Tuesday, 2019-01-08

*** rlandy|rover is now known as rlandy|rover|bbl00:23
clarkband now I've got through the ansible version spec00:24
*** bhavikdbavishi has joined #zuul02:06
*** bhavikdbavishi has quit IRC02:09
*** bhavikdbavishi has joined #zuul02:20
*** rlandy|rover|bbl is now known as rlandy|rover02:33
dmsimardthe new stuff in the ui is genius02:50
dmsimardtristanC: when browsing jobs (ex: http://zuul.openstack.org/job/zuul-quick-start ).. would the next logical step be to list (recent) builds for that job or something like that ?02:53
dmsimardbecause that'd be pretty cool.02:53
dmsimardI know you can go the other way and filter builds by job02:53
tristanCdmsimard: yes sure, we could add a reducer to update the page with recents build02:56
tristanCdmsimard: i'd like to recursively fetch parent job to display breadcrumbs at the top too02:58
dmsimardthe job graph doesn't provide it already ?02:58
tristanCdmsimard: yes, but corvus -2 the implementation... and this would be complementary, as a graph for a single job page02:59
*** spsurya has joined #zuul04:22
*** bhavikdbavishi has quit IRC04:25
*** rfolco has quit IRC04:34
*** rfolco has joined #zuul04:34
*** rf0lc0 has joined #zuul04:37
*** rfolco has quit IRC04:39
*** rf0lc0 has quit IRC04:40
*** rfolco has joined #zuul04:41
*** bhavikdbavishi has joined #zuul04:44
*** spsurya has quit IRC05:10
*** spsurya has joined #zuul05:13
*** sshnaidm has quit IRC06:52
*** quiquell|off is now known as quiquell07:09
*** pcaruana has joined #zuul07:42
*** quiquell is now known as quiquell|brb07:44
openstackgerritMatthieu Huin proposed openstack-infra/zuul-jobs master: upload-pypi: add option to register packages  https://review.openstack.org/62901807:49
*** dkehn has quit IRC07:58
*** quiquell|brb is now known as quiquell08:11
*** ssbarnea has joined #zuul08:14
*** ssbarnea|bkp2 has quit IRC08:16
*** themroc has joined #zuul08:26
*** sshnaidm has joined #zuul08:31
*** sshnaidm has quit IRC08:47
*** evrardjp_ has joined #zuul08:48
*** evrardjp__ has joined #zuul08:49
*** jpena|off is now known as jpena08:50
*** evrardjp has quit IRC08:51
*** evrardjp_ has quit IRC08:52
*** evrardjp__ has quit IRC08:53
*** evrardjp has joined #zuul09:03
*** hashar has joined #zuul09:04
*** gtema has joined #zuul09:05
*** sshnaidm has joined #zuul09:55
*** bhavikdbavishi has quit IRC10:25
*** dkehn has joined #zuul11:52
*** bhavikdbavishi has joined #zuul11:55
*** jpena is now known as jpena|lunch12:37
*** bhavikdbavishi has quit IRC13:00
*** bhavikdbavishi has joined #zuul13:01
*** zigo has joined #zuul13:28
*** bhavikdbavishi has quit IRC13:32
*** jpena|lunch is now known as jpena13:34
*** jhesketh has quit IRC13:45
*** jhesketh has joined #zuul13:47
*** AlbinVass has joined #zuul14:25
AlbinVassAnyone here that has experience setting up nodepool with winrm?14:26
ShrewsAlbinVass: maybe ask tobiash (i think he added that feature)14:26
AlbinVassBecause I think I've found a bug in executor/server.py. It seems like it wants to add an ssh key to known hosts even though it doens't use ssh14:27
AlbinVassAny way to get in contact with him?14:27
tobiashAlbinVass: did you set the connection-type property in nodepool?14:28
AlbinVassyep, to winrm14:28
AlbinVassit's on row 1262 if that's any help14:29
tobiashAlbinVass: zuul adds all host keys that it got from nodepool14:31
tobiashhowever this list should be empty if you set the connection type to winrm in nodepool14:31
AlbinVassgetting 'NoneType' is not iterable14:31
tobiashas nodepool skips getting the keys from the node in this case14:31
tobiashdo you have a log?14:31
Shrewsversion mismatch between zuul and nodepool maybe?14:33
tobiashI don't see that exception in my deployment14:33
AlbinVassit could be an old docker image14:37
ShrewsAlbinVass: i'm guessing you have a more recent version of nodepool than zuul. when the winrm feature was added to nodepool, there are zuul changes that were added that you need to actually use that feature.14:39
AlbinVassthat could be it actually since I had to set the port to 5986 manually otherwise it was still checking default ssh port14:42
tobiashthe port needs to be set manually because it's not yet derived from the connection type14:43
AlbinVassDocumentation says it's 5986 if connection-type is set to winrm14:44
tobiashoh, maybe I forgot about that14:45
tobiashAlbinVass: you're right, https://git.zuul-ci.org/cgit/nodepool/tree/nodepool/driver/openstack/config.py#n21914:48
tobiashAlbinVass: are you on recent versions of zuul and nodepool?14:48
openstackgerritTobias Henkel proposed openstack-infra/zuul master: Default ssh keys to empty list  https://review.openstack.org/62920114:59
tobiashAlbinVass: this should resolve your exception, but I think there might be more wrong in your setup ^15:00
tobiashAlbinVass: so it would be useful to see what versions you use and having a log15:01
AlbinVassyeah probably, still trying to set it up to ping a server to start with15:05
*** bhavikdbavishi has joined #zuul15:13
dmsimardI could probably hack something together but is there a way to tell wha tthe log URL will be during job runtime ?15:24
pabelangerdmsimard: yup, emit-job-header will do it. See: http://git.openstack.org/cgit/openstack-infra/project-config/tree/playbooks/base/pre.yaml#n115:27
AlbinVassDoesn't like like updating the docker containers fixes it so it's probably something else15:27
dmsimardpabelanger: ah, thanks, I felt like it had been done before but forgot where. Thanks.15:28
AlbinVasstobiash: any chance I get can in touch with you tomorrow?15:28
tobiashjust ping me here, I'll respond as soon as I'm available then15:31
AlbinVassSure, thanks15:32
openstackgerritMerged openstack-infra/zuul master: Default ssh keys to empty list  https://review.openstack.org/62920115:53
*** pcaruana has quit IRC16:20
*** AlbinVass has quit IRC16:20
*** themroc has quit IRC16:37
*** hashar has quit IRC17:07
*** quiquell is now known as quiquell|off17:19
*** gtema has quit IRC17:50
*** rf0lc0 has joined #zuul18:14
*** rfolco has quit IRC18:15
*** rf0lc0 is now known as rfolco18:16
*** panda is now known as panda|off18:17
*** jpena is now known as jpena|off18:30
fungithe kata containers community is abandoning pullapprove for their github workflow, and looking for alternatives. for completeness i suggested that zuul covers a similar featureset or at least is extensible enough through simple jobs to do most of it. unfortunately i'm not very familiar with github-oriented workflows so not sure how other zuul+gh users here do it. they're asking about "keyword matching for merge18:46
fungiblocking" and "gathering ack's before merge" http://lists.katacontainers.io/pipermail/kata-dev/2019-January/000644.html18:46
clarkbpabelanger SpamapS and tobiash ^ are likely best positioned to talk about their approaches18:48
*** bhavikdbavishi has quit IRC18:52
mordredfungi: the awx project is using zuul for gating18:53
mordredon github18:53
mordredfungi: https://github.com/ansible/awx18:53
tobiashwe use githubs branch protection with required status for check, gate and in combination with codeowners review18:54
tobiashour gate is label triggered18:54
pabelangerwe even have a job for dco-license: https://github.com/ansible-network/zuul-config/blob/master/zuul.d/jobs.yaml#L70 based on some logic that bonnyci was doing. Been meaning on pushing it up into zuul-jobs18:55
tobiashalso note that zuul respects other required status checks too (we use e.g the wip bot which also reports as a status check)18:55
tobiashso this could be combined with anything further that sets status18:55
pabelangeryes, out gate and awx is also label triggered18:55
pabelangerour* (ansible-network)18:56
fungiokay, so basically there is something in github called a "label" which you can set acls around who's allowed to add/remove them, and that's how you "vote" to approve and/or block pull requests?18:58
tobiashlabels unfortunately don't have acls18:59
tobiashso we really use it just as a trigger18:59
fungiwell, what they seem to be describing is leaving formatted strings in pr comments, which i don't suppose have acls either18:59
pabelangerfungi: yes, correct. but what tobiash, no ACL anybody with commit access can apply a label18:59
pabelangerin our case, we call it gate :)18:59
pabelangerfor example: https://github.com/ansible-network/vyos/pull/5118:59
fungi"with commit access" sounds like an acl19:00
tobiashok, not a really fine grained acl ;)19:00
tobiashif you want something more sophisticated you can leverage e.g. codeowners with required reviews19:01
pabelangerYah, they are tied, so if you want to label, they also need commit. which means they can force push tags, I've learned19:01
pabelangerseems branch protections don't help for that19:01
tobiashor write/use some external 'approve-bot' that sets a successful status once all needed approvals are there19:01
tobiashbut I suppose using review requirements with codeowners19:04
tobiashwould be what I'd recommend19:04
tobiashcombined with setting a label to trigger the gate19:04
tobiashbut I'm always open for better suggestions19:05
pabelangerI haven't looked at codeowners yet myself, on the list to try19:05
tobiashit's really simple, just enable code review requirements and add a CODEOWNERS file that just lists some group for * in the most simple case19:06
tobiashbut with that self-review won't be possible19:06
*** hashar has joined #zuul20:08
*** spsurya has quit IRC20:41
*** jesusaur has quit IRC23:02
*** hashar has quit IRC23:09
*** rlandy|rover is now known as rlandy|rover|bbl23:16
corvustristanC: do you have a change to make the build url the result url for a build?  are there any pre-requisites for that?23:59

Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!