a bit.  Though it won't wind up bearing a lot of resemblance to what you've done.19:34
*** thorst has quit IRC19:37
*** thorst has joined #openstack-powervm19:39
*** thorst has quit IRC19:44
*** smatzek has joined #openstack-powervm19:44
*** burgerk has quit IRC19:48
*** thorst has joined #openstack-powervm19:59
esbergluthorst: efried: edmondsw: So I'm still trying to figure this log publishing out. We are using this weird snapshot of an scp plugin20:30
esbergluI looked at it a while back with adreznec and we had no idea where it actually came from or why it worked and the official scp plugin did not20:31
esbergluI'm nearly certain that it is that coming back to bite us20:31
esbergluI'm working on getting the latest scp plugin working20:31
*** thorst has quit IRC20:33
edmondswesberglu weird... good luck20:34
esbergluedmondsw: Also I think another piece of the intree local.conf got left behind on that bad gerrit commit20:34
edmondswI mean, do you have a patch number for me to check?20:35
esbergluAbout to push it up20:37
esbergluedmondsw: ^20:38
edmondswesberglu should we move the Q_PLUGIN_CONF_FILE line to be with that, since I think they're highly related20:40
esbergluedmondsw: I was going to go through and comment/organize all of the local.conf files20:41
esbergluedmondsw: I can though quick20:41
edmondswsounds good20:41
*** smatzek has quit IRC20:42
esbergluedmondsw: Done20:42
edmondswesberglu +220:44
esbergluedmondsw: thx20:44
edmondswesberglu I'm headed out... have a good weekend!20:44
esbergluedmondsw: Same to you20:44
*** k0da has joined #openstack-powervm20:51
*** edmondsw has quit IRC20:54
*** jpasqualetto has quit IRC20:55
*** esberglu has quit IRC21:00
*** mdrabe has quit IRC21:35
*** tjakobs has quit IRC21:53
*** dwayne has quit IRC21:55
*** thorst has joined #openstack-powervm21:59
*** thorst has quit IRC22:04
*** k0da has quit IRC22:05
*** thorst has joined #openstack-powervm22:20
*** jwcroppe has quit IRC22:20
*** thorst has quit IRC22:20
*** dwayne has joined #openstack-powervm22:40
*** jwcroppe has joined #openstack-powervm22:54
*** thorst has joined #openstack-powervm23:51
*** thorst has quit IRC23:56

Generated by 2.15.3 by Marius Gedminas - find it at!