IRC logs for #baserock for Tuesday, 2017-10-17

*** gtristan has quit IRC05:04
*** gtristan has joined #baserock05:31
*** gtristan has quit IRC05:35
*** gtristan has joined #baserock05:51
*** jude_ has joined #baserock07:36
*** rdale has joined #baserock07:42
*** edcragg has quit IRC08:07
*** edcragg has joined #baserock08:07
*** toscalix has joined #baserock08:11
*** jonathanmaw has joined #baserock08:20
*** ssam2 has joined #baserock09:10
*** ChanServ sets mode: +v ssam209:10
*** tiagogomes has quit IRC09:22
*** locallycompact has joined #baserock09:23
*** tiagogomes has joined #baserock09:23
gitlab-br-botybd: merge request (benbrown/fix-config-access->master: Fix config access, it is in the app module) #395 changed state ("opened"): https://gitlab.com/baserock/ybd/merge_requests/39509:48
gitlab-br-botybd: merge request (benbrown/fix-config-access->master: Fix config access, it is in the app module) #395 changed state ("opened"): https://gitlab.com/baserock/ybd/merge_requests/39509:49
gitlab-br-botybd: merge request (benbrown/fix-config-access->master: Fix config access, it is in the app module) #395 changed state ("merged"): https://gitlab.com/baserock/ybd/merge_requests/39509:53
gitlab-br-botdefinitions: merge request (benbrown/sam/fix-ci->master: Fix caching in CI and some other optimizations) #63 changed state ("opened"): https://gitlab.com/baserock/definitions/merge_requests/6309:58
benbrown_ssam2: The issue is likely that the job gets a runner that's been used for ybd, which stores non-gits in /cache10:00
ssam2between builds ?10:01
ssam2oh, between builds if the runner is reused10:02
ssam2but not in the shared cache10:02
ssam2yeah i see the issue i guess10:02
* benbrown_ nods10:02
benbrown_So I was considering adding a new manager instance that was explicitly used for bst pipelines10:03
ssam2is that difficult ?10:03
benbrown_Not really10:03
ssam2ok. an easier approach would be to just `rm cache/ybd/artifacts` before running buildstream builds10:03
benbrown_that also works10:03
benbrown_just means that we're not getting the full benefit of /cache10:03
ssam2yeah10:04
ssam2ha, my diagnostic broke everything anyway10:04
ssam2ok, well you've got to sort out separate runners then i shall leave it in your capable hands10:04
ssam2*if you've got time10:05
gitlab-br-botdefinitions: merge request (benbrown/sam/fix-ci->master: Fix caching in CI and some other optimizations) #63 changed state ("opened"): https://gitlab.com/baserock/definitions/merge_requests/6310:06
benbrown_ssam2: Aye, I'll aim to get it sorted today10:10
jjardonbenbrown_: no , only one manager, please10:30
*** adds68 has quit IRC10:33
*** adds68 has joined #baserock10:36
*** toscalix has quit IRC10:40
benbrown_jjardon: Any particular reason why? Not after a new digitalocean instance, just another docker+machine registered from the same machine.10:41
jjardonI would prefer not to have to tag jobs to run in specific hardware10:46
benbrown_jjardon: I'd rather we didn't have to nuke ybd cache every time for bst jobs10:46
benbrown_tags are designed for this, why not use them?10:47
jjardonThink in gitlab.com runners; you don't use a specific runner10:47
benbrown_Don't we have shared runners disabled?10:48
benbrown_Regardless, we would use a specific runner if we used tags10:49
benbrown_jjardon: Yeah, we don't use shared runners. Not sure what you mean?10:50
jjardonWhen you use gitlab runners you don't specify what runner you want to run on10:50
jjardonI'd like to keep that way (in case we do not have more funding for ours own runners, for example)10:51
benbrown_jjardon: do shared runners even provide /cache?10:53
benbrown_jjardon: actually, do the shared set privileged to true, because that's a requirement for ybd10:55
ssam2shared runners do set privileged to true10:56
ssam2buildstream is using the shared runners10:56
ssam2(for the buildstream/ repos)10:56
benbrown_jjardon: since we already require our manager for ybd, we may as well set that as the tagged instance10:57
benbrown_jjardon: and enable shared runners for buildstream jobs10:57
ssam2that would make buildstream builds really slow though10:58
jjardonIs this sth that definitely can't be fix in .gitlab-ci.yml or in builstream?10:59
benbrown_jjardon: the only fix in the CI yaml is nuking /cache before each bst task10:59
ssam2nuking cache/ybd/artifacts/ should be enough ?10:59
ssam2maybe cache/buildstream/artifacts too11:00
benbrown_ssam2: sure, but that's currently all that's in there11:00
ssam2no, there are also gits11:00
ssam2and ostree sources for buildstream builds11:00
benbrown_separate mechanism11:00
ssam2and those are the things which are really slow to re-fetch11:00
ssam2oh OK11:01
*** adds68 has quit IRC11:01
ssam2those are the only things i care about caching though really11:01
ssam2pulling artifacts from the remote cache is a pretty fast operation11:01
ssam2whereas pulling gits from git.baserock.org takes ages11:01
*** adds68 has joined #baserock11:01
benbrown_true, just a nice to have I guess11:01
ssam2and pulling ostree sources from sdk.gnome.org is similarly dog slow11:01
benbrown_I'll go for nuking /cache/ybd/ then :)11:02
gitlab-br-botdefinitions: merge request (benbrown/sam/fix-ci->master: Fix caching in CI and some other optimizations) #63 changed state ("opened"): https://gitlab.com/baserock/definitions/merge_requests/6311:09
benbrown_ssam2: This error is seemingly non-full disk related: https://gitlab.com/baserock/definitions/-/jobs/3629198911:13
ssam2yes, that one is weird11:13
ssam2i think it may be path length related ... as that system has definitely built in the past11:14
ssam2i can try and reproduce locally; but it may depend on the filesystem used on that machine11:19
ssam2i moved my cache to an even longer path name and built that element (unmaintained/openstack-services/posix-ipc-tarball.bst) and it worked ifne12:09
ssam2*fine12:09
ssam2oh, maybe it's not quite as long12:11
ssam2also, seems that the CI is not pushing buildstream artifacts any more12:18
ssam2which is weird12:18
ssam2oh, because it's not on a protected branch so it doesn't get the private key12:19
benbrown_We only allow pushes for protected branches, I unprotected so I could force push.12:19
ssam2right12:20
ssam2i was just wondering why buildstream was going so slow12:20
benbrown_I'd hoped once it got to that step, it would see that the branch is now re-protected, but that's not happened.12:21
*** gtristan has quit IRC13:24
*** gtristan has joined #baserock13:26
*** toscalix has joined #baserock14:55
*** toscalix has quit IRC15:56
*** ssam2 has quit IRC16:28
*** jude_ has quit IRC16:29
*** locallycompact has quit IRC17:02
*** locallycompact has joined #baserock17:02
*** locallycompact has quit IRC17:02
*** locallycompact has joined #baserock17:03
*** jonathanmaw has quit IRC17:08
*** locallycompact has quit IRC18:37
*** rdale has quit IRC18:58
*** gtristan has quit IRC21:13

Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!