IRC logs for #baserock for Wednesday, 2015-07-15

*** radiofree has quit IRC00:52
*** radiofree has joined #baserock00:54
*** rdale has quit IRC01:07
*** rdale has joined #baserock01:27
*** zoli__ has joined #baserock02:01
*** zoli__ has quit IRC02:06
*** zoli__ has joined #baserock05:46
*** mike has joined #baserock06:19
*** mike is now known as Guest6784806:19
*** Guest67848 has quit IRC06:57
*** rdale has quit IRC07:05
*** mike has joined #baserock07:07
*** mike is now known as Guest1397607:07
*** rdale has joined #baserock07:26
*** dabukalam has quit IRC07:30
*** rdale has quit IRC07:31
*** dabukalam has joined #baserock07:32
*** rdale has joined #baserock07:32
*** zoli__ has quit IRC07:50
*** bashrc_ has joined #baserock07:54
*** mariaderidder has joined #baserock07:55
*** paulw has joined #baserock08:11
*** mdunford has joined #baserock08:18
*** zoli__ has joined #baserock08:44
*** ssam2 has joined #baserock08:47
*** ChanServ sets mode: +v ssam208:47
*** mdunford has quit IRC08:49
*** mariaderidder has quit IRC09:03
*** mdunford has joined #baserock09:03
*** mariaderidder has joined #baserock09:09
*** lachlanmackenzie has joined #baserock09:13
*** mariaderidder has quit IRC09:15
* persia is pondering proposing a code of conduct for Baserock, and wondered if others have opinions09:25
KinnisonIn general I think CoCs are good09:27
jmacs+109:28
KinnisonI have this odd argument with myself between "We're too small to need one" and "Without one, how can we hope to grow safely"09:28
KinnisonI'd err on the side of having one09:28
ssam2seems fine if it is short09:28
*** mariaderidder has joined #baserock09:29
persiaThey are never short, but I'll take a survey of a few, and submit a candidate to the mailing list later in the week.09:30
ssam2ah, if it's a pre-existing one that we don't need to maintain then there is less need for it to be short09:30
petefoth_A good idea to have one at some point, so long as it doesn't increase friction for possible new contributors. I'm not sure we need one yet though, given the relativelky low numbers of current contributors09:30
persiaI think of the key points as being 1) raise problems when they are known, 2) respect others, and 3) don't block others work09:31
ssam2I think "assume people mean well" is key09:32
petefoth_ssam2: if there is a pre-existing one that works, juts reference it, don't create a copy09:32
* persia encourages others to share opinions of key points to help find the right document09:32
persiapetefoth_: Unreliable: source could change without our agreement.  That said, we can create a copy using technologies that allow us to update our copy if the source changes.09:33
persia(if we want)09:33
jmacsI.e. lorry it. Sounds sensible.09:33
petefoth_If it changes, check it still works for us: if not either submit changes upstream or fork09:34
straycathow does 3) work in practice? i've had work rejected from other projects because i was wrong, why would baserock behave differently here?09:47
pedroalvareznobody looks at https://mason-x86-64.baserock.org/ these days :(09:54
tiagogomesMason sending IRC notifications, would help09:55
rjekCould perhaps do that with a generic off-the-shelf bot and having Mason generate an RSS feed?09:57
persiastraycat: In other projects I've worked in, 3) is mostly about being clear in communications when someone stops working on something on which others rely for their work, so that the others can step in, rather than being blocked.  When something is wrong, it is wrong.10:02
ratmice___not CoC stuff as its focused on patch review, anyhow I agree with all the points raised here: http://tromey.com/blog/?p=43610:02
straycatpersia, okay that makes sense10:03
persiaratmice___: Thanks!  Those are all great points.10:03
straycatthose are good points thanks ratmice___10:07
ssam2pedroalvarez: mostly because when the output gets long it causes my normal web browser to eat all my RAM10:10
pedroalvarez:)10:11
* pedroalvarez will send a couple of fixes10:11
pedroalvarezhttps://gerrit.baserock.org/#/c/975/10:16
ssam2oh, that breakage was my fault too!10:27
ssam2good spot10:27
persiaI've been watching storyboard upstream, and apparently OpenStack is running trunk, so that the fixes made by the developers end up being available quickly.10:30
persiaIs there any way the Baserock infrastructure could have this sort of auto-update, to take advantage of recent (and ongoing) improvements?10:31
persiaSome of storyboard upstream appears to also be in this channel, so I expect they would be responsive to any issues encountered.10:31
SotKpersia: Do you mean auto-updating Baserock's StoryBoard instance or auto-updating all of our infrastructure?10:33
persiaSotK: I meant only storyboard: I think we'd need a closer relationship with more upstreams to autoupdate everything.10:34
* SotK thinks it should definitely be updated from whatever version it is currently running when someone has time, even if auto-updating isn't wanted10:35
paulsherwoodam i right that OpenStack is confident in trunk because of their test/ci regime?10:36
paulsherwoodand in effect we could/should have the same level of confidence10:36
paulsherwoodif so +1 on adopting trunk10:37
paulsherwoodit would be an interesting use-case anyway10:37
persiapaulsherwood: Yes, OpenStack is fairly aggressive about that, and some folk run lots of OpenStack services off trunk.10:37
paulsherwoodso +110:37
persiaSo yeah, if we trust their CI/review process, then we should trust their trunk, especially since we don't have to go far to make noise if someone approves something that breaks something.10:38
persiaThat said, I don't know how aggressive the testing is for Storyboard: at least some recent patches seem to have not passed some of the non-voting tests.10:38
*** Guest13976 has quit IRC10:39
*** pdar_ is now known as pdar10:40
* SotK thinks that the non-voting test in question may be broken10:41
Zara_yeah, it seems the 'check-storyboard-js-integration' test always fails.10:42
ssam2our current storyboard instance is a bit of a hack10:53
ssam2i'll deploy anything at baserock.org that is less of a hack :)10:53
*** Guest13976 has joined #baserock10:53
persiassam2: Shall I take that to mean that you welcome patches to the infra repo for an auto-updating storyboard?10:55
ssam2yes10:55
Zara_ooi, how was baserock's storyboard set up? (ie: built from a reference 'storyboard' system, or some other system with storyboard then downloaded via pip and npm and the rest?)10:55
persiaHeh.  I was hoping for something more active, but that certainly meets my needs.10:56
ssam2I'm not sure that feel comfortable just pulling the latest storyboard on the current instance, it might well break, and doesn't have much of a rollback mechanism10:56
persiassam2: rollback would be 1) yell at upstream, 2) pull the new HEAD of master with the revert commit10:56
persia(or at least, that is how OpenStack seems to do it)10:56
ssam2right. I'm cagey about updating from its current state, even10:57
ssam2it may have grown more dependencies, require different apache config, etc.10:57
ssam2Zara: I tried to use the puppet-storyboard scripts on Fedora, which mostly worked, but I had to edit the storyboard config and apache config manually (and maybe other stuff, I forget)10:57
ssam2I also tried to use a tool call Packer for this, which it turns out isn't actually very useful for us10:58
ssam2as everything it does can be done by Ansible with much less faff10:58
ssam2the files I used to deploy it are here: http://git.baserock.org/cgi-bin/cgit.cgi/baserock/baserock/infrastructure.git/tree/baserock_storyboard10:58
ssam2but I think throwing that away and starting again is probably a good bet :)10:59
ssam2one issue is that they seem to use forked versions of Puppet modules10:59
ssam2another is that they clearly only test deployment on Ubuntu. I did some kind of hacking to get systemd units for the services it provides, but never got them in a state I could send them upstream10:59
ssam2so, help here is much appreciated. The infrastructure is all hosted in OpenStack, so any set of build and deploy instructions that produce a Storyboard VM in OpenStack are suitable for baserock.org infrastructure11:01
* persia wonders if the openstack-infra puppet modules could just be used entire11:01
ssam2Hopefully. Maybe the issues I found have now been resolved, or maybe someone will have time to resolve them11:02
ssam2My lack of knowledge about Puppet didn't help11:02
pedroalvarezAnsible!11:03
* pedroalvarez shuts up11:03
ssam2i'd prefer Ansible :) it's much easier to understand than Puppet in my opinion11:04
persiaAnsible use in OpenStack is growing, but I don't believe infra has considered any migration yet, despite affection by many.11:18
Zara_hm, would it be possible to make a dummy storyboard with the same settings as the existing one, then update that and see if things break?11:20
persiaPossibly, although if there were manual steps, it may be difficult to achieve that.11:21
paulsherwoodcould we do the ansible approach and offer it to openstack?11:21
persiaNo: infra is very opposed to large code drops.11:22
persiaIt might be possible to do the ansible approach inside openstack infra, and adopt it though.11:22
ssam2zara: that's a good idea. I can provide you with a snapshot of the current image11:24
ssam2if you'd like11:24
*** petefoth has joined #baserock11:28
*** doffm_ has joined #baserock11:31
*** mariaderidder_ has joined #baserock11:31
Zara_I think it'd be good to make it available; I'm not sure if I'll have time to test it myself right now, but I might in the future, or someone else might.11:36
*** petefoth_ has quit IRC11:37
*** mariaderidder has quit IRC11:37
*** doffm has quit IRC11:37
ssam2Zara_: makes sense, I'll put it on my list of things to do11:42
Zara_:) I'm hoping it works perfectly on the first try, buuut...11:45
ssam2it seems the 'run-parts' in Busybox actually has more features than 'run-parts' in Fedora12:06
ssam2there's a first12:06
ssam2what I want is `run-parts --exit-on-error`12:06
ssam2does that feature exist in the run-parts in Debian/Ubuntu ?12:07
ssam2or any other distro (Arch, Gentoo etc)?12:07
ssam2seems so12:09
*** Guest13976 has quit IRC12:39
*** straycat has left #baserock12:53
*** zoli__ has quit IRC12:59
*** nowster has quit IRC14:05
paulsherwooditstool upstream is now officially github... https://github.com/itstool/14:17
* paulsherwood has sent a patch14:17
ssam2thanks!14:33
*** petefoth has quit IRC14:42
*** nowster has joined #baserock14:49
*** paulw has quit IRC14:55
*** paulw has joined #baserock14:55
paulsherwoodssam2: is that a +1 ? :)15:09
*** nowster has quit IRC15:36
KinnisonOn CoCs: https://twitter.com/sarahsharp/status/62133468415738265615:40
*** nowster has joined #baserock15:42
persiaShe has a fairly good view of these things: she, herself, regularly plays in high-temperature communities, but receives input from a large constituency of folk who wonder how she can stand it, helping inform her insight.15:43
*** paulw has quit IRC15:57
*** rdale has quit IRC16:02
paulsherwoodanyone know why writeexts.py now tries --json as a flag on system-version-manager list?16:10
KinnisonIt was part of some work done to allow it to detect preexisting systems16:11
KinnisonThere were two parts, one for writeexts.py and one for system-version-manager16:11
Zara_(Agree that CoCs are generally a good idea and we should have one, more to reassure new people who might read significance into the absence.)16:11
paulsherwoodok. this is another example of the back-n-forth problems - we can't guarantee two parts get updated at once16:12
ssam2paulsherwood: does it fail if the --json flag isn't supported?16:18
KinnisonBased on the code I reviewed at the time, it falls back cleanly16:19
paulsherwoodthe deployment succeeds, but the backtrace is ugly16:28
paulsherwoodi wondered if the deployment might have only succeeded because ybd may ignore error codes (i know folks have tsked about ybd's carelessness regarding errors)16:29
*** nowster has quit IRC16:35
*** mariaderidder_ has quit IRC16:41
ssam2looking at http://git.baserock.org/cgi-bin/cgit.cgi/baserock/baserock/definitions.git/tree/extensions/ssh-rsync.check, it should neither fail nor give a backtrace to stdout16:55
ssam2it'll give a backtrace to the log file, which I think is OK16:55
ssam2i didn't try it though, so i could be talking nonsense16:55
ssam2need to leave now, have a good evening!16:55
*** ssam2 has quit IRC16:55
*** nowster has joined #baserock16:57
paulsherwoodgnite... that may be ybd... it tails logs to std_out on error (which seems useful for builds - this is the first time i've seen it in deploy)16:58
radiofreeyou probably want to do a sys.stdout.flush after printing16:59
*** bashrc_ has quit IRC17:05
*** mdunford has quit IRC17:10
paulsherwood:)17:16
*** nowster has quit IRC17:39
*** nowster has joined #baserock17:51
*** lachlanmackenzie has quit IRC18:43
*** mdunford has joined #baserock21:54
*** mdunford has quit IRC22:07

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