| *** markwash has quit IRC | 00:00 | |
| *** senk has quit IRC | 00:12 | |
| *** ericw has joined #openstack-infra | 00:15 | |
| *** che-arne has quit IRC | 00:16 | |
| *** matsuhashi has joined #openstack-infra | 00:19 | |
| mordred | fungi: re: https://review.openstack.org/#/c/52367/ - do you feel strongly about trying to do the other format? | 00:27 |
|---|---|---|
| *** rakhmerov has quit IRC | 00:28 | |
| fungi | mordred: nah, what you have is more compact. my original goal was merely to shrink the resulting changelog size anyway | 00:30 |
| *** matsuhashi has quit IRC | 00:31 | |
| fungi | mordred: though per your and lifeless's comments in there, i think my algorithm got the tags while omitting the merges by making multiple passes | 00:32 |
| *** matsuhashi has joined #openstack-infra | 00:32 | |
| mordred | fungi: I agree | 00:32 |
| fungi | first a git tag pass, then a git log for each tag range | 00:32 |
| * fungi looks back at his script | 00:33 | |
| *** dcramer_ has quit IRC | 00:36 | |
| *** matsuhashi has quit IRC | 00:37 | |
| fungi | yeah, i did a loop over the git tag output, filtered for only release tag patterns, and then did a git log (well, shortlog in mine) --no-merges over each successive pair of tags to get the changes between them | 00:37 |
| fungi | seemed to work even if the tag was on a merge commit | 00:38 |
| mordred | fungi: yeah - I agree that will work. I'm not sure the extra author info is worth the double loop thing though. I think lifeless point was that it's sad we can't just do a single git invocation and get the log output in the format we want | 00:41 |
| lifeless | right | 00:41 |
| *** michchap has quit IRC | 00:41 | |
| lifeless | parsing is overhead | 00:41 |
| lifeless | manual recursive descent more so | 00:41 |
| fungi | ahh, right. yeah, parsing or loop are the two options | 00:41 |
| lifeless | if we can have git DTRT we avoid all that | 00:41 |
| *** michchap has joined #openstack-infra | 00:42 | |
| lifeless | if we can't, then I start suggesting parser generators (like pyparsing or parsley) to mordred | 00:42 |
| mordred | yah. except this is pbr, so I'm not going to add a dep on pyparsing or parsley | 00:43 |
| mordred | to get potentially more elegant code implementing the changelog writing :) | 00:43 |
| mordred | (in fact, I'm not going to add a depend on anything) | 00:44 |
| mordred | I do agree that getting git to do it directly would be baller | 00:44 |
| lifeless | pbr isn't allowed to depend on anything ? | 00:45 |
| fungi | i agree that, as a module targeting basic packaging functionality, its dependency list should be as tight as possible | 00:46 |
| *** ericw has quit IRC | 00:46 | |
| *** SergeyLukjanov has joined #openstack-infra | 00:47 | |
| mordred | lifeless: at the moment, I believe that all depends to pbr need to be MASSIVELY scruitinzed | 00:47 |
| mordred | lifeless: because pbr comes in via setup_requires, which is operated by easy_install, which is a pile of bork | 00:47 |
| mordred | the more complex we make pbr installation, the more opportunities we introduce for corner cases and fundamental breaking | 00:48 |
| *** SergeyLukjanov has quit IRC | 00:49 | |
| mordred | so, I'm not going to say 'not allowed' - but I think the chances of finding a case where it's justified are slim | 00:49 |
| mordred | and this one is certainly not one | 00:49 |
| *** matsuhashi has joined #openstack-infra | 00:49 | |
| fungi | mordred: out of curiosity have you compared the size of nova's changelog before and after? | 00:49 |
| mordred | fungi: I have not - that's an excellent idea | 00:49 |
| fungi | i was seeing something like a fourfold reduction in size, so i'd expect similar | 00:50 |
| mordred | here it is with trunk pbr | 00:50 |
| mordred | -rw-r--r-- 1 mordred mordred 7162597 Oct 27 20:50 ChangeLog | 00:50 |
| lifeless | mordred: why is it pbr that does ChangeLog generation? | 00:52 |
| mordred | and condensed | 00:52 |
| mordred | -rw-r--r-- 1 mordred mordred 915127 Oct 27 20:52 ChangeLog | 00:52 |
| mordred | lifeless: because it handles setup.py interactions | 00:52 |
| mordred | of which sdist is one | 00:52 |
| *** nosnos has joined #openstack-infra | 00:52 | |
| mordred | pbr is a collection of repeated code that otherwise would go into setup.py | 00:53 |
| fungi | nice, so eightfold or so | 00:53 |
| mordred | yah | 00:53 |
| mordred | that should make zigo happier | 00:53 |
| fungi | of course the actual space savings will depend on the relative verbosity of commit messages in a given project | 00:54 |
| mordred | sure. I think the real win is that the result is MUCH more useful to read | 00:54 |
| fungi | very much so | 00:55 |
| mordred | and also underscores the point of having nicely readable commit summaries | 00:55 |
| mordred | oh. you know what? I should strip trailing periods in the output. OR I should add periods if they aren't there | 00:55 |
| mordred | so that the ChangeLog reads consistently | 00:55 |
| mordred | regardless of commit subject style choice | 00:55 |
| fungi | both, commit subhects with no period should get one added, and those with one should have it stripped. that way you can annoy everyone | 00:56 |
| mordred | fungi: ++ | 00:56 |
| *** markwash has joined #openstack-infra | 00:58 | |
| openstackgerrit | Monty Taylor proposed a change to openstack-dev/pbr: Make ChangeLog more condensed https://review.openstack.org/52367 | 01:01 |
| mordred | fungi: also - I was thinking - perhaps if we included support for not including commit lines that contained something | 01:02 |
| mordred | like "NIT: Fix typo" or something | 01:02 |
| fungi | it's a savings which wouldn't be realized in established projects for a long time, but it's not a terrible idea | 01:03 |
| mordred | yah. was mainly thinking in terms of making the ChangeLog something that communicates things that were done | 01:04 |
| mordred | there are usually commits like that which do not communicate anything | 01:04 |
| *** markwash has quit IRC | 01:05 | |
| mordred | maybe NIT: and AUTO: so we can also mark things like requirements changes that are just mechanical system commits | 01:05 |
| fungi | probably needs a bit of an rfc to find out the variety of ways devs might be interested in filtering commits out of the changelog | 01:07 |
| mordred | yah | 01:07 |
| mordred | let's make that a follow on change | 01:07 |
| Shrews | mordred: go as expected last night? | 01:08 |
| *** dims has quit IRC | 01:11 | |
| openstackgerrit | A change was merged to openstack-dev/pbr: Move base test case logic out of __init__.py https://review.openstack.org/48860 | 01:13 |
| mordred | Shrews: yeah - it was really good | 01:16 |
| *** dkranz has quit IRC | 01:21 | |
| *** yamahata has joined #openstack-infra | 01:24 | |
| *** rakhmerov has joined #openstack-infra | 01:25 | |
| *** sdake has quit IRC | 01:26 | |
| *** alexpilotti has quit IRC | 01:28 | |
| *** fallenpegasus has joined #openstack-infra | 01:38 | |
| *** guohliu has joined #openstack-infra | 01:39 | |
| *** senk has joined #openstack-infra | 01:46 | |
| *** markwash has joined #openstack-infra | 01:46 | |
| *** bingbu has joined #openstack-infra | 01:47 | |
| openstackgerrit | Monty Taylor proposed a change to openstack-dev/pbr: Make ChangeLog more condensed https://review.openstack.org/52367 | 01:50 |
| mordred | fungi: ^^ | 01:50 |
| mordred | that updates the logic around lines with multiple commits | 01:50 |
| *** michchap has quit IRC | 01:53 | |
| *** ericw has joined #openstack-infra | 01:53 | |
| *** michchap has joined #openstack-infra | 01:55 | |
| openstackgerrit | Monty Taylor proposed a change to openstack-dev/pbr: Ignore boring commit messages https://review.openstack.org/54048 | 01:57 |
| *** rakhmerov has quit IRC | 01:59 | |
| *** sdake has joined #openstack-infra | 02:00 | |
| *** wenlock has joined #openstack-infra | 02:00 | |
| mordred | fungi, clarkb: fyi - I have released a new pbr version so that the windows guys won't be screwed anymore | 02:01 |
| *** dims has joined #openstack-infra | 02:01 | |
| mordred | it SHOULD not break anything | 02:01 |
| mordred | but heads up anyway | 02:01 |
| fungi | mordred: when you say lines with multiple commits, are you talking about tags on merge commits? or something else? | 02:04 |
| mordred | fungi: what I meant is a single commit with more than one tag | 02:05 |
| fungi | aha! yes, okay now i see what you're solving | 02:05 |
| mordred | which is the case with 2013.2 and 2013.2.rc1 | 02:05 |
| lifeless | mordred: speaking of mailmap | 02:05 |
| lifeless | did you see my RFC about it ? | 02:05 |
| mordred | speaking of - the merge-tags-back-to-master script didn't work because we were tryin to do something bad in the script which I beliveve we fixed | 02:06 |
| mordred | lifeless: I did not | 02:06 |
| lifeless | mordred: 'consolidating .mailmap? | 02:06 |
| lifeless | ' | 02:06 |
| mordred | fungi: a) that landed, right? and b) should we retrigger that job now that the script sucks less? | 02:06 |
| mordred | lifeless: hrm. yah. it's definitley still needed. copying from centrally certainly seems like a decent idea | 02:07 |
| mordred | especially since we have user to email mappings elsewhere too | 02:07 |
| lifeless | consolidating .mailmap? | 02:07 |
| mordred | yah | 02:07 |
| lifeless | bah, mouse fail | 02:07 |
| fungi | mordred: i can't remember what was broken in the tag merge script now, but pretty sure the fix did merge | 02:07 |
| lifeless | anyhow yeah, having 40 separate mailmaps seens like a pain | 02:08 |
| fungi | mordred: oh, right, we weren't merging to the right branch or something | 02:08 |
| mordred | lifeless: ++ | 02:08 |
| fungi | we were skipping milestone-proposed which was the only place it would have done anything | 02:08 |
| fungi | yeah | 02:08 |
| mordred | lifeless: similar to requirements update, we could write the sync script such that it only copies in relevant entries | 02:09 |
| lifeless | mordred: whats .mailmap used for ? | 02:09 |
| mordred | fungi: yup. because we were filtering on branch, but tag events don't have a branch | 02:09 |
| fungi | right-o | 02:09 |
| mordred | lifeless: most directly for AUTHORS file (removes duplicates) | 02:09 |
| lifeless | mordred: and AUTHORS is generated when /how ? | 02:09 |
| mordred | also, if people want their git log output to show | 02:09 |
| mordred | lifeless: sdist time | 02:10 |
| mordred | lifeless: by pbr | 02:10 |
| lifeless | mordred: ok, so if we copy mailmap in before sdist happens | 02:10 |
| lifeless | then we can do it lazily | 02:10 |
| lifeless | but I don't think we can otherwise, because race conditions | 02:10 |
| mordred | hrm. that would mean that a person running sdist locally would get a potentially surprising result | 02:10 |
| lifeless | you don't know who you need until they start contributing to that repo | 02:10 |
| mordred | I'm not sure how much I care | 02:10 |
| *** ArxCruz has quit IRC | 02:10 | |
| mordred | (about the local surprising result) | 02:11 |
| *** harlowja_at_home has joined #openstack-infra | 02:11 | |
| harlowja_at_home | fungi, u want to try my new zuul curses viewer?? :-P | 02:11 |
| harlowja_at_home | or others, https://github.com/harlowja/gerrit_view/#czuul :) | 02:11 |
| mordred | lifeless: if we did that, we might wind up shipping a large .mailmap file in all of our tarballs | 02:12 |
| mordred | lifeless: at least until we put it in as a global exclude | 02:12 |
| mordred | or - actually, it wouldn't e in git, so no we wouldn't | 02:12 |
| mordred | nevermind | 02:12 |
| *** rcleere has joined #openstack-infra | 02:12 | |
| *** ljjjustin has joined #openstack-infra | 02:14 | |
| fungi | mordred: out of curiosity, why does pbr manually remap from the mailmap file? git log already claims in its manpage to respect it | 02:15 |
| *** guohliu has quit IRC | 02:15 | |
| mordred | fungi: uhm. now that you ask that question, I do not know | 02:15 |
| fungi | man git-log briefly mentions it, but directs you to the git-shortlog manpage for details | 02:16 |
| fungi | harlowja_at_home: installing it into a venv now to take it for a spin | 02:18 |
| harlowja_at_home | :) | 02:19 |
| harlowja_at_home | cool | 02:19 |
| lifeless | mordred: we shouldn't ship .mailmap in tarballs anyway | 02:20 |
| lifeless | mordred: so meh | 02:20 |
| *** lnxnut has quit IRC | 02:20 | |
| harlowja_at_home | fungi, i could probably use screen space better, but pgup, pgdown, up/down arrow keys to move around (when u try it out) | 02:24 |
| openstackgerrit | Monty Taylor proposed a change to openstack-dev/pbr: Let git do mailmap mapping for us https://review.openstack.org/54050 | 02:25 |
| openstackgerrit | Monty Taylor proposed a change to openstack-dev/pbr: Ignore boring commit messages https://review.openstack.org/54048 | 02:25 |
| mordred | fungi: there you go. there's a patch to let git do the mailmap mapping for us | 02:25 |
| mordred | wow. merge conflict. that's so sexy | 02:28 |
| openstackgerrit | Monty Taylor proposed a change to openstack-dev/pbr: Make ChangeLog more condensed https://review.openstack.org/52367 | 02:30 |
| openstackgerrit | Monty Taylor proposed a change to openstack-dev/pbr: Ignore boring commit messages https://review.openstack.org/54048 | 02:30 |
| openstackgerrit | Monty Taylor proposed a change to openstack-dev/pbr: Let git do mailmap mapping for us https://review.openstack.org/54050 | 02:30 |
| mordred | and I had to rebase the stack. sigh | 02:31 |
| mordred | lifeless: if you want to review https://review.openstack.org/52367 and https://review.openstack.org/54048 I will not scream at you | 02:31 |
| mordred | I eam https://review.openstack.org/54050 not https://review.openstack.org/54048 - https://review.openstack.org/54048 is not ready yet | 02:32 |
| *** guohliu has joined #openstack-infra | 02:32 | |
| *** UtahDave has quit IRC | 02:32 | |
| *** bingbu has quit IRC | 02:32 | |
| *** rakhmerov has joined #openstack-infra | 02:32 | |
| mordred | fungi: also, there are several sexy git-review patches outstanding, :) | 02:34 |
| harlowja_at_home | sexy | 02:34 |
| *** michchap has quit IRC | 02:36 | |
| *** ljjjustin has quit IRC | 02:37 | |
| *** ljjjustin has joined #openstack-infra | 02:37 | |
| notmyname | mordred: still around? I see you're referencing me on twitter and proposing ideas on the mailing list. :-) | 02:39 |
| *** xchu has joined #openstack-infra | 02:39 | |
| mordred | notmyname: I constantly reference you on twitter | 02:39 |
| mordred | notmyname: although sadly it does not appear that that SAT solver is going to be immediately helpful in our current SAT solver interest | 02:40 |
| notmyname | ah. too bad | 02:40 |
| notmyname | cause auto requirements is pretty useless without it | 02:40 |
| mordred | yeah. the author has a rather weird hatred of python versioning | 02:40 |
| mordred | it's not useless. it just doesn't solve your use case | 02:40 |
| mordred | I do agree though that it can be improved to the point where it does | 02:40 |
| mordred | notmyname: in any case - yes! I'm still here | 02:41 |
| *** vipul has quit IRC | 02:41 | |
| notmyname | at best I'd call the current state "unwieldy", and boardering on useless. eg we may have some patches in eventlet that would be of interest, but that doesn't mean everyone needs to force upgrade to bleeding-edge eventlet (actual case, but substitute whatever package and get the same result) | 02:42 |
| notmyname | anyway, dependencies isn't what I wanted to focus on tonight. we've covered that plently before | 02:42 |
| *** vipul has joined #openstack-infra | 02:42 | |
| notmyname | and it seems the new thing you found doesn't meet our needs :-( | 02:43 |
| mordred | not yet. we'll get there | 02:43 |
| *** dims has quit IRC | 02:43 | |
| mordred | dstufft is still working on a SAT solver that will | 02:43 |
| notmyname | like you, I'm always hopeful | 02:43 |
| notmyname | first, the probably less contentious one: consolidated mailmap. wouldn't tarball makers care very much about the AUTHORS file? | 02:44 |
| notmyname | if for nothing else, it's necessary for making the same hash | 02:44 |
| notmyname | and it contains info that is otherwise not in the tarball (VCS info) | 02:45 |
| mordred | it's a good question | 02:45 |
| mordred | the question is - how important is the contents of AUTHORS to people making their own tarballs? | 02:45 |
| mordred | (I do not have a strong opinion on this one) | 02:46 |
| mordred | so it's an honest question | 02:46 |
| mordred | lifeless: ^^ (we're talking about you) | 02:46 |
| clarkb | mailmap is for deduping right? | 02:46 |
| clarkb | do we need to bother? | 02:46 |
| mordred | clarkb: yes. | 02:46 |
| clarkb | the list is correct without dedup | 02:46 |
| mordred | clarkb: sure. but it's a little ugly, and we do have the dedup machinery | 02:47 |
| mordred | I certainly don't think we should replace the mailmap files with nothing | 02:47 |
| clarkb | it seems to solve all the problems but uglyness :) | 02:47 |
| mordred | I think there are a few different options | 02:47 |
| *** harlowja_at_home has quit IRC | 02:48 | |
| notmyname | I'm a fan of keeping .mailmap local to a project, if for nothing else to keep the dependency trees shorter when building packages and running stuff | 02:48 |
| notmyname | (that's my opening position) | 02:48 |
| mordred | a) keep the individually managed mailmap files in the repos | 02:48 |
| mordred | b) have a central list from which mailmap entries can be synced by a computer as needed, with the results stored in the repo | 02:48 |
| *** ericw has quit IRC | 02:49 | |
| mordred | c) have a central list which is used at infra tarball creation time that does not have any relationship to the git repo | 02:49 |
| *** bingbu has joined #openstack-infra | 02:49 | |
| mordred | d) have a central list which is used at infra tarball creation time that does not have any relationship to the git repo but which is included in the tarball | 02:50 |
| mordred | any other combos? I'm missing? | 02:50 |
| mordred | a) is clearly the easiest, since it requires no changes, but is the most effort in terms of ongoing maint of the lists | 02:51 |
| notmyname | it doesnt' seem that git shortlog respects .mailmap being in .gitignore, FWIW | 02:51 |
| mordred | notmyname: what do you mean? | 02:52 |
| mordred | like, if you put it in gitignore, it doesn't see it? | 02:52 |
| notmyname | (b) seems to be the worst of both worlds: per project file + extra dependency and syncing problem | 02:52 |
| notmyname | mordred: as in even with it ignored, shortlog applies it | 02:52 |
| *** matsuhashi has quit IRC | 02:53 | |
| mordred | notmyname: can you explain what you mean by "extra dependency" ? | 02:53 |
| *** matsuhashi has joined #openstack-infra | 02:53 | |
| mordred | (I think th emore I think about it, btw, I agree that b is not a great choice - just want to make sure I'm capturing your thoughts) | 02:54 |
| notmyname | similar to pbr now being a build-time dependency (and I'm hopeful that one day it won't be). depends if this is separate from pbr or not. if part of pbr, then it's an extra "hook" that makes it harder to decouple later. if not, then it's a new dependency for builders/deployers | 02:54 |
| mordred | oh - I was TOTALLY thinking that this would be a jenkins related thing | 02:56 |
| mordred | that just copies a file in place before doing something | 02:56 |
| mordred | not that it would be something that setup.py knew anything about | 02:56 |
| notmyname | or pbr sdist hooks? | 02:56 |
| mordred | pbr already does the authors processing with the mailmap file | 02:56 |
| mordred | so all that is required is for the file to be present | 02:57 |
| mordred | if it's not, it's a no op | 02:57 |
| notmyname | FTFY: pbr already does what git shortlog does ;-) | 02:57 |
| mordred | notmyname: it actually is deferring to git to do the mailmap work in a commit above, fwiw | 02:57 |
| mordred | I think that running code in pbr to _fetch_ a global mailmap from somewhere else would be HORRIBLE | 02:57 |
| notmyname | good. we agree on that :-) | 02:57 |
| *** matsuhashi has quit IRC | 02:58 | |
| mordred | I believe the mechanism choices are "do we havea mechanism to mupdate the currently existing mailmap file automatically" | 02:58 |
| clarkb | notmyname: curious about the pbr not being a build time dependency. aiui that is pbr's reason for existence | 02:58 |
| mordred | or "do we inject a mailmap file on the build hosts" | 02:58 |
| mordred | if we do the second thing, it would mean that poeple other than us making sdist tarballs would have a different and possibly less featureful experience | 02:59 |
| notmyname | so we make an -infra jenkins-ish thingy to create a .mailmap file. how do people who want to know who did what get along? they have to pull from the centralized jenkins-ish thingy? | 02:59 |
| notmyname | clarkb: happy to talk about that, but I'm only smart enough for one fun conversation at a time ;-) | 02:59 |
| lifeless | mordred: you're talking about me? | 02:59 |
| mordred | notmyname: yah. that's ultimately the problem with not committing the changes to the repo | 03:00 |
| lifeless | mordred: I will review those things for you | 03:00 |
| mordred | lifeless: we're discussing your mailmap thoughts | 03:00 |
| notmyname | mordred: wouldn't git blame break too? | 03:00 |
| mordred | notmyname: does git blame use mailmap? | 03:00 |
| lifeless | mordred: I don't know that I have thoughts | 03:00 |
| lifeless | mordred: I just know that maintaining the same mapping in 40 repos is a waste of a good life | 03:00 |
| mordred | notmyname: so it does. that changes my opinion of the options | 03:00 |
| notmyname | not sure, but AFAIK all of git that looks at authors does | 03:00 |
| lifeless | mordred: if I was doing the work I'd be inclined to just copy the contents into every relevant repo, no editing | 03:01 |
| notmyname | mordred: so while a per-project .mailmap may have some duplicated info, keeping it the repo on a per-project basis may be the best | 03:01 |
| mordred | I think given that, it is important for the appropriate contents to be in the git repos | 03:01 |
| mordred | now - having a tool or script that could, if desired, grab a global openstack mailmap and copy it into your repo seems very simple | 03:02 |
| mordred | or, screw a tool | 03:02 |
| mordred | just having an accessible globla mailmap is all you'd need | 03:02 |
| mordred | the tool is wget :) | 03:02 |
| mordred | if you, as a project, wanted to copy it in and commit it - neat. if you don't? neat. | 03:02 |
| mordred | as this is not an operational set of information that is related to actual build things, I do not feel that we need to have a strong opinion on enforcing an automated sync | 03:03 |
| mordred | right? | 03:03 |
| lifeless | mordred: I don't follow | 03:03 |
| fungi | proposal job which submits a review to sync up your .mailmap file if entries in your git log contain authors in the central mailmap file? | 03:04 |
| notmyname | mordred: I support you having that opinion (while my own is stronger to keeping them per-project) | 03:04 |
| lifeless | mordred: there is a mapping which is needed to get correct behaviour from multiple tools (pbr, git). | 03:04 |
| mordred | that behavior is cosmetic at best | 03:04 |
| lifeless | mordred: the mapping should be the same for the same person in all git repos in OpenStack | 03:04 |
| mordred | lifeless: because notmyname and swift will actively become unhappy if jenkins started proposing mailmap changes | 03:05 |
| mordred | lifeless: I do not feel that it is worth a fight | 03:05 |
| *** guohliu has quit IRC | 03:05 | |
| lifeless | mordred: what would make them unhappy? | 03:05 |
| mordred | requirements is | 03:05 |
| notmyname | lifeless: "should be"? why? | 03:05 |
| mordred | it has crossproject functional effects on the testing of openstack | 03:05 |
| *** guohliu has joined #openstack-infra | 03:05 | |
| clarkb | notmyname: because gerrit | 03:05 |
| notmyname | mordred: well, I started this topic by saying it's less contentious | 03:05 |
| mordred | authors file contents is - less important | 03:05 |
| mordred | notmyname: :) | 03:05 |
| clarkb | gerrit does enforce it to a degree | 03:05 |
| notmyname | clarkb: hmm...good point. | 03:06 |
| lifeless | notmyname: what aspect of having the same mailmap as nova (either same for the same people, or same verbatim) would make you/swift unhappy? | 03:06 |
| mordred | lifeless: it would be commits that potentially suggest changes to their local file that are not relevant to their repo | 03:07 |
| notmyname | while I'd be mildly annoyed at having hundreds/thousands of lines in a .mailmap file in swift for people who have never even looked at the code, syncling that isn't worth too much of a fight at this point IMO | 03:07 |
| notmyname | and "syncling" is a weird-sounding word | 03:07 |
| mordred | I like it! | 03:07 |
| mordred | similarly, I have other fish to fry, so I don't feel strongly enough about it to make notmyname mildly annoyed. if we could sync only authors who have commits in the target repo... | 03:08 |
| mordred | then I doubt we'd see significant pushbach | 03:08 |
| lifeless | that surely doable | 03:08 |
| mordred | or pushback | 03:08 |
| notmyname | I'm happy to hear that the decision at this point is to auto-proposed updates or not. I'm fine to leave it there (I don't really care too much). I'm strongly against having a central dependency requirement to pull a .mailmap from | 03:09 |
| lifeless | just minor race condition (and strictly speaking no worse than the current often-stale dataset we have today) | 03:09 |
| *** dcramer_ has joined #openstack-infra | 03:09 | |
| *** matsuhashi has joined #openstack-infra | 03:09 | |
| fungi | what if it only proposed updates to the mailmap if authors in that project's git log were present in the global mailmap file? in the way that requirements sync only proposes updates for entries in your project's requirements lists? | 03:09 |
| mordred | notmyname: I 100% agree. pulling a dep would be awful | 03:09 |
| mordred | fungi: yes | 03:09 |
| mordred | lifeless: well, and if someone did choose to add themselves to a mailmap file in their first commit from a different email than their previous commit, it would not break things | 03:10 |
| notmyname | lifeless: fungi: what caused this question to arise? what's the pain point? | 03:10 |
| fungi | dryness | 03:10 |
| notmyname | not sure I follow | 03:10 |
| lifeless | notmyname: DRY; hidden non-discoverable datasources suck, etc | 03:10 |
| notmyname | gotcha | 03:10 |
| notmyname | IOW you don't want to keep your .mailmap entry up to date across N projects | 03:11 |
| lifeless | right | 03:11 |
| lifeless | plus several separate metric systems | 03:11 |
| lifeless | plus the foundation web page | 03:11 |
| lifeless | STAB STAB STAB STAB | 03:11 |
| * mordred fixes this problem by committing with the same email address everywhere | 03:12 | |
| notmyname | ya, the metrics and the foundation seems that they could benefit, but isn't that info already in gerrit anyway? shouldn't they pull from there rather than git shortlog? | 03:12 |
| clarkb | the foundation webpage is what got me to thinking we could just stop carring (there are tons of dups on the havana release page) | 03:12 |
| mordred | notmyname: oh - the metrics systems should pull from somewhere else for sure | 03:12 |
| lifeless | notmyname: well, I'd like to have just one source of all this info | 03:12 |
| mordred | clarkb: well, god only knows where the foundation got that data | 03:12 |
| mordred | it certainly wasn't from ANY system managed by the project | 03:12 |
| lifeless | notmyname: and clear instructions on the new contributor signup page saying 'do X to set this up, and Y if you change affiliation || email in future' | 03:12 |
| notmyname | lifeless: agreed. IMO the foundation (and gerrit) is or should be tracking all that and have the authoritative source | 03:13 |
| notmyname | lifeless: sure, but docs for new contribs don't have to spell out every eventuality. | 03:13 |
| lifeless | notmyname: so basically I'm bringing up any redundant source I see and going 'hey, this is perpetuating the overall problem' | 03:13 |
| fungi | gerrit doesn't make a lot of that queryable without being a gerrit admin, so we'd have to expose data from it through some additional mechanism | 03:13 |
| lifeless | notmyname: they don't, but there are only two situations. New, and changed. | 03:13 |
| mordred | fungi: once the user db system stops being stupid | 03:13 |
| lifeless | notmyname: point it new contributors don't even *know* that it's a thing at the moment. | 03:13 |
| fungi | mordred: yep | 03:14 |
| *** vipul is now known as vipul-away | 03:14 | |
| lifeless | notmyname: so a) reduce the number of things that we all need to care about, and b) then it's sane to tell people. | 03:14 |
| notmyname | lifeless: they don't know that what's a thing at the moment? .mailmap? | 03:14 |
| lifeless | notmyname: the big set of data sources; AFAICT there is no list of all the things that store this mapping data | 03:15 |
| lifeless | notmyname: and there should need to be; it should be maintained in one place | 03:15 |
| notmyname | except the currently-hard-to-get-but-it's-there gerrit system | 03:15 |
| mordred | I'm going to push VERY hard for there to be a central, authoitative and accessilbe source for this this cycle | 03:15 |
| notmyname | which actually has the CLA info etc | 03:15 |
| mordred | because right now if we did something it would be ADDING another source to the list | 03:15 |
| mordred | rather than reducing the number | 03:16 |
| *** gyee has joined #openstack-infra | 03:16 | |
| lifeless | notmyname: gerrit is one such system | 03:16 |
| mordred | but if we get the membership system to stop sucking, then we can build thigns that can query it when we need, etc | 03:16 |
| notmyname | ok. /me is done talking about sync'ing .mailmap files until mordred fixes the world ;-) | 03:16 |
| lifeless | notmyname: but it's not authoritative; *that* is the problem. | 03:16 |
| *** gyee has quit IRC | 03:16 | |
| notmyname | mordred: next thing I want to talk about is changelog files | 03:17 |
| mordred | notmyname: awesome! | 03:17 |
| fungi | it's almost like he's reading through your list of pending pbr patches | 03:17 |
| notmyname | mordred: (I say this because we know each other and in this rare case I think it will translate in a text-based medium) filtering "boring" commit messages from an auto-generated changelog seems like such a waste of time. go solve a real problem | 03:18 |
| notmyname | /rant | 03:18 |
| mordred | notmyname: only thought about it at all because I was right there in the code anyway | 03:19 |
| * notmyname is a strong proponent of hand-curated CHANGELOG files | 03:19 | |
| notmyname | I agree that changelogs that are simply the result of `git shortlog` are useless | 03:19 |
| mordred | well, we don't use git shortlog | 03:19 |
| notmyname | well, "the git changelog" | 03:20 |
| notmyname | from your email | 03:20 |
| mordred | gotcha. | 03:20 |
| *** guohliu has quit IRC | 03:21 | |
| mordred | in general, rant noted. I do not plan on spending much energy on the topic (it's good plane hacking fodder, tbh) | 03:21 |
| * mordred WANTS to be hacking on wheel support, but that requires way less laggy network than he has now | 03:21 | |
| notmyname | the audience for CHANGELOG is people who need to see what's the high-level status of a project, what's important, and what to watch out for. it's an executive summary for the deployer, packager, and admin | 03:21 |
| notmyname | auto-generated files can't do that | 03:21 |
| mordred | I kinda think of NEWS files or release notes as being that | 03:22 |
| mordred | but honestly, on the fast moving projects, manually curated changelog entries would be very difficult to manage | 03:24 |
| mordred | either you're adding entries as you go, in which case you're increasing merge conflict fodder | 03:24 |
| mordred | or you're adding them at release time, which means the file is out of date in the repo until then, and someone has to track things for longer | 03:25 |
| notmyname | IOW the changelog doesn't have info about a release until the release happens? | 03:25 |
| notmyname | for the 2nd point | 03:25 |
| mordred | right. | 03:25 |
| mordred | but for people running from trunk | 03:25 |
| notmyname | how is that a problem? | 03:25 |
| mordred | you grab it, and you look for the curated information, since git log is not good enough for deployers | 03:26 |
| mordred | but it's not filled in, because of merge problems | 03:26 |
| notmyname | if a deployer is running at trunk, they have the chops to see what's going on since the last release was cut | 03:26 |
| mordred | how can anyone run openstack and NOT have those chops? | 03:27 |
| mordred | in a real prod env | 03:27 |
| notmyname | they run someone's product/distro/bundling of it | 03:28 |
| notmyname | piston/nebula/red hat/mirantis/cloudscaling/etc | 03:28 |
| mordred | in which case, they are reading the product release notes | 03:28 |
| mordred | from the value add company | 03:28 |
| lifeless | mordred: release time doesn't exist | 03:28 |
| mordred | who is curating things for them | 03:28 |
| lifeless | mordred: its a fallacy. | 03:28 |
| mordred | lifeless: it does. we're not to your world yet | 03:28 |
| lifeless | mordred: enough folk are that from a planning perspective... | 03:29 |
| mordred | OpenSTack makes 6 month releasesuntil we decide other wise | 03:29 |
| lifeless | mordred: yes, it does. | 03:29 |
| notmyname | lifeless: true :-) when are we going to do away with this whole release cycle thing anyway? ;-) | 03:29 |
| lifeless | mordred: but 'fix stuff up at the end' is already not an option | 03:29 |
| mordred | lifeless: indeed. which is another reason that write the changelog at release time is a no-go | 03:29 |
| lifeless | notmyname: J timeframe I hope | 03:29 |
| mordred | essentially, it's an even more complex issue than finding the right time to land the commit that changes the version number - the real process just doesn't work that way | 03:30 |
| mordred | which means manual curation would be back to a thing needing to be done continaully | 03:30 |
| mordred | which has merge conflict issues big-time | 03:31 |
| mordred | due to the form of the thing | 03:31 |
| fungi | run-parts style changelog assembly ;) | 03:31 |
| mordred | fungi: boom! you win! :) | 03:31 |
| notmyname | mordred: our goal for swift now is that master is deployable at any point (we've missed this recently, sadly). every so often, we bundle together what's happened and cut a "release". at this point I curate the AUTHORS and CHANGELOG file from what's happened (BTW, this happens much more often than every 6 months) | 03:31 |
| mordred | notmyname: right. because YOU have a sooner than 6 month release cycle | 03:32 |
| mordred | curating that file for you is not nearly as much of a nightmare | 03:32 |
| mordred | imagine curating it for a nova release | 03:32 |
| notmyname | mordred: well, importantly for conflicts sake, I'm the one who does it | 03:32 |
| mordred | right | 03:32 |
| mordred | you also probably actually know everytihng that landed in swift in a cycle | 03:33 |
| *** guohliu has joined #openstack-infra | 03:33 | |
| notmyname | mordred: only if I keep on top of it nearly every day | 03:33 |
| notmyname | I get very far behind on all the trips ;-) | 03:33 |
| mordred | oy. tell me about it | 03:33 |
| mordred | notmyname: so you're saying you keep up a set of notes about what the changelog is going to be? | 03:34 |
| notmyname | perhaps. or even have the PTL work on in as it goes | 03:34 |
| notmyname | point being is that russellb does a great job knowing what's going on. I guarantee you that any PTL can tell you what's going on at the drop of a hat. personally, I have to look through the VCS logs at every release to know everything that's merged | 03:35 |
| mordred | nod | 03:35 |
| notmyname | I've done both: updating it at release time and keeping it updated during a release | 03:35 |
| mordred | well, I mean, ultimately skip_changelog is an option in setup.cfg for a reason | 03:36 |
| notmyname | either way, the PTL is the one updating it and making sure that there is a good high-level summary for stuff. this is especially important with features that cross many commits | 03:36 |
| notmyname | mordred: it's not there just because you like me so much? ;-) | 03:36 |
| mordred | nothing is preventing anyone from manually curating one - but if that is not a thing a person wants to do, making one that contains _something_ that might be informative is better than no information in the tarball | 03:37 |
| mordred | notmyname: well, there are two reasons :) | 03:37 |
| notmyname | lo | 03:37 |
| notmyname | lol | 03:37 |
| mordred | bcwaldon also preferred manaul curation back when he still hacked on openstack | 03:37 |
| notmyname | I've heard interest from jgriffith too | 03:38 |
| mordred | but if we can get best-effort from automation and it's good enough for a wide variety of things, then awesome, one less thing for folks | 03:38 |
| mordred | jgriffith: you are more than welcome to turn off ChangeLog generation in your repo should you like | 03:38 |
| mordred | altohugh, what bcwaldon wound up doing was this: http://docs.openstack.org/developer/python-glanceclient/#release-notes | 03:39 |
| mordred | maintaining the curated thing in the sphinx docs | 03:39 |
| mordred | and then having the ChangeLog file be the autogenerated thing | 03:39 |
| *** nati_ueno has quit IRC | 03:40 | |
| mordred | that way there is also, in addition ot the tarball, a published web version of the notes | 03:40 |
| mordred | that can be linked places | 03:40 |
| notmyname | do you worry that an autogenerated changelog may give too much weight to "raw" data? IOW, if it's not there and someone has to go digging in VCS for it, they kinda know that thye have to figure out what the big stuff is. that implicit understanding is gone with an easily available changelog file that is essentially git shortlog output | 03:40 |
| *** lnxnut has joined #openstack-infra | 03:40 | |
| *** nati_ueno has joined #openstack-infra | 03:40 | |
| mordred | I'm not sure I follow the concern | 03:41 |
| mordred | I also, btw, think that git is a pretty frequently used vehicle from which to get the source code | 03:41 |
| mordred | and honestly think that changelog files are a dinosaur from a time when that was not the case | 03:41 |
| mordred | and when the tarball was actually the only mechanism you had to get the code, and the VCS history was simply not available | 03:42 |
| notmyname | so you've got a big feature that spans a few commits or one that landed early in the cycle. take a look at the ChangeLog (that was generate from git shortlog), and you've got to wade through hundreds to thousands of messages before you get to the big stuff | 03:42 |
| mordred | sure. that's why I actually like http://docs.openstack.org/developer/python-glanceclient/#release-notes as a compromise for the folks who want to maintain it | 03:43 |
| notmyname | ya, I guess my point is that hand-curated is awesome and should be required. if that's in sphinx or the actual CHANGELOG file is someone less important (especially if considering greenfield projects) | 03:44 |
| notmyname | I don't make a hand-curated changelog for python-swiftclient. I should. | 03:45 |
| mordred | yeah. I can totally see that, and I would not be unhappy if everyone started doing a release notes style hand curated thing | 03:45 |
| *** nati_ueno has quit IRC | 03:46 | |
| notmyname | mordred: so where does that leave us? please don't work on filtering changelog mesages. at least until you figure out the really important stuff. like vim modelines in files | 03:47 |
| mordred | notmyname: :) | 03:47 |
| mordred | notmyname: I did not bring that one up | 03:47 |
| notmyname | I know :-) | 03:47 |
| mordred | but I've automated all of the things that were the source of my commits in the last cycle - I've got some up with _something_ to keep my commit count high! ;) | 03:48 |
| mordred | else people might think I've turned into a manager | 03:48 |
| notmyname | horror! | 03:48 |
| notmyname | clarkb: so pbr as not a build-time dependency... | 03:48 |
| mordred | speaking of - python-glanceclient does not seem to be using the openstack sphinx theme... | 03:48 |
| * mordred sits back to watch this one | 03:49 | |
| clarkb | ya, mostly just curious how that would work | 03:49 |
| clarkb | since pbr is python build reasonableness | 03:49 |
| notmyname | clarkb: in my ideal world, pbr would be a tool that is used to build stuff (specifically for pip, not packages), similar to debuild for debian. ie feed a repo in and get out a thing that can be installed | 03:49 |
| notmyname | I think it's most egregious in client libraries (where it becomes another dependency to install when especially for clients we should be _reducing_ these things) | 03:50 |
| mordred | notmyname: that's actually possible, but I thnk it would REALY piss off the packagers working from git and not from tarballs we generate | 03:51 |
| notmyname | mordred: how so? | 03:51 |
| clarkb | notmyname: I think we may have some philosophical differences about that :) (I think swiftclient should include its "optional" dependencies at least when installing from pip | 03:51 |
| mordred | notmyname: because it would be a tool you'd HAVE to use to interact with the source code repository | 03:51 |
| mordred | whereas right now you can interact with setup.py like other python projects do | 03:52 |
| notmyname | mordred: it wouldn't be `pbr /path/to/some/git/repo`? | 03:52 |
| mordred | yup. THAT would work | 03:52 |
| mordred | but you'd need pbr at build time still | 03:52 |
| notmyname | clarkb: python-*client should run anywhere. our APIs are HTTP! we should be runnable on py2.0 on every computer ever made | 03:53 |
| mordred | notmyname: we are | 03:53 |
| mordred | pbr does not prevent that | 03:53 |
| clarkb | notmyname: I agree, but that doesn't rule out dependencies | 03:53 |
| mordred | it ONLY prevents using pre-secure pip | 03:53 |
| mordred | and I do not feel the need to support that | 03:53 |
| mordred | as it's a horrible idea | 03:53 |
| mordred | and actively destructive | 03:53 |
| notmyname | mordred: why would you need pbr for build time? | 03:54 |
| mordred | in any case, with the tool approach | 03:54 |
| mordred | you could remove the need for pbr at pip install time | 03:54 |
| mordred | BUT | 03:54 |
| mordred | you'd increase the surface area of people who woul dhave to learn about pbr directly | 03:54 |
| notmyname | clarkb: sure, but as a general idea in client libs (for any project), less deps >> more deps | 03:54 |
| mordred | because every single person wanting to install from git | 03:54 |
| mordred | would have to learn about running pbr | 03:55 |
| clarkb | notmyname: but broken features isn't good either | 03:55 |
| notmyname | mordred: that's what I don't get yet. why? I don't need debuild to run from source on a debian machine? | 03:55 |
| clarkb | it is really ocnfusing when you just want swiftclient to push objects to $publiccloud and it can't in a default install | 03:55 |
| notmyname | clarkb: nobody is promoting broken features | 03:55 |
| mordred | you DO need it to build debian packages though | 03:55 |
| mordred | and the way pip and friends work right now | 03:56 |
| mordred | is via source-level installs | 03:56 |
| mordred | so it's not analogous to how debuild/dpkg work | 03:56 |
| mordred | once we start using wheels, btw | 03:56 |
| mordred | you will not need pbr on your machien to install things | 03:56 |
| mordred | so, actually, that ^^ is probably what you really want | 03:57 |
| notmyname | could you say that when we have wheels the install process will be on rails? | 03:57 |
| mordred | oy | 03:57 |
| * notmyname doesn't actually know what "wheels" in this context means | 03:57 | |
| mordred | at that point, you will not need "build" depends to install | 03:57 |
| mordred | notmyname: it's a new binary distribution mechanism for python | 03:57 |
| mordred | it allows us to build "built" distributions of software | 03:58 |
| mordred | that can be installed without running setup.py | 03:58 |
| notmyname | gems? | 03:58 |
| *** harlowja has quit IRC | 03:58 | |
| mordred | I believe gems still run ruby code at install time | 03:58 |
| mordred | this is something that can run _no_ python code at install time | 03:59 |
| notmyname | I don't actually know much about how the ruby word works.. | 03:59 |
| mordred | me either | 03:59 |
| mordred | the upshot is, the only people who will need pbr | 03:59 |
| notmyname | interesting...so where is it supported? | 03:59 |
| mordred | pip >= 1.4 | 03:59 |
| mordred | so, first step is | 03:59 |
| mordred | we're going to start to upload wheels (and only wheels) of pre-release versions of things to pypi | 04:00 |
| *** bingbu has quit IRC | 04:00 | |
| mordred | (this goes hand in hand with pip 1.4 not downloading pre-release software by default) | 04:00 |
| mordred | which means we'll have by-default non-downloaded pre-release things no matter what version of pip you happen to be running | 04:00 |
| mordred | since pip pre 1.4 will not see the wheels at all | 04:01 |
| notmyname | "pre-release versions" == every commit? or == milestones? | 04:01 |
| mordred | milestones. things like oslo.config==1.2b4 | 04:01 |
| *** fifieldt has joined #openstack-infra | 04:01 | |
| notmyname | so "releases" that aren't part of the six-month cycle, essentially | 04:01 |
| notmyname | what about client libs? | 04:01 |
| mordred | client lib releases are releases | 04:02 |
| mordred | so for those, we'll upload both things | 04:02 |
| *** changbl has joined #openstack-infra | 04:02 | |
| notmyname | and what happens when pip >1.4 sees a wheel? | 04:02 |
| notmyname | oh, ok. go on | 04:02 |
| mordred | so, when pip >1.4 sees a wheel, if you have set PIP_USE_WHEEL or --use-wheel, it will prefer the wheel over the tarbal | 04:02 |
| mordred | (there are still flags around this because it's still a new feature) | 04:03 |
| mordred | and if there is a wheel that matches the system on which you are running, it will download the wheel and just unpack it directly | 04:03 |
| mordred | rather than running the setup.py code that was in the source repo | 04:03 |
| notmyname | ok. and what about earlier versions of pip? | 04:03 |
| mordred | (setup.py gets run at wheel creation time, similar to debuild) | 04:03 |
| mordred | earlier versions of pip will default to grabbing the tarball as usual | 04:04 |
| notmyname | ok, so pbr would be needed for "wheel builders"? | 04:04 |
| mordred | yes | 04:04 |
| notmyname | or pip pre 1.4 | 04:04 |
| mordred | yes | 04:04 |
| notmyname | when is pip 1.4 released? | 04:04 |
| mordred | a few months ago | 04:05 |
| mordred | it's the version of pip in latest ubuntu | 04:05 |
| mordred | and, if you are using pip, it's the version you should be using | 04:05 |
| mordred | there is no valid reason to use pip < 1.4 | 04:05 |
| notmyname | "should" is often not what our users are doing | 04:05 |
| mordred | because, by definitoin, if you are using it, you are using it to install software via pip | 04:05 |
| mordred | so you, by definition, are ok with instaling softare via pip :) | 04:06 |
| *** lnxnut has quit IRC | 04:06 | |
| mordred | well, for regular users, installing one of our things right now on old pip will give you a dependency error saying to upgrade your pip | 04:06 |
| mordred | which is pretty clear | 04:06 |
| notmyname | yup | 04:06 |
| *** lnxnut has joined #openstack-infra | 04:06 | |
| notmyname | I'm quite aware of that one ;-) | 04:06 |
| mordred | I could investigate figuring out a way to print a more verbose warning | 04:07 |
| mordred | which explains "seriously, dude. upgrade pip. you are opening yourself to being rooted right now" | 04:07 |
| notmyname | even when building packages and there is an older version of pip installed (even if you aren't using it...) | 04:07 |
| mordred | that should not be the behavior | 04:07 |
| mordred | if you are building packages | 04:08 |
| mordred | and if it is, it's a bug | 04:08 |
| notmyname | right. packagers should use the pbr env flags | 04:08 |
| notmyname | is that what you mena? | 04:08 |
| mordred | actually, they should not be necessary | 04:08 |
| mordred | it turns out | 04:08 |
| *** primeministerp has quit IRC | 04:08 | |
| mordred | as dh_python already passes --single-versoin-externaly-managed and I _think_ that should triger the SKIP_PIP_INSTALL codepath as well | 04:09 |
| mordred | it's on my todo list to tighten that up and get more tests to ensure that it is the case | 04:09 |
| mordred | but _certainly_ the packagers can be explicit and use the var | 04:09 |
| notmyname | mordred: on lucid, not using pip (but distro version installed), `python ./setup.py develop` doesn't work because pip isn't new enough | 04:10 |
| mordred | that is correct | 04:10 |
| mordred | because pip is a requirement of pbr | 04:10 |
| *** dcramer_ has quit IRC | 04:10 | |
| mordred | SKIP_PIP_INSTALL=1 python setup.py develop should work on lucid though | 04:11 |
| *** lnxnut has quit IRC | 04:11 | |
| mordred | I do not understand who the theoretical lucid user we're targetting here is though | 04:11 |
| mordred | so it's hard to design a solution for them | 04:11 |
| notmyname | in this case, it's me since my precise VM is hosed after my OS X update | 04:12 |
| notmyname | and I have a lucid VM in the cloud already | 04:12 |
| mordred | heh. good. well, I assume you're smarter than average | 04:12 |
| mordred | so that's helpful | 04:12 |
| *** bingbu has joined #openstack-infra | 04:13 | |
| *** dripton_ has quit IRC | 04:13 | |
| notmyname | nope. the env var doesn't help | 04:13 |
| * mordred looking for his lucid image | 04:14 | |
| mordred | notmyname: that not working i s a bug | 04:14 |
| notmyname | hmm..pbr isn't installed at all on this box | 04:14 |
| notmyname | maybe a bootstrapping thing? | 04:14 |
| mordred | do you have a pbr egg in that directory? | 04:15 |
| notmyname | I love the tracebacks with this line: "raise VersionConflict(dist,req) # XXX put more info here" | 04:15 |
| mordred | yah | 04:15 |
| notmyname | mordred: no. no traces of pbr that I can find | 04:15 |
| *** oubiwann has quit IRC | 04:15 | |
| mordred | it's very odd that any code is executing then, as setup.py has no functionality at all without pbr | 04:16 |
| *** wenlock has quit IRC | 04:16 | |
| *** yamahata has quit IRC | 04:16 | |
| mordred | as in, will fail to operate with an argument error | 04:16 |
| mordred | I dont' seem to have my lucid node any more - lemme spin one up | 04:16 |
| notmyname | setuptools is there and will execute. that's the call that setup.py is making | 04:17 |
| *** matsuhashi has quit IRC | 04:17 | |
| *** yamahata has joined #openstack-infra | 04:17 | |
| *** rcleere has quit IRC | 04:18 | |
| notmyname | to the cloud! | 04:18 |
| *** sdake has quit IRC | 04:18 | |
| mordred | but setuptools can't do anything without pbr | 04:18 |
| *** ryanpetrello has joined #openstack-infra | 04:18 | |
| mordred | the first thing it should try to do is download it because of setup_requires | 04:18 |
| mordred | notmyname: which repo are you doing, btw | 04:19 |
| notmyname | swift | 04:19 |
| mordred | k | 04:19 |
| mordred | and you've installed all of swift's depends by packages? | 04:19 |
| mordred | is ther e a place I can copy pate a command to do that (want to duplicate env) | 04:20 |
| notmyname | It's been a long time since I've logged into this box. I'm honestly not sure | 04:20 |
| *** matsuhashi has joined #openstack-infra | 04:20 | |
| notmyname | I copied the code over to the lucid box, cd'd to the dir, and `sudo python ./setup.py develop` | 04:20 |
| mordred | wow. SO OLD SO OLD | 04:22 |
| mordred | ubuntu@lucidtest:~$ sudo apt-get install git-core | 04:22 |
| mordred | :) | 04:22 |
| notmyname | yup | 04:23 |
| *** matsuhashi has quit IRC | 04:23 | |
| jgriffith | mordred: notmyname only catching part of the context in scroll back... | 04:23 |
| jgriffith | mordred: notmyname changelog generation? | 04:23 |
| notmyname | jgriffith: once upon a time you mentioned that a curated might be nice | 04:23 |
| jgriffith | notmyname: ahh... yes, likely while I was updating release notes for python-cinderclient | 04:24 |
| notmyname | jgriffith: mordred was mentioning that you can easily turn off the auto-generation for cinder ifyou want | 04:24 |
| jgriffith | notmyname: or release notes for cinder G | 04:24 |
| mordred | jgriffith: there are two optoins for that - you can either turn off auto generatoin - or you can do them in sphinx docs like glanceclient: http://docs.openstack.org/developer/python-glanceclient/#release-notes | 04:24 |
| mordred | notmyname: ok. I see the bug. I will fix it | 04:25 |
| notmyname | mordred: ah really? I was sure I was doing something wrong | 04:25 |
| jgriffith | mordred: so I kinda decided I liked having them in sphinx and published on pypi | 04:25 |
| mordred | notmyname: nope. it's a bug. I apologize | 04:25 |
| jgriffith | for the client at least | 04:25 |
| mordred | jgriffith: awesome | 04:25 |
| jgriffith | mordred: notmyname auto-magically generating is the topic I think we talked about | 04:25 |
| jgriffith | mordred: notmyname that would still be quite nice | 04:26 |
| mordred | jgriffith: wait - what do you want? | 04:26 |
| jgriffith | mordred: haha... depends on when you ask | 04:26 |
| mordred | hahahaahahahaha | 04:26 |
| jgriffith | mordred: so... I *think* IIRC the conversation with notmyname was around trying to post process the release notes | 04:26 |
| jgriffith | and we had talked about a way to aut-generate off of say like BP or Bug tags in the commit | 04:27 |
| jgriffith | as it stands now... for cinder-client I go through commit history and manually try and fill in what's happened since the last release | 04:27 |
| jgriffith | in docs/index.rst | 04:28 |
| jgriffith | which I like because it shows up in pypi when you browse to the package | 04:28 |
| *** matsuhashi has joined #openstack-infra | 04:28 | |
| jgriffith | err...hmmm | 04:28 |
| * jgriffith lies | 04:29 | |
| jgriffith | it does not | 04:29 |
| jgriffith | I would have to modify the README for that | 04:31 |
| jgriffith | anyway... not sure if that's the context notmyname was recalling or not? | 04:31 |
| jgriffith | bahhh!! | 04:31 |
| jgriffith | mordred: continuous upgrade session conflicts :( | 04:31 |
| notmyname | jgriffith: I remember something much more general about just how changelogs are done in swift. but I need to start handling the swiftlclient changelog, and I'm leaning towards the sphinx docs like you use | 04:32 |
| *** primeministerp has joined #openstack-infra | 04:33 | |
| jgriffith | notmyname: cool... I think I was complaining to you late one night when I was trying to pull everything together :) | 04:33 |
| notmyname | jgriffith: I think we all do that every time any of us sees someone else in person ;-) | 04:34 |
| jgriffith | notmyname: LOL... indeed! | 04:34 |
| mordred | notmyname: you havea found a fascinating edge case | 04:37 |
| notmyname | mordred: oh goody | 04:37 |
| *** yamahata has quit IRC | 04:40 | |
| mordred | notmyname: it might be unfixable | 04:41 |
| notmyname | mordred: what's the issue? need pbr to resolve but pbr isn't installed? | 04:42 |
| mordred | nope. it's that setuptools itself is reading the requires list out of the pbr egg it downloads from EGG-INFO/requires.txt | 04:42 |
| mordred | which means that pbr's suppression of things hasn't been gotten to yet | 04:43 |
| * mordred trying a few evil things real quick | 04:43 | |
| notmyname | mordred: kk. which would also be an issue for package building since that would require setup.py to be run | 04:44 |
| mordred | actually, not really | 04:44 |
| mordred | package building gets build deps via packages | 04:44 |
| notmyname | ah, right | 04:45 |
| mordred | which means that the easy_install codepath in setuptools is not triggered | 04:45 |
| mordred | it's setuptools easy_install that is the culprit in everything | 04:45 |
| notmyname | hmm...but even then, packaging (or at least the lucid packaging I know) still uses setup.py for sphinx docs generation. so that would trigger it, at least | 04:47 |
| mordred | yes. but that's fine | 04:47 |
| notmyname | not sure of other setup.py references | 04:47 |
| mordred | becuase in _that_ context you're driving the setup.py in the conetxt of the SKIP_PIP_INSTALL | 04:47 |
| mordred | (there is a patch I need to make still - but different problem) | 04:48 |
| mordred | at least, I think I need to make it - I have a patch which will fix a thing I _thought_ was your problem :) | 04:48 |
| mordred | but that's not your problem | 04:48 |
| *** reed has quit IRC | 04:49 | |
| notmyname | mordred: thanks for looking | 04:53 |
| mordred | notmyname: I'm going to keep beating my head against it, because the sequence doesn't make me happy | 04:53 |
| *** ljjjustin has quit IRC | 04:53 | |
| lifeless | mordred: so, why does pbr need to be the place ChangeLog generation happens in? | 04:53 |
| mordred | but the best thing I can come up with is to remove the version specification from pip in requirements.txt | 04:54 |
| lifeless | mordred: like, why couldn't be in (say) pbr_utils which pbr arranges for install of ? | 04:54 |
| mordred | lifeless: hang on - lemme finish this thought | 04:54 |
| *** vipul-away is now known as vipul | 04:54 | |
| *** ljjjustin has joined #openstack-infra | 04:55 | |
| mordred | notmyname: I worry about doing that because while it will allow you to skip past your current thing, it widens the possibility that people will operate software with a known insecure version of things | 04:55 |
| mordred | notmyname: I'm landing - let me noodle on it further and see what the options might be | 04:56 |
| notmyname | mordred: kk. thanks again. (I'm out for the night too) | 04:56 |
| *** senk has quit IRC | 05:06 | |
| *** nati_ueno has joined #openstack-infra | 05:15 | |
| *** nicedice__ has quit IRC | 05:16 | |
| *** yamahata has joined #openstack-infra | 05:22 | |
| *** nosnos_ has joined #openstack-infra | 05:37 | |
| *** nosnos has quit IRC | 05:37 | |
| *** SergeyLukjanov has joined #openstack-infra | 05:44 | |
| mordred | notmyname: my pleasure! let me know when you're back on, I have further thoughts | 05:54 |
| mordred | lifeless: morning! | 05:54 |
| *** matsuhashi has quit IRC | 05:56 | |
| *** matsuhashi has joined #openstack-infra | 05:57 | |
| *** katyafervent has joined #openstack-infra | 06:00 | |
| mordred | lifeless: I now have the bandwidth to discuss pbr and changelog generation | 06:01 |
| *** denis_makogon has joined #openstack-infra | 06:01 | |
| openstackgerrit | Monty Taylor proposed a change to openstack-dev/pbr: Skip the install of pip https://review.openstack.org/54061 | 06:02 |
| *** mattymo has joined #openstack-infra | 06:02 | |
| openstackgerrit | Monty Taylor proposed a change to openstack-dev/pbr: Make ChangeLog more condensed https://review.openstack.org/52367 | 06:05 |
| openstackgerrit | Monty Taylor proposed a change to openstack-dev/pbr: Let git do mailmap mapping for us https://review.openstack.org/54050 | 06:05 |
| openstackgerrit | A change was merged to openstack-infra/config: Add havana branch to gerritbot for #openstack-doc https://review.openstack.org/53192 | 06:11 |
| openstackgerrit | A change was merged to openstack-infra/config: add stable/grizzly for openstack-chef cookbooks https://review.openstack.org/52416 | 06:12 |
| *** davidhadas_ has quit IRC | 06:16 | |
| *** tian has joined #openstack-infra | 06:18 | |
| lifeless | mordred: cool | 06:46 |
| *** vipul is now known as vipul-away | 06:57 | |
| *** davidhadas has joined #openstack-infra | 07:01 | |
| *** matsuhashi has quit IRC | 07:02 | |
| *** matsuhashi has joined #openstack-infra | 07:02 | |
| *** amotoki has joined #openstack-infra | 07:03 | |
| *** nosnos_ has quit IRC | 07:05 | |
| *** ryanpetrello has quit IRC | 07:05 | |
| *** nosnos has joined #openstack-infra | 07:05 | |
| *** matsuhashi has quit IRC | 07:07 | |
| *** matsuhashi has joined #openstack-infra | 07:07 | |
| rakhmerov | mordred: could you please have a look at https://review.openstack.org/#/c/51205? Most likely you'll disapprove it for a couple of things, but I want to make sure before sending another patchset :) | 07:09 |
| *** ljjjustin has quit IRC | 07:10 | |
| *** ben_duyujie has joined #openstack-infra | 07:16 | |
| *** ljjjustin has joined #openstack-infra | 07:21 | |
| *** ljjjustin has quit IRC | 07:28 | |
| *** davidhadas_ has joined #openstack-infra | 07:32 | |
| *** davidhadas_ has quit IRC | 07:32 | |
| *** davidhadas has quit IRC | 07:33 | |
| *** davidhadas has joined #openstack-infra | 07:34 | |
| *** DinaBelova has joined #openstack-infra | 07:44 | |
| *** dkliban has joined #openstack-infra | 07:49 | |
| *** flaper87|afk is now known as flaper87 | 08:01 | |
| *** dkliban has quit IRC | 08:02 | |
| *** ben_duyujie has quit IRC | 08:02 | |
| *** davidhadas_ has joined #openstack-infra | 08:02 | |
| *** davidhadas has quit IRC | 08:02 | |
| *** nati_ueno has quit IRC | 08:09 | |
| *** matsuhashi has quit IRC | 08:09 | |
| *** nati_ueno has joined #openstack-infra | 08:09 | |
| *** matsuhashi has joined #openstack-infra | 08:10 | |
| *** matsuhas_ has joined #openstack-infra | 08:12 | |
| *** tkammer_ has joined #openstack-infra | 08:12 | |
| *** matsuhashi has quit IRC | 08:13 | |
| *** nati_ueno has quit IRC | 08:13 | |
| *** yamahata has quit IRC | 08:19 | |
| *** marun has joined #openstack-infra | 08:22 | |
| *** che-arne has joined #openstack-infra | 08:24 | |
| ogelbukh | yay, now we have the change request: https://review.openstack.org/#/c/53858/ | 08:26 |
| *** boris-42 has quit IRC | 08:28 | |
| *** che-arne has quit IRC | 08:32 | |
| *** dizquierdo has joined #openstack-infra | 08:43 | |
| *** alexpilotti has joined #openstack-infra | 08:43 | |
| *** Bada has joined #openstack-infra | 08:45 | |
| *** marun has quit IRC | 08:47 | |
| *** denis_makogon has quit IRC | 08:49 | |
| openstackgerrit | Andreas Jaeger proposed a change to openstack-infra/config: Split Install Guide into Debian and Ubuntu guides https://review.openstack.org/54076 | 08:52 |
| *** hashar has joined #openstack-infra | 08:54 | |
| *** DinaBelova has quit IRC | 09:02 | |
| *** marun has joined #openstack-infra | 09:06 | |
| *** yassine has joined #openstack-infra | 09:12 | |
| openstackgerrit | Dzmitry Horbach proposed a change to openstack-infra/jenkins-job-builder: Added default value of -1 for all logrotate attributes https://review.openstack.org/54079 | 09:19 |
| *** ljjjustin has joined #openstack-infra | 09:19 | |
| *** fallenpegasus has quit IRC | 09:22 | |
| *** fbo_away is now known as fbo | 09:23 | |
| *** fallenpegasus has joined #openstack-infra | 09:27 | |
| katyafervent | mordred, Hi) I have tiny question) Do we need someone else approval or it may happened that somebody forget to check approved?) Please take a look here https://review.openstack.org/#/c/52092/ | 09:28 |
| ekarlso- | Anyone care to sign off on this ? https://review.openstack.org/#/c/51066/ | 09:30 |
| ekarlso- | it's been there for 2 weeks and is pretty small :) | 09:30 |
| *** ljjjustin has quit IRC | 09:32 | |
| *** bingbu has quit IRC | 09:32 | |
| *** dkliban has joined #openstack-infra | 09:34 | |
| *** boris-42 has joined #openstack-infra | 09:36 | |
| *** dizquierdo has quit IRC | 09:36 | |
| *** saschpe has quit IRC | 09:36 | |
| *** saschpe has joined #openstack-infra | 09:40 | |
| *** che-arne has joined #openstack-infra | 09:40 | |
| *** xchu has quit IRC | 09:47 | |
| *** SergeyLukjanov has quit IRC | 09:50 | |
| *** SergeyLukjanov has joined #openstack-infra | 09:51 | |
| *** SergeyLukjanov has quit IRC | 09:51 | |
| *** DinaBelova has joined #openstack-infra | 09:54 | |
| *** matsuhas_ has quit IRC | 09:58 | |
| *** matsuhashi has joined #openstack-infra | 09:59 | |
| *** matsuhashi has quit IRC | 10:03 | |
| openstackgerrit | Peter Liljenberg proposed a change to openstack-infra/jenkins-job-builder: Added support for Jenkins plugin Blame upstream committers https://review.openstack.org/54085 | 10:06 |
| *** pcm_ has joined #openstack-infra | 10:10 | |
| *** davidhadas_ has quit IRC | 10:12 | |
| *** pcm_ has quit IRC | 10:12 | |
| *** pcm_ has joined #openstack-infra | 10:12 | |
| Steap | Hey, I'm not familiar with StackForge and have just read this comment: https://review.openstack.org/#/c/53306/6/modules/openstack_project/files/gerrit/acls/stackforge/sphinxcontrib-docbookrestapi.config . Grepping through the sources, I'm under the impression that not many projects use a "-ptl" group, am I right ? | 10:18 |
| Steap | Since I don't think my project will be extremely huge, is it fine to keep a "-core" group for the "pushTag" action, and move to "-ptl" if, for some reason, it becomes quite big ? | 10:19 |
| *** amotoki has quit IRC | 10:37 | |
| *** davidhadas has joined #openstack-infra | 10:48 | |
| *** shardy has joined #openstack-infra | 10:50 | |
| mattymo | ttx, ping | 10:52 |
| *** nosnos has quit IRC | 10:53 | |
| ttx | mattymo: pong | 10:53 |
| *** nosnos has joined #openstack-infra | 10:54 | |
| mattymo | ttx, I saw your reply about #openstack-fuel | 10:56 |
| mattymo | ttx, think it's worth waiting for others to voice their opinion on this or just agree that your view is what the majority will hold? | 10:56 |
| ttx | mattymo: probably worth waiting | 10:57 |
| ttx | mattymo: have no idea what the others think about that | 10:57 |
| *** ArxCruz has joined #openstack-infra | 10:58 | |
| *** nosnos has quit IRC | 10:58 | |
| mattymo | ttx, I was hoping the fact that Mirantis is a foundation member and that we're in stackforge that it would influence the group | 10:59 |
| mattymo | but of course a line must be drawn somewhere | 10:59 |
| *** guohliu has quit IRC | 11:10 | |
| ekarlso- | anyone here familiar with the config file generation stuff ? | 11:10 |
| ekarlso- | https://review.openstack.org/54101 < I do "./tools/config/generate_sample.sh -b ./ -p libra -o etc" and I get "RuntimeError: Unable to find group for option daemon, maybe it's defined twice in the same group?" | 11:15 |
| *** flaper87 is now known as flaper87|afk | 11:15 | |
| *** DinaBelova has quit IRC | 11:20 | |
| *** dizquierdo has joined #openstack-infra | 11:22 | |
| *** davidhadas has quit IRC | 11:30 | |
| *** davidhadas has joined #openstack-infra | 11:31 | |
| *** dims has joined #openstack-infra | 11:34 | |
| *** Bada has quit IRC | 11:40 | |
| *** Bada has joined #openstack-infra | 11:41 | |
| *** adalbas has joined #openstack-infra | 11:46 | |
| *** fallenpegasus has quit IRC | 11:49 | |
| jog0 | grenade is trying to upgrade from grizzly | 11:49 |
| *** davidhadas has quit IRC | 11:49 | |
| jog0 | sdague: ^ | 11:49 |
| *** davidhadas has joined #openstack-infra | 11:50 | |
| sdague | jog0: yes, it is | 11:50 |
| sdague | there is a whole bunch of stuff in process to try to fix that | 11:50 |
| jog0 | sdague: is that why grenade is failing ATM? | 11:50 |
| sdague | jog0: I don't know | 11:50 |
| jog0 | or at least making the issue worse | 11:50 |
| jog0 | sdague: http://logs.openstack.org/59/53859/5/check/check-grenade-devstack-vm/2bea2b1/ | 11:51 |
| sdague | that's a failure to validate old | 11:52 |
| sdague | it's not even tried to upgrade | 11:52 |
| sdague | someone broke agregates in grizzly? | 11:52 |
| jog0 | sdague: right, but its a grizzly failure | 11:52 |
| * jog0 pulls up griz git logs | 11:52 | |
| jog0 | this may explain how the issue got in. do we run grenade on stable branches? | 11:53 |
| sdague | no, again, spent many days working on this last week :) | 11:53 |
| sdague | to try to get us there, bug apparently someone managed to break it first | 11:53 |
| jog0 | sdague: btw I saw your email blogs this weekend, do you ever take time off? | 11:53 |
| *** luisg has quit IRC | 11:54 | |
| sdague | sometimes :) honestly my brain really wanted to work through some of these issues this weekend | 11:54 |
| *** luisg has joined #openstack-infra | 11:54 | |
| jog0 | we need more people like you | 11:54 |
| sdague | I find that I just let my interest take me where it wants, otherwise I just end up doing one thing but thinking about something else | 11:54 |
| *** fallenpegasus has joined #openstack-infra | 11:55 | |
| jog0 | https://review.openstack.org/#/q/status:merged+project:openstack/nova+branch:stable/grizzly,n,z | 11:55 |
| jog0 | nothing looks obvious in that list, or even timewise | 11:56 |
| *** simonmcc has quit IRC | 11:56 | |
| *** simonmcc has joined #openstack-infra | 11:56 | |
| jog0 | sdague: where is grenade configured to set old=grizzly | 11:56 |
| *** fallenpegasus has quit IRC | 11:56 | |
| *** dkranz has joined #openstack-infra | 11:57 | |
| sdague | devstack-gate | 11:58 |
| sdague | this is the change we need to get sorted - https://review.openstack.org/#/c/53940/ | 11:58 |
| sdague | but that's having issues at the moment, and not sure why | 11:58 |
| sdague | also there are related grenade changes, look for the changes by mauro | 11:59 |
| jog0 | sdague: so grenade isn't version agnostic? | 12:00 |
| sdague | no, it includes the specific fixups needed to get between two versions | 12:01 |
| * jog0 cries a little | 12:01 | |
| sdague | because, you can't upgrade openstack without that | 12:01 |
| jog0 | sigh | 12:01 |
| *** fallenpegasus has joined #openstack-infra | 12:01 | |
| sdague | hey, more hands making all services roll forward with no changes required to configs, appreciated :) | 12:02 |
| jog0 | sdague: hope to work on that for TripleO in the next few months | 12:02 |
| jog0 | but in the CD story | 12:02 |
| sdague | the point of grenade was at least we'd have them all documented | 12:02 |
| sdague | yeh, I still don't see how you get around some of it | 12:02 |
| jog0 | sdague: yeah I am not disagreeing just sad | 12:02 |
| sdague | like turning on v3 api | 12:02 |
| jog0 | sdague: anyway back to the critical issue unbarring gate | 12:03 |
| jog0 | unbreaking* | 12:03 |
| sdague | so the aggregates call fails | 12:03 |
| jog0 | sdague: but why | 12:03 |
| sdague | http://logs.openstack.org/59/53859/5/check/check-grenade-devstack-vm/2bea2b1/console.html#_2013-10-28_10_41_58_612 is the problem, I think | 12:04 |
| jog0 | we know this broke about 6 hours ago | 12:04 |
| jog0 | grenade uses trunk clients right? | 12:07 |
| sdague | the exercises, yeh | 12:08 |
| sdague | oh, look at that | 12:08 |
| sdague | commit a1391b2ead18c617bc5ce373a621f8a74c9cd952 | 12:08 |
| sdague | Merge: fbb01be 6374ee0 | 12:08 |
| sdague | Author: Jenkins <jenkins@review.openstack.org> | 12:08 |
| sdague | Date: Mon Oct 28 06:57:07 2013 +0000 | 12:08 |
| sdague | Merge "Nova aggregate-details should be more human friendly" | 12:08 |
| sdague | last commit in python-novaclient | 12:08 |
| jog0 | sdague: I tried taht | 12:08 |
| jog0 | I tried reverting | 12:09 |
| jog0 | didn't fix | 12:09 |
| sdague | and? | 12:09 |
| jog0 | https://review.openstack.org/#/c/54105/ | 12:09 |
| jog0 | I checked jenkins for the results so they didn't get pasted back | 12:09 |
| jog0 | that 409 is expected | 12:11 |
| jog0 | here is a passing grenade http://logs.openstack.org/35/51135/5/check/check-grenade-devstack-vm/fa4fcb1/console.html | 12:11 |
| jog0 | http://logs.openstack.org/35/51135/5/check/check-grenade-devstack-vm/fa4fcb1/console.html#_2013-10-28_03_31_36_670 | 12:12 |
| jog0 | this may be a dep issue | 12:12 |
| jog0 | sdague: pypi was updated around the time things started dying (6AM) http://pypi.openstack.org/openstack/?C=M;O=D | 12:13 |
| sdague | actually, I'm pretty sure it's that nova client change | 12:13 |
| sdague | I'm going to push the revert again | 12:13 |
| jog0 | sdague: can't hurt | 12:14 |
| sdague | and let everything run through to see | 12:14 |
| jog0 | sdague: hold on | 12:14 |
| jog0 | I'll unabandon and recheck | 12:14 |
| sdague | nova aggregate-details $AGGREGATE_ID | egrep "{u'availability_zone': u'$AGGREGATE_A_ZONE'}|{}" | 12:15 |
| sdague | jog0: I did another push already of the revert | 12:15 |
| jog0 | sdague: just noticed | 12:15 |
| sdague | but that's the line where we die, so the fact a display change was pushed would definitely trigger that | 12:16 |
| jog0 | https://jenkins02.openstack.org/job/check-grenade-devstack-vm/3467/consoleFull | 12:17 |
| jog0 | that was the output from my patch | 12:17 |
| *** dprince has joined #openstack-infra | 12:18 | |
| jog0 | plus that change should have been self-gating | 12:18 |
| jog0 | scratch tat | 12:18 |
| jog0 | no it should have: https://review.openstack.org/#/c/50438/ | 12:18 |
| sdague | so assuming this passes - https://review.openstack.org/#/c/54111/1 - I'll push that in | 12:22 |
| sdague | it will at least give us more info | 12:22 |
| *** weshay has joined #openstack-infra | 12:22 | |
| jog0 | sdague: cool | 12:23 |
| jog0 | sdague: yours failed too http://status.openstack.org/zuul/ | 12:26 |
| jog0 | any chance we are seeing an async issue? | 12:26 |
| jog0 | http://logs.openstack.org/59/53859/5/check/check-grenade-devstack-vm/2bea2b1/logs/pip-freeze.txt.gz | 12:27 |
| jog0 | and aren't actually using proposed change to novaclient? | 12:28 |
| sdague | yeh, that's what I'm thinking | 12:28 |
| jog0 | sdague: GAH | 12:28 |
| jog0 | http://logs.openstack.org/59/53859/5/check/check-grenade-devstack-vm/2bea2b1/console.html#_2013-10-28_10_39_07_145 | 12:28 |
| sdague | that's going to need more experienced folks than I | 12:28 |
| katyafervent | fungi, Thanks a lot!) | 12:30 |
| openstackgerrit | A change was merged to openstack-infra/config: Add murano-metadataclient to stackforge https://review.openstack.org/52092 | 12:30 |
| fungi | Steap: it's fine if your core group is just going to be a couple of people who will participate in tagging releases on it. and my other concern on that change (pypi registration) seems addressed now | 12:30 |
| sdague | jog0: yeh, so we knew that this assymetric with stable was going to eventually be a problem | 12:31 |
| sdague | so I guess we know that it is today | 12:31 |
| sdague | is this a jeblair vacation week? | 12:32 |
| jog0 | also its using github? | 12:32 |
| jog0 | sdague: yup | 12:32 |
| jog0 | in the Scottish Highlands | 12:32 |
| sdague | jog0: no, I don't think so | 12:32 |
| jog0 | http://logs.openstack.org/59/53859/5/check/check-grenade-devstack-vm/2bea2b1/console.html#_2013-10-28_10_41_55_285 | 12:32 |
| sdague | jog0: yeh, but the trees are preppred in the wrap script | 12:33 |
| jog0 | ? | 12:34 |
| Steap | fungi: yes :) | 12:36 |
| *** thomasem has joined #openstack-infra | 12:36 | |
| Steap | fungi: honestly, I'm currently the only member of the project; I'll probably be joined by some Ceiloemter devs, but I doubt we'll ever be "a lot of people" | 12:36 |
| sdague | jog0: we're never actually cloning in grenade | 12:37 |
| sdague | in the gate, all the trees are created for us | 12:37 |
| *** alaski has joined #openstack-infra | 12:37 | |
| fungi | katyafervent: no problem. i expect he didn't approve on +2 simply because he was on an airplane at the time | 12:38 |
| fungi | Steap: sounds good, and i see the pypi registration looks correct. thanks! | 12:38 |
| Steap | fungi: np :) | 12:39 |
| Steap | fungi: thx for the review | 12:39 |
| jog0 | sdague: but you clone the other repos still right? | 12:39 |
| sdague | not in the gate | 12:40 |
| sdague | the entire environment is set up in advanced by the devstack-gate wrapper scripts | 12:40 |
| jog0 | ohh the git_clone script doesn't clone git haha | 12:41 |
| sdague | yeh | 12:41 |
| sdague | it's a function for a reason, and it has a lot of ways to bail early | 12:41 |
| sdague | which we take advantage of | 12:41 |
| sdague | so yeh, it's the novaclient change | 12:41 |
| sdague | it changed formatting | 12:41 |
| sdague | fungi: can you ninja merge something for us? | 12:41 |
| fungi | sdague: i can | 12:42 |
| sdague | https://review.openstack.org/#/c/54108/ | 12:42 |
| jog0 | sdague: can we fix gate to pull that in instead? | 12:42 |
| fungi | the novaclient revert? | 12:42 |
| sdague | fungi: yeh | 12:42 |
| jog0 | to use novaclient proposed change in grenade? | 12:42 |
| sdague | jog0: it will probably take days to get that sorted correctly, especially with jeblair not around | 12:42 |
| jog0 | sdague: but don't we do that for other projects already | 12:43 |
| sdague | jog0: we've never sorted out the stable side of the grenade equation | 12:43 |
| jog0 | sdague: this isn't stable this is clients | 12:43 |
| sdague | this was actually the thing I was working on for two days last week :) | 12:43 |
| fungi | yeah, if the grenade job is currently using released novaclient when testing novaclient, i agree the gate breakage isn't going to be immediately fixable without an override. | 12:43 |
| sdague | jog0: except.... I don't think the old tree is actually getting latest changes | 12:44 |
| sdague | there are two issues of stable side, stable jobs, and the stable /...../old/... tree | 12:44 |
| sdague | honestly, this is matrix hell in some ways to get all the moving parts sorted :) | 12:45 |
| sdague | fungi: so lets see if the novaclient revert gets us sorted, then we can deal with the rest of it when we aren't on fire | 12:46 |
| fungi | it's merged now | 12:46 |
| jog0 | sdague: sounds good to me | 12:47 |
| jog0 | didn't realize how bad it was | 12:47 |
| sdague | fungi: cool, thanks | 12:47 |
| jog0 | sdague: want to send the ML response out saying life is good again | 12:47 |
| sdague | yeh, it's going to be a rough week with jeblair gone and mordred on a plan again to somewhere | 12:47 |
| *** flaper87|afk is now known as flaper87 | 12:47 | |
| sdague | jog0: only once we know that it is | 12:47 |
| sdague | https://review.openstack.org/#/c/54111/ - recheck, should be a good test | 12:47 |
| jog0 | sdague: heh I had no idea how delicate this stuff was | 12:48 |
| jog0 | btw why doesn't neutron have grenade? | 12:49 |
| jog0 | do we not have a neutron version? | 12:49 |
| sdague | because grenade only really checks 5 services - nova, cinder, swift, keystone, glance | 12:51 |
| sdague | and runs in nova-network mode | 12:51 |
| jog0 | more matrix hell :) | 12:52 |
| sdague | yeh, and honestly, until neutron can run under tenant isolation, parallel, and a full tempest run, thinking it might upgrade test well... is *optimistic* | 12:52 |
| jog0 | sdague: fair enough | 12:53 |
| *** yamahata has joined #openstack-infra | 12:55 | |
| *** anteaya has joined #openstack-infra | 12:55 | |
| sdague | and, basically there are very few people that help with grenade, so coverage is what it is | 12:55 |
| *** lnxnut has joined #openstack-infra | 12:57 | |
| jog0 | sdague: which 'project' ownes it? | 12:58 |
| jog0 | ci devstack qa? | 12:58 |
| sdague | qa | 12:58 |
| sdague | the reality is it's me and dtroyer, and whoever else I can con into helping :) | 12:59 |
| ArxCruz | clarkb: fungi can you guys help me with a zuul/jenkins issue ? | 12:59 |
| sdague | that set of ci stats for openstack really made the rounds yesterday | 13:00 |
| jog0 | sdague: I may have conned myself into helping -- I want to get the RPC / rolling upgrade test up | 13:00 |
| jog0 | but using grenade will be a bit of a burdon at least at first | 13:00 |
| sdague | honestly, with the complexity of getting the gate setup correctly for 2 trees, I think the only sane path is to work it into grenade scope | 13:01 |
| *** zul has quit IRC | 13:02 | |
| sdague | I really think any other approach just ends up building a grenade like fork of a thing, which is then 2 upgrade test frameworks with not enough reviewers to handle the crazy matrix | 13:02 |
| sdague | i.e. this is a harder problem than you realize, to make it gate friendly | 13:03 |
| *** ericw has joined #openstack-infra | 13:03 | |
| *** lnxnut has quit IRC | 13:04 | |
| *** dripton has joined #openstack-infra | 13:04 | |
| *** julim has joined #openstack-infra | 13:04 | |
| jog0 | sdague: its not an upgrad test at first | 13:04 |
| jog0 | its a RPC compat test | 13:04 |
| sdague | jog0: I get that | 13:04 |
| *** lnxnut has joined #openstack-infra | 13:04 | |
| sdague | but it is a test that needs 2 different versions of openstack checked out | 13:05 |
| jog0 | sdague: but yeah I tend to agree, in the long term for sure | 13:05 |
| jog0 | in the short term not sure | 13:05 |
| sdague | as you can see... getting that right in the gate is hard | 13:05 |
| *** zul has joined #openstack-infra | 13:05 | |
| sdague | starting over from scratch on that is just silly | 13:05 |
| *** lnxnut has quit IRC | 13:05 | |
| *** lnxnut has joined #openstack-infra | 13:06 | |
| sdague | because you'll just end up having learn, the hard way, all the things you need to do to make that work right in the gate | 13:06 |
| sdague | and build another crazy set of ifdefs in devstack-gate | 13:06 |
| *** ruhe has joined #openstack-infra | 13:06 | |
| *** fallenpegasus has quit IRC | 13:06 | |
| jog0 | sdague: I did POC that ran after devstack finished | 13:07 |
| jog0 | not during ... eww | 13:07 |
| jog0 | anyway we don't have to sort this out now | 13:08 |
| jog0 | won't happen before summit anyway | 13:08 |
| *** lnxnut has quit IRC | 13:10 | |
| *** lnxnut has joined #openstack-infra | 13:10 | |
| fungi | ArxCruz: what's the issue? | 13:11 |
| *** fallenpegasus has joined #openstack-infra | 13:11 | |
| *** lnxnut has quit IRC | 13:15 | |
| anteaya | sdague: that blog post was great | 13:15 |
| anteaya | for those who haven't seen it yet: http://dague.net/2013/10/27/openstack-ci-by-the-numbers/ | 13:16 |
| *** annegentle has joined #openstack-infra | 13:18 | |
| *** b3nt_pin has joined #openstack-infra | 13:20 | |
| *** b3nt_pin is now known as beagles | 13:20 | |
| *** guohliu has joined #openstack-infra | 13:20 | |
| sdague | jog0: the thing is, if your setup ever relies on "then we git clone" ... -2, because zuul has to do all the cloning in the gate | 13:21 |
| sdague | or git anything, honestly | 13:21 |
| jog0 | sdague: :( | 13:22 |
| fungi | sdague: well, you can clone and then fetch the zuul ref passed in as an envvar | 13:22 |
| sdague | zuul has to set up all the refs | 13:22 |
| *** mriedem has joined #openstack-infra | 13:22 | |
| jog0 | the basic worflow I tested was: given a devstack | 13:22 |
| sdague | fungi: yeh, but the reality is zuul needs to do all the tree setup | 13:22 |
| jog0 | checkout grizzly, kick nova-compute, run tempest | 13:23 |
| *** SergeyLukjanov has joined #openstack-infra | 13:23 | |
| sdague | jog0: so how do you know to checkout grizzly? | 13:23 |
| jog0 | sdague: on my own env 'git checkout ...' | 13:23 |
| sdague | and what if it's actually supposed to be a gerrit tree | 13:23 |
| sdague | because it's a grizzly change | 13:24 |
| jog0 | sdague: yeah ... I agree | 13:24 |
| sdague | and what if there is code in the nova tree which hasn't dynamically loaded yet | 13:24 |
| jog0 | anyway lets punt this to a HK chat | 13:24 |
| sdague | sure :) | 13:24 |
| sdague | but I'm a big -2 on another framework here :) | 13:24 |
| sdague | and you are going to need a lot of evidence that grenade *can't* be made to cover this issue to get me to change my mind | 13:25 |
| fungi | thinking through the latest excitement and how it specifically could have been prevented, i think once there are proper backward-compat jobs running for clients (so that the novaclient change in question would have failed to merge because it broke stable/grizzly, right?) this specific bit of the pain could be avoided? | 13:25 |
| sdague | so, I actually know why the current situation failed | 13:25 |
| sdague | because I was in this code friday, I just didn't understand this until then | 13:25 |
| sdague | but we currently *never* set zuul refs on old | 13:25 |
| * fungi goes to find and review the client backward-compat testing change again | 13:26 | |
| *** jgriffith has quit IRC | 13:26 | |
| sdague | which is basically the issue | 13:26 |
| sdague | and I think it might be subtle about how we do that | 13:26 |
| sdague | as the gate has never been set up to do tat | 13:26 |
| *** jgriffith has joined #openstack-infra | 13:27 | |
| fungi | sdague: what if there was already a separate job which ensured that what was set up for old was functional? (e.g. backward-compat devstack tempest-full) | 13:27 |
| sdague | so old actually was functional | 13:27 |
| jog0 | mordred: is working on that AFAIK | 13:28 |
| *** SergeyLukjanov has quit IRC | 13:28 | |
| sdague | fungi: maybe that's a beer conversation? | 13:28 |
| sdague | so here's the thing | 13:28 |
| *** SergeyLukjanov has joined #openstack-infra | 13:28 | |
| sdague | even in the best of cases, we are probably only testing 20% of surface | 13:29 |
| fungi | oh, was it only grenade tripping over the change in novaclient behavior and devstack-tempest on the starting stable branch would have been fine? if so, yeah that's a different kettle of fish | 13:29 |
| fungi | jog0: yes, he is, i was merely recalling that i need to review it again | 13:29 |
| zul | no new pbr yet? | 13:29 |
| jog0 | greande is working again https://review.openstack.org/#/c/54092/ | 13:29 |
| fungi | zul: i thought he tagged it yesterday | 13:30 |
| sdague | so I remain skeptical that we could keep the kind of throughput that we have today if we pulled the base case validation further away from the tests | 13:30 |
| zul | fungi: still not pypi | 13:30 |
| sdague | because it's not like we have 100% behavior definition | 13:30 |
| jog0 | sdague: want to respond to the ML thread or should I | 13:30 |
| fungi | zul: i see it as 0.5.22 (uploaded today, utc)... https://pypi.python.org/pypi/pbr | 13:30 |
| sdague | jog0: I need to head to the office, so go for it (back on in 30) | 13:30 |
| jog0 | kk | 13:30 |
| zul | fungi: oh duh i totally need more caffine | 13:31 |
| fungi | zul: me too, me too | 13:31 |
| *** rpodolyaka has left #openstack-infra | 13:40 | |
| fungi | bodepd_: on https://review.openstack.org/51079 you're importing a git repository which has what look like a bunch of topic branches in it. those will be imported automatically with it unless you a) delete them there first or b) point your review.projects.yaml.erb entry at a fork with only the branch(es) you want imported | 13:42 |
| *** jaypipes has joined #openstack-infra | 13:45 | |
| *** SergeyLukjanov has quit IRC | 13:47 | |
| fungi | bodepd_: i'll follow up in the review since you're probably not in the office yet | 13:49 |
| mattymo | fungi, fyi bodepd_ was online until 3am your time. he might not show up for a few hrs | 13:51 |
| fungi | mattymo: yeah, that was only midnight his time | 13:52 |
| fungi | i had merely forgotten for a moment which tz he's in | 13:52 |
| mattymo | it happens | 13:52 |
| mattymo | fungi, regarding my channel request, it looks like I have resistance on both sides | 13:52 |
| mattymo | #fuel was registered by someone 7 years ago so I get to wait for freenode staff to review and convert it | 13:53 |
| fungi | mattymo: i feel bad--i'm so far behind on ml threads i haven't actually read it yet. taking a review break to look at it now | 13:53 |
| *** SergeyLukjanov has joined #openstack-infra | 13:54 | |
| *** nati_ueno has joined #openstack-infra | 13:54 | |
| fungi | off-topic rant... shell server vm on a host which is rebuilding its software raid-1 after a failure. i/o starvation city: population me | 13:56 |
| mattymo | as we say in Russia, it's normal | 13:56 |
| *** DinaBelova has joined #openstack-infra | 13:57 | |
| zul | fungi: heh | 13:59 |
| *** beekneemech is now known as bnemec | 14:00 | |
| fungi | zul: i would call the sysadmin and yell at him, but yelling at myself isn't going to solve anything i'm afraid | 14:00 |
| rakhmerov | :) | 14:00 |
| *** nati_ueno has quit IRC | 14:00 | |
| *** nati_ueno has joined #openstack-infra | 14:01 | |
| zul | fungi: you could look at yourself in the mirror and then start to yell | 14:01 |
| *** boris-42 has quit IRC | 14:02 | |
| *** b3nt_pin has joined #openstack-infra | 14:02 | |
| sdague | fungi: you got some time today to help on diving through the d-g scripts to see if we can come up with a "once and for all" approach that would have preventing the novaclient break? | 14:02 |
| *** dcramer_ has joined #openstack-infra | 14:03 | |
| rakhmerov | fungi: thanks for reviewing Mistral stuff at https://review.openstack.org/#/c/51205. I have a question on that. You left some comments some time ago and after that it's been there for a while and nobody else reviewed it. Do you think it makes sense to wait a little longer or it's better to fix those non-critical (if I understood your overall feedback) asap? Just not sure what to do next :( | 14:03 |
| *** yamahata has quit IRC | 14:05 | |
| *** nati_ueno has quit IRC | 14:05 | |
| *** oubiwann has joined #openstack-infra | 14:06 | |
| fungi | sdague: if i understand the problem you're trying to solve, i don't think it can be done with one job | 14:06 |
| *** yamahata has joined #openstack-infra | 14:06 | |
| fungi | at least not trivially | 14:06 |
| *** fallenpegasus has quit IRC | 14:07 | |
| fungi | consider, for a client/library change you want grenade to test two things... that the change isn't breaking the base scenario, and that the change works when applied to the upgraded version | 14:07 |
| fungi | well, actually i guess with libraries, you're not actually testing an upgrade of the library right? | 14:08 |
| sdague | correct | 14:08 |
| sdague | what I'm trying to do is prevent this wedge | 14:08 |
| fungi | only upgrades of servers | 14:08 |
| sdague | and what I'm trying to figure out is if setting 0 => 1 there is going to ruin the world - https://github.com/openstack-infra/devstack-gate/blob/master/devstack-vm-gate-wrap.sh#L490 | 14:09 |
| *** alcabrera has joined #openstack-infra | 14:09 | |
| fungi | so we need to use the proposed change, if it's a library, in both the base scenario and the upgraded one. but if it's a server we only want the change in one or the other (in the base scenario for a stable branch test and in the upgraded scenario for a master branch change) | 14:09 |
| sdague | right, well the issue right now is on "old" we never update with zuul refs | 14:10 |
| sdague | ever | 14:10 |
| sdague | which is how we got the wedge | 14:10 |
| sdague | as novaclient was able to run through old verification with an old version, and new verification is different tests | 14:10 |
| fungi | okay, and doing so is going to require that the base scenario setup considers the zuul branch (to decide whether this is a test on stable or master so we know which direction we're testing in) and the zuul project (so we can decide whether it's one which gets the zuul ref in only one scenario or both) | 14:12 |
| *** fallenpegasus has joined #openstack-infra | 14:12 | |
| *** ruhe has quit IRC | 14:14 | |
| *** oubiwann_ has joined #openstack-infra | 14:15 | |
| fungi | note i'm mainly thinking in terms of "upgrade from change on stable" just to make sure we don't paint ourselves into a hole which makes that harder to implement later. i know it's not actually being done yet | 14:15 |
| sdague | yeh, though I actually think it's all related | 14:16 |
| fungi | dig ourselves into a corner, whatever ;) | 14:16 |
| sdague | the point being, we have 2 trees | 14:16 |
| sdague | and we get a zuul change in | 14:16 |
| sdague | actually, let's etherpad this | 14:16 |
| fungi | definitely | 14:17 |
| sdague | https://etherpad.openstack.org/p/dg-for-grenade | 14:17 |
| mordred | morning all | 14:17 |
| anteaya | morning | 14:17 |
| *** oubiwann has quit IRC | 14:17 | |
| *** dkliban has quit IRC | 14:19 | |
| sdague | mordred: so, it finally hit us, the stable upgrade grenade wedge | 14:20 |
| *** b3nt_pin has quit IRC | 14:20 | |
| mordred | sdague, fungi: I am not fully awake enough to be super helpful yet | 14:20 |
| sdague | fungi: honestly that third thing is the thing I'm scared about understanding | 14:20 |
| mordred | but, I think we have to consider zuul refs for both old and new | 14:20 |
| sdague | when we have a deep zuul queue | 14:20 |
| fungi | mordred: sdague yep | 14:20 |
| *** beagles has quit IRC | 14:21 | |
| mordred | I don't think the 3rd thing is too terrible | 14:21 |
| sdague | fungi: so if we blindly just say "apply to old" are we smart enough to only do that with the right branches? | 14:21 |
| fungi | sdague: there is logic in devstack which does that | 14:21 |
| *** b3nt_pin has joined #openstack-infra | 14:21 | |
| fungi | or rather in devstack-gate i think | 14:21 |
| *** b3nt_pin is now known as beagles | 14:21 | |
| fungi | first tries to find the zuul ref on the target branch then falls back | 14:21 |
| mordred | oh - you know what - jeblair and I already kinda stepped through the logic on this a little bit for the stable-client code | 14:21 |
| *** oubiwann_ is now known as oubiwann | 14:22 | |
| sdague | mordred: ok, cool | 14:22 |
| mordred | I think. I'm still caffinated | 14:22 |
| mordred | uncaffinated | 14:22 |
| sdague | fungi: yeh I guess it's unclear to me if the fallback branch is doing the right things, or only covers the client case | 14:23 |
| fungi | sdague: well, i think the fallback branch selection gets more complex in grenade | 14:23 |
| *** ruhe has joined #openstack-infra | 14:23 | |
| *** markmcclain has joined #openstack-infra | 14:24 | |
| fungi | but point was we can use similar logic around the zuul ref existing, we just need to have a bit more work surrounding fallback branch selection | 14:24 |
| *** SergeyLukjanov is now known as _SergeyLukjanov | 14:24 | |
| *** _SergeyLukjanov has quit IRC | 14:24 | |
| *** fifieldt has quit IRC | 14:24 | |
| *** senk has joined #openstack-infra | 14:25 | |
| *** johnthetubaguy1 has joined #openstack-infra | 14:25 | |
| *** ryanpetrello has joined #openstack-infra | 14:26 | |
| *** rcleere has joined #openstack-infra | 14:26 | |
| *** dkliban has joined #openstack-infra | 14:27 | |
| *** johnthetubaguy1 is now known as johnthetubaguy | 14:27 | |
| sdague | yeh, man, I want to really refactor the wrap script to make it make more sense to me. I'm so used to our heavy function style in devstack | 14:27 |
| *** datsun180b has joined #openstack-infra | 14:27 | |
| fungi | well, it *is* self-gating, now backward on on stable release branches of tempest jobs too | 14:28 |
| *** che-arne has quit IRC | 14:28 | |
| fungi | so if we want to refactor it some, we can have a fair amount of confidence we're not breaking the gate (though less confidence that it's still testing the things we think it's testing) | 14:29 |
| *** senk has quit IRC | 14:30 | |
| anteaya | rakhmerov: I would suggest that you address fungi's comments with a new patchset | 14:30 |
| anteaya | basically his comments amount to, "This patch won't achieve your goal the way it is written." | 14:31 |
| rakhmerov | anteaya: ok, will do | 14:33 |
| rakhmerov | thanks | 14:33 |
| anteaya | np | 14:33 |
| anteaya | thought I would jump in since he is in the middle of another conversation | 14:33 |
| fungi | yeah, i'm reading back over what i wrote now. it may work as written, but i don't know | 14:33 |
| *** senk has joined #openstack-infra | 14:36 | |
| *** SergeyLukjanov has joined #openstack-infra | 14:36 | |
| *** prad_ has joined #openstack-infra | 14:36 | |
| *** davidhadas_ has joined #openstack-infra | 14:37 | |
| *** davidhadas has quit IRC | 14:39 | |
| openstackgerrit | Sean Dague proposed a change to openstack-infra/devstack-gate: extract git_checkout into common function https://review.openstack.org/54132 | 14:40 |
| *** che-arne has joined #openstack-infra | 14:40 | |
| *** davidhadas_ has quit IRC | 14:41 | |
| *** wenlock has joined #openstack-infra | 14:43 | |
| *** beagles has quit IRC | 14:44 | |
| openstackgerrit | Sean Dague proposed a change to openstack-infra/devstack-gate: move single host setup activities to setup_host https://review.openstack.org/54133 | 14:45 |
| ArxCruz | ALL: I'm listening review.openstack.org in my intern infrastructure, and my zuul is returning ZUUL_REF with a Z in front which make my jobs fail since it can't find the ref. I'm using layout.yaml from dev just to listen instead of send reviews back to review.openstack.org. Does anyone know what's could be wrong? | 14:46 |
| *** yamahata has quit IRC | 14:47 | |
| mordred | ArxCruz: zuul keeps its own copies of the repos where it constructs the refs to be changed | 14:48 |
| *** fallenpegasus has quit IRC | 14:48 | |
| fungi | ArxCruz: zuul refs have a Z at the start of their name so we can tell them apart from other refs (branches, tags, shas...) | 14:48 |
| mordred | ArxCruz: you need to make sure that you clone/fetch from your zuul | 14:48 |
| *** vipul-away is now known as vipul | 14:48 | |
| ArxCruz | mordred: and how I do that? where do i configure it ? | 14:49 |
| mordred | ArxCruz: in whatever job zuul is triggering, you need to fetch the ZUUL_REF right? | 14:49 |
| mordred | (you are already trying to do that, because you said it's failing) | 14:49 |
| fungi | ArxCruz: if you follow the puppetry for our zuul.openstack.org configuration, you'll see we have apache serving git repositories for it | 14:49 |
| mordred | when you do that, point it at your zuul | 14:49 |
| ArxCruz | mordred: fungi i'm cloning to my zuul, when zuul start he's cloning to /var/lib/zuul/git in my server | 14:50 |
| mordred | ArxCruz: yes. that's right | 14:50 |
| openstackgerrit | Sean Dague proposed a change to openstack-infra/devstack-gate: Let us gate grenade forward https://review.openstack.org/53940 | 14:50 |
| mordred | ArxCruz: your jobs need to fetch the Z refs from the repos in /var/lib/zuul/git | 14:51 |
| ArxCruz | mordred: okay, i will check here | 14:51 |
| ArxCruz | by the way, there's a z repository here that's failing, is that okay ? | 14:51 |
| mordred | I don't know what you mean be failing? | 14:51 |
| fungi | i don't know what you mean by a z repository | 14:52 |
| mordred | sdague: Co-Authored-With: ... our AUTHORS script checks for Co-Authored-By: ... should it check for both? | 14:52 |
| sdague | mordred: doh | 14:52 |
| sdague | let me fix | 14:52 |
| ArxCruz | ERROR zuul.Repo: Unable to initialize repo for ssh://arxcruz@review.openstack.org:29418/z/tempest | 14:52 |
| openstackgerrit | Sean Dague proposed a change to openstack-infra/devstack-gate: Let us gate grenade forward https://review.openstack.org/53940 | 14:52 |
| ArxCruz | fungi: mordred ^ | 14:52 |
| *** mrodden has joined #openstack-infra | 14:53 | |
| *** b3nt_pin has joined #openstack-infra | 14:53 | |
| *** fallenpegasus has joined #openstack-infra | 14:53 | |
| mordred | ArxCruz: what's z/tempest ? | 14:56 |
| ArxCruz | mordred: nevermind, it's just a project in layout.yaml in zuul-dev | 14:56 |
| ArxCruz | :) | 14:56 |
| mordred | :) | 14:56 |
| fungi | ArxCruz: i'd need to read back through the zuul source, but first guess is that there's a config option to specify the path to where you want zuul to build its working repositories and you've specified ssh://arxcruz@review.openstack.org:29418/z as the path? | 14:56 |
| ArxCruz | fungi: i didn't specify nothing :) just puppet it | 14:57 |
| fungi | ph, that | 14:57 |
| *** b3nt_pin is now known as beagles | 14:57 | |
| fungi | oh, that | 14:57 |
| ArxCruz | in layout.yaml have a project z/tempest | 14:57 |
| fungi | z/tempest is a fake project to force shared queues for the jobs in it | 14:57 |
| *** senk has quit IRC | 14:58 | |
| fungi | i bet we have the same error in our zuul logs at startup | 14:58 |
| *** senk has joined #openstack-infra | 14:58 | |
| *** senk has quit IRC | 14:58 | |
| *** senk has joined #openstack-infra | 14:59 | |
| *** dcramer_ has quit IRC | 15:00 | |
| openstackgerrit | Matthew Mosesohn proposed a change to openstack-infra/config: add fuel-dev channel to gerritbot https://review.openstack.org/53656 | 15:00 |
| *** Ryan_Lane has quit IRC | 15:00 | |
| *** xeyed4good has joined #openstack-infra | 15:01 | |
| *** xeyed4good has left #openstack-infra | 15:01 | |
| ArxCruz | fungi: mordred it seems the problem is in devstack-vm-gate-wrap.sh | 15:02 |
| ArxCruz | export ZUUL_URL=${ZUUL_URL:-http://zuul.openstack.org/p} | 15:02 |
| mordred | yes. you'll want to set ZUUL_URL to point to your zuul instead of our zuul | 15:02 |
| ArxCruz | mordred: okay, the problem is that my vm is being created on demand through nodepool | 15:03 |
| ArxCruz | there's no way to set that automatically | 15:03 |
| ArxCruz | correct ? | 15:03 |
| mordred | sure there is | 15:03 |
| *** nati_ueno has joined #openstack-infra | 15:03 | |
| mordred | what's calling devstack-vm-gate-wrap.sh ? | 15:03 |
| ArxCruz | the jenkins job | 15:03 |
| mordred | yup. so set it in the jenkins job | 15:03 |
| ArxCruz | mordred: aha! but, if I do that, next time puppet service begin, it will fix that ;D | 15:04 |
| ArxCruz | got it ? | 15:04 |
| mordred | ArxCruz: set it in the jenkins-job-builder description of the job | 15:04 |
| mordred | so that puppet sets it | 15:04 |
| ArxCruz | mordred: i didn't get it :/ | 15:05 |
| mordred | ArxCruz: you have a set of yaml files in your local env that contain the descriptions of the jenkins jobs, right? | 15:05 |
| mordred | ArxCruz: such as one called devstack-gate.yaml ? | 15:05 |
| ArxCruz | mordred: yes | 15:06 |
| mordred | ArxCruz: so edit them, and before it calls devstack-vm-gate-wrap.sh, do "export ZUUL_URL=http://my.local.zuul.server/p" | 15:06 |
| *** changbl has quit IRC | 15:06 | |
| mordred | ArxCruz: then puppet will see the change, apply it to your jenkins, and when jenkins tries to run the script, it will set the ZUUL_URL for you | 15:07 |
| ArxCruz | mordred: I got it, thanks | 15:07 |
| mordred | fungi: ^^ we may want to think about a way to parameterize that interaction slightly better | 15:07 |
| *** dcramer_ has joined #openstack-infra | 15:07 | |
| ArxCruz | mordred: yes! need it :) | 15:07 |
| mordred | fungi: in terms of people like ArxCruz setting up local testing | 15:07 |
| mordred | fungi: especially for the ones who do the thing we tell them they shoudl do - which is just use a copy of our stuff :) | 15:08 |
| *** UtahDave has joined #openstack-infra | 15:09 | |
| ArxCruz | mordred: and it's pretty easy to set | 15:09 |
| fungi | mordred: it is absolutely the case that we have lots of silent defaults to our own settings in various places, which makes them less discoverable since we don't need to set them in our configs | 15:09 |
| mordred | yup | 15:09 |
| * mordred bets wenlock has found many of those :) | 15:09 | |
| ArxCruz | I can change the yaml to export ZUUL_URL=<? some_config > | 15:09 |
| ArxCruz | if there's no some_config | 15:10 |
| mordred | wenlock: speaking of - when you get to zuul, you may hit this ^^ | 15:10 |
| *** reed has joined #openstack-infra | 15:10 | |
| ArxCruz | the export ZUUL_URL=${ZUUL_URL:-http://zuul.openstack.org/p} line in devstack-vm-gate-wrap.sh will change it to the default one | 15:10 |
| mattymo | fungi, I already managed to get #fuel registered. It seemed much harder to just go around you guys than to go through | 15:10 |
| mordred | ArxCruz: you could - except I don't think we're doing any of our jenkins_job_builder things as templates, only as raw files | 15:10 |
| mattymo | err much easier just to go around* | 15:10 |
| wenlock | reading back... | 15:10 |
| wenlock | indeed we have hit this! | 15:10 |
| *** SergeyLukjanov has quit IRC | 15:10 | |
| wenlock | we were going to ask for help this week on that | 15:11 |
| ArxCruz | i found this issue with zuul before, and what I did was a workaround let me show you | 15:11 |
| mordred | heh. great timing | 15:11 |
| openstackgerrit | sebastian marcet proposed a change to openstack-infra/config: Review and create Puppet configuration for dev server blueprint sso-openid-provider https://review.openstack.org/53644 | 15:11 |
| ArxCruz | mordred: i did it: http://paste.openstack.org/show/50058/ | 15:12 |
| fungi | mattymo: yeah, it's become a bit of a pain point now that we have 80 stackforge projects and counting | 15:12 |
| mordred | mattymo: there are a few things we haven't designed for with stackforge popularity - irc channels are clearly one of them :) | 15:12 |
| ArxCruz | so, I ran a sed after jenkins job is installed switching the zuul.openstack.org to my own | 15:12 |
| mordred | ArxCruz: hah. nice hack | 15:12 |
| krtaylor | ArxCruz, maybe a blueprint for this work, we can come up with a plan to parameterize as we get all this running in our environment | 15:13 |
| krtaylor | mordred, fungi ^^^ | 15:14 |
| ArxCruz | krtaylor: to parameterize we need to move the yaml files from files to templates in puppet | 15:14 |
| fungi | krtaylor: in many cases (such as this one) it is parameterized already, but not necessarily discoverable without reading through the source | 15:15 |
| mordred | fungi: it's not though | 15:15 |
| mordred | fungi: because chaning the param would involve forking the puppet, which would make it exceedingly hard to track upstream changes | 15:15 |
| fungi | it's at least parameterized in the devstack-gate scripts in this particular case, and defaulting to our preferred config | 15:15 |
| fungi | but yes, not parameterized in jjb configuration nor in puppetry | 15:16 |
| krtaylor | ArxCruz and I were brainstorming how we could do this, maybe with heira | 15:16 |
| mordred | the question is - how can one put all localized parameters into site.pp - which has NO expecetations of reusability | 15:16 |
| mordred | I don't think you need to go heira | 15:16 |
| mordred | if you can just get it to be something that you put in manifests/site.pp and passed in to the modules, it should be ok | 15:17 |
| ArxCruz | mordred: like that sed ? I think that's a very ugly workaround :) | 15:18 |
| mordred | ArxCruz: it is an ugly workaround - I think we can figure out something better | 15:19 |
| *** mgagne has joined #openstack-infra | 15:21 | |
| ArxCruz | mordred: it seems this problem is only on devstack-gate, would be possible, move devstack-gate.yaml to templates, adding a new property on puppet like zuul_server_url, with the default to zuul.openstack.org and put this as an parameter in devstack-gate.yaml, then change in devstack-vm-gate-wrap.sh check if the parameter is set or not and change if necessary ? | 15:23 |
| *** markmcclain has quit IRC | 15:23 | |
| *** ryanpetrello has quit IRC | 15:23 | |
| *** changbl has joined #openstack-infra | 15:24 | |
| krtaylor | Just curious, how many 3rd party CI test environments exist? It would be useful to know so we could discuss the snags and how they were solved | 15:27 |
| fungi | krtaylor: https://review.openstack.org/#/admin/groups/91,members | 15:27 |
| krtaylor | maybe that is more of an email thread, or summit bof | 15:27 |
| mordred | krtaylor: summit bof! | 15:27 |
| mordred | krtaylor: I think figuring out the best way to discuss issues with third-party copies of infra without killing the infra team is a great idea | 15:28 |
| *** sarob has quit IRC | 15:28 | |
| krtaylor | prob too late for next week :) | 15:28 |
| mordred | ArxCruz: I think that wold work - but I tihnk it might make the repo weird. I think that's close though | 15:28 |
| *** sarob has joined #openstack-infra | 15:28 | |
| krtaylor | mordred, agreed, we can help each other, bring forward plan to generalize things | 15:29 |
| fungi | i agree that there's a great long-tail benefit to be realized from helping other groups get copies of our different tools/systems up and running (since that means more people who can understand and improve on it eventually), but in the short term it's definitely very time-consuming to help those people get started | 15:29 |
| ArxCruz | i can't see this group 91,members :( | 15:29 |
| krtaylor | nor I | 15:30 |
| fungi | oh, right, it's not visible. fixing | 15:30 |
| ArxCruz | mordred: or we can move everything to templates already, even if it's note necessary for all the yaml's file | 15:30 |
| fungi | try again. there's really no reason we shouldn't make the list of bot accounts in our gerrit public | 15:31 |
| ArxCruz | will became more easy if in the future need to change something | 15:31 |
| *** SergeyLukjanov has joined #openstack-infra | 15:31 | |
| mordred | fungi, clarkb: ^^ thoughts on moving jjb things from files to templates? | 15:31 |
| *** davidhadas has joined #openstack-infra | 15:31 | |
| mordred | does directory inclusion work on templates? or do we have to list each one individually? | 15:32 |
| fungi | mordred: i don't think it does. at least that was the reason i assumed we were using flat directories for templates in our modules | 15:32 |
| *** yamahata has joined #openstack-infra | 15:33 | |
| *** sarob has quit IRC | 15:33 | |
| fungi | perhaps if jjb could read from multiple directories, we's be able to put the templates in one and manage the other with just a file tree | 15:33 |
| fungi | er, we'd | 15:33 |
| fungi | though it may be that you can put a template in a puppet-managed directory and the resulting file counts as being a puppet-managed file for that directory | 15:34 |
| fungi | if the docs say that works, i'm not looking in the right place | 15:35 |
| fungi | on a separate note, fyi github's api endpoint is all fsck'd up again. manage-projects isn't running successfully today because of random github api errors | 15:36 |
| *** yassine has quit IRC | 15:36 | |
| ArxCruz | fungi: that would be the same as move only the devstack-gate.yaml to a template huh? | 15:36 |
| fungi | ArxCruz: right | 15:36 |
| fungi | ArxCruz: because i think it would still be nice for people to be able to add jjc files in the config repo without having to make accompanying edits to a manifest for each | 15:37 |
| fungi | er, jjb | 15:37 |
| ArxCruz | fungi: Is it possible use a directory as a template ? | 15:38 |
| fungi | ArxCruz: not sure. i've not tried that before, but there is documentation for puppet | 15:39 |
| openstackgerrit | A change was merged to openstack-infra/elastic-recheck: refactor templates into query_builder https://review.openstack.org/52818 | 15:40 |
| *** ruhe has quit IRC | 15:40 | |
| fungi | again, i don't see anything which says that's a feature, but i could be looking in the wrong part of the puppet docs | 15:41 |
| *** chuck_ has joined #openstack-infra | 15:41 | |
| ArxCruz | fungi: http://projects.puppetlabs.com/issues/6649 | 15:41 |
| ArxCruz | it seems there's a module to do that | 15:41 |
| ArxCruz | but it's not a feature in puppet | 15:41 |
| *** miqui has joined #openstack-infra | 15:41 | |
| fungi | right, wontfix'd in puppet core anyway | 15:43 |
| ArxCruz | fungi: I believe it would be possible just move the devstack-vm.yaml to template | 15:43 |
| ArxCruz | creating a new puppet config only for jjb who needs parameters | 15:44 |
| ArxCruz | and another to read the jenkins_jobs_builder/config | 15:44 |
| ArxCruz | so, if you just want to add a new file without any changes, put in /config | 15:44 |
| ArxCruz | if you know it will need paramerter to be configured, change the puppet script | 15:45 |
| ArxCruz | it will split the jjb but it's the best solution I can see so far :/ | 15:45 |
| fungi | right, just hoping puppet management of that directory won't step on puppet managing a specific file within the directory. would need testing | 15:45 |
| *** krotscheck has joined #openstack-infra | 15:45 | |
| ArxCruz | fungi: say no more, I will test in my infra | 15:46 |
| ArxCruz | :) | 15:46 |
| ArxCruz | if works, I will submit a patch :) | 15:46 |
| krtaylor | ArxCruz, +1, consider it added to our weekly spring backlog :) | 15:46 |
| fungi | it's far easier to discuss these things when there's concrete code behind them. thanks ArxCruz! | 15:46 |
| ArxCruz | fungi: :) | 15:46 |
| *** krotscheck has quit IRC | 15:47 | |
| ArxCruz | krtaylor: working on that right now, this will improve our work in like 90% for this week :) | 15:47 |
| *** sdake has joined #openstack-infra | 15:50 | |
| *** fallenpegasus has quit IRC | 15:51 | |
| *** hogepodge has joined #openstack-infra | 15:51 | |
| *** yassine has joined #openstack-infra | 15:52 | |
| *** ruhe has joined #openstack-infra | 15:53 | |
| *** boris-42 has joined #openstack-infra | 15:56 | |
| *** fallenpegasus has joined #openstack-infra | 15:56 | |
| *** senk has quit IRC | 15:57 | |
| *** ruhe has quit IRC | 16:01 | |
| *** mrodden1 has joined #openstack-infra | 16:06 | |
| *** ^d has joined #openstack-infra | 16:06 | |
| *** ^d has quit IRC | 16:07 | |
| *** ^d has joined #openstack-infra | 16:07 | |
| *** mrodden has quit IRC | 16:07 | |
| *** senk has joined #openstack-infra | 16:08 | |
| *** gyee has joined #openstack-infra | 16:09 | |
| anteaya | so reading up on how a puppet module is supposed to be created, should this be in modules/seafile/files rather than in modules/seafile/templates: https://review.openstack.org/#/c/54030/12/modules/seafile/templates/ccnet/seafile.ini | 16:10 |
| *** senk1 has joined #openstack-infra | 16:10 | |
| anteaya | since it isn't a template | 16:10 |
| mattymo | anteaya, are you staff? | 16:10 |
| anteaya | ha ha ha | 16:10 |
| *** marun has quit IRC | 16:10 | |
| anteaya | I can answer questions | 16:10 |
| mattymo | but not approve change requests :) | 16:11 |
| anteaya | I am not core infra so I can't kick a server | 16:11 |
| mattymo | fungi, maybe you can https://review.openstack.org/#/c/53656/ | 16:11 |
| anteaya | no, I am not infra-core so I can not approve patches for merging | 16:12 |
| *** senk has quit IRC | 16:12 | |
| *** nati_ueno has quit IRC | 16:12 | |
| *** nati_ueno has joined #openstack-infra | 16:13 | |
| fungi | mattymo: did you /msg chanserv access #fuel add jeblair +AFRfiorstv (i can't check since chanserv tells me i'm not authorized to pull an access list for that channel) | 16:17 |
| *** nati_ueno has quit IRC | 16:17 | |
| *** jcoufal has joined #openstack-infra | 16:18 | |
| *** chuck_ has quit IRC | 16:18 | |
| *** DinaBelova has quit IRC | 16:20 | |
| mattymo | fungi, done | 16:20 |
| *** esheffield_ has joined #openstack-infra | 16:20 | |
| anteaya | mattymo just to give you a sense, fungi clarkb mordred and jeblair (who is on holidays this week) are all infra-core | 16:23 |
| anteaya | myself pleia2 and zaro are also infra team members though not infra-core, but pleia2 and zaro are core on other projects | 16:24 |
| anteaya | and we are all paid to be here | 16:24 |
| *** ben_duyujie has joined #openstack-infra | 16:26 | |
| *** dstanek has joined #openstack-infra | 16:26 | |
| *** larsbutler has joined #openstack-infra | 16:26 | |
| *** tsufiev has joined #openstack-infra | 16:28 | |
| *** esheffield_ has quit IRC | 16:28 | |
| larsbutler | hi there, I noticed that the *-coverage jobs on openstack CI have not been run in while: https://jenkins.openstack.org/search/?q=coverage&max=200 | 16:29 |
| tsufiev | hi there! is this channel a suitable places for asking questions about pbr? | 16:29 |
| larsbutler | Many of them, including swift-coverage, keystone-coverage, and nova-coverage, have not been run since around August 5th/6th | 16:29 |
| larsbutler | Is this a bug? | 16:29 |
| *** ryanpetrello has joined #openstack-infra | 16:30 | |
| larsbutler | I ask because I stumbled upon this bug just a while ago, which probably would have been caught if the coverage job for keystone had been running: https://bugs.launchpad.net/keystone/+bug/1245539 | 16:30 |
| uvirtbot | Launchpad bug 1245539 in keystone "Error running test coverage: "no such option: --coverage"" [Medium,In progress] | 16:30 |
| SpamapS | hm, PBR people.. any idea what this problem is: https://bugs.launchpad.net/bugs/1245372 ? | 16:30 |
| uvirtbot | Launchpad bug 1245372 in trove "kick-start mysql is not working" [Undecided,Confirmed] | 16:30 |
| anteaya | tsufiev: it is, though one person knows the most about pbr, the author mordred, so ask away - be prepared for pauses | 16:31 |
| anteaya | mordred: you have a pbr audience | 16:31 |
| tsufiev | anteaya: thank you, will ask him | 16:31 |
| anteaya | tsufiev: post your question in channel and he will read it when he gets to the backscroll | 16:32 |
| anteaya | who knows, maybe someone else can answer too | 16:32 |
| tsufiev | anteaya: yeah, I need a few minutes to type my issue in :) | 16:33 |
| anteaya | type away, or consider a paste.openstack.org | 16:33 |
| anteaya | for code samples and such | 16:33 |
| fungi | larsbutler: the coverage jobs run on either jenkins01 or jenkins02 now that we have multiple masters, and jenkins.o.o is only for security-sensitive jobs at this point | 16:33 |
| *** markmcclain has joined #openstack-infra | 16:34 | |
| *** moted has joined #openstack-infra | 16:34 | |
| larsbutler | fungi: ah, I see | 16:35 |
| *** jpeeler has quit IRC | 16:35 | |
| *** senk1 has quit IRC | 16:35 | |
| *** julim has quit IRC | 16:35 | |
| larsbutler | fungi: I found the keystone build failure associated with my issue: https://jenkins01.openstack.org/job/keystone-coverage/127/console | 16:35 |
| *** jpeeler has joined #openstack-infra | 16:35 | |
| larsbutler | so, is coverage just not a critical part of the build? | 16:36 |
| *** guohliu has quit IRC | 16:36 | |
| *** thedodd has quit IRC | 16:37 | |
| fungi | larsbutler: i think the discussion has come up before, as far as making coverage jobs latch or ratchet style enforcement, but we currently don't gate on it no | 16:37 |
| ArxCruz | fungi: mordred it works for me changing the devstack-vm.yaml to templates and adding a File(...) on puppet | 16:37 |
| ArxCruz | http://paste.openstack.org/show/50064/ | 16:37 |
| fungi | ArxCruz: puppet doesn't complain or blow away the templated file then, eh? sounds like a potential solution | 16:38 |
| ArxCruz | fungi: I remove the devstack-vm.yaml from files/jenkins_job_builder/config directory | 16:38 |
| ArxCruz | and add it on templates/jenkins_job_builder/config | 16:39 |
| *** adalbas has quit IRC | 16:39 | |
| tsufiev | mordred: have you ever seen the following pbr-ed setup.py strange behaviour? I have an 'ordereddict' requirement line in project's requirements.txt, but when i run python setup.py egg_info (python --version -> 2.7.3), ordereddict line is silently ignored and doesn't get to the <project_name>.egg-info/requires.txt. | 16:40 |
| larsbutler | fungi: okay, understood. Do you think it would be worthwhile better integrate coverage runs into the builds? Or is there a good reason to not do this? | 16:40 |
| tsufiev | mordred: but if i change requirement line to the 'ordereddict==1.1', it gets there | 16:40 |
| ArxCruz | fungi: mordred and in my job I have this now: http://paste.openstack.org/show/50065/ | 16:40 |
| fungi | larsbutler: one problem is that jenkins only has two ways to finish a job. success state or failure state. so we'd need two jobs--one which runs coverage and reports its results, a second which tries to run coverage and gates on whether or not you've broken the job | 16:41 |
| mattymo | fungi, what do we need to do for jeblair on #fuel-dev | 16:43 |
| fungi | larsbutler: that basically means always running coverage twice, and it's slow/expensive on larger code bases. also worth noting, we have the same situation on a lot of other non-voting jobs | 16:43 |
| mattymo | ? | 16:43 |
| fungi | mattymo: oh, i just realized that was #fuel-dev and not #fuel | 16:44 |
| fungi | mattymo: /msg chanserv access #fuel-dev add jeblair +AFRfiorstv | 16:45 |
| mattymo | fungi, it's done | 16:46 |
| fungi | chanserv will actually give me an access listing for that channel, so i should be able to confirm it for him | 16:46 |
| larsbutler | fungi: I don't quite understand. Why does there have to be one to run coverage & report, plus one more to run coverage & vote? (I'm totally new to the project, so I'm just trying to wrap my head around things) | 16:47 |
| fungi | mattymo: yep! i see him in there now. thanks! | 16:47 |
| mattymo | fungi, why do we need this? #puppet-openstack doesn't have any core members | 16:47 |
| mattymo | and it has gerritbot | 16:48 |
| *** guohliu has joined #openstack-infra | 16:48 | |
| ryanpetrello | https://review.openstack.org/#/c/42393/ anybody want to finally push this one through :)? | 16:48 |
| clarkb | morning | 16:50 |
| fungi | mattymo: i think it was recently realized that if that's not done pretty much as soon as the channel is registered, it becomes much, much harder for us to get access later. apparently jeblair has a backlog of channels he's trying to get founder status on so we can help troubleshoot issues with the bots we're managing on them | 16:50 |
| *** lnxnut has joined #openstack-infra | 16:51 | |
| mattymo | fungi, ok | 16:51 |
| fungi | in a lot of cases the channel gets registered by a lone founder who later disappears and then nobody has control over it any longer | 16:51 |
| mattymo | fungi, so can it be finished today and pushed through? are we waiting on him? | 16:51 |
| mattymo | fungi, and I trust you folks to be the benevolent dictators) | 16:52 |
| *** vipul has quit IRC | 16:53 | |
| fungi | mattymo: since i was able to confirm it in jeblair's absence (he's vacationing at the moment), if clarkb wants to get a double-check in on it we can probably put it through | 16:53 |
| *** vipul has joined #openstack-infra | 16:53 | |
| *** dcramer_ has quit IRC | 16:53 | |
| *** ArxCruz has quit IRC | 16:54 | |
| mattymo | clarkb, are you feeling generous today? | 16:55 |
| fungi | https://review.openstack.org/53656 being the change we're talking about | 16:55 |
| *** ArxCruz has joined #openstack-infra | 16:55 | |
| *** dcramer_ has joined #openstack-infra | 16:55 | |
| *** lnxnut has quit IRC | 16:55 | |
| fungi | i suppose https://review.openstack.org/52085 is in the same boat now too | 16:56 |
| *** lnxnut has joined #openstack-infra | 16:56 | |
| *** markmcclain has quit IRC | 16:57 | |
| *** harlowja has joined #openstack-infra | 16:57 | |
| *** ryanpetrello has quit IRC | 16:58 | |
| openstackgerrit | A change was merged to openstack-infra/config: Add IRC bot to #openstack-climate for all Gerrit changes https://review.openstack.org/52085 | 16:58 |
| fungi | not sure why #openstack-climate didn't need our intervention to register though. i wonder whether freenode staff are selectively enforcing namespace registrations? | 16:59 |
| *** ArxCruz has quit IRC | 17:00 | |
| clarkb | mattymo: fungi: looking | 17:00 |
| clarkb | fungi: may have been registered before the restrictions | 17:00 |
| fungi | good point | 17:00 |
| mordred | tsufiev: well, that's definitley a bug - but not the bug you think it is :) | 17:01 |
| *** dcramer_ has quit IRC | 17:01 | |
| mordred | tsufiev: I thought ordereddict was not needed on 2.7 ? | 17:01 |
| reed | which reminds me that #openstack-community is not registered with you guys | 17:02 |
| *** dizquierdo has quit IRC | 17:03 | |
| fungi | mordred: also, in case you missed SpamapS's comment earlier (and/or bug e-mails), today's pbr release is apparently causing problems for (at least) heat and trove... https://launchpad.net/bugs/1245372 | 17:04 |
| uvirtbot | Launchpad bug 1245372 in trove "kick-start mysql is not working" [Undecided,Confirmed] | 17:04 |
| mordred | fungi: awesome | 17:04 |
| fungi | [pbr] Processing SOURCES.txt ... error: No such file or directory | 17:05 |
| *** julim has joined #openstack-infra | 17:05 | |
| fungi | probably their setup mentions a nonexistent file--i haven't checked yet | 17:05 |
| *** markwash has quit IRC | 17:05 | |
| *** ArxCruz has joined #openstack-infra | 17:06 | |
| SpamapS | heat-cfntools specifically | 17:06 |
| tsufiev | mordred: yeah, it is not needed in 2.7, but we have one requirements.txt both for ubuntu (python2.7) and CentOS (python2.6). I can bypass this, nevertheless it's still a bug :) | 17:06 |
| mordred | tsufiev: it's actually a feature :) | 17:06 |
| mordred | tsufiev: it explicitly doesn't install it on 2.7 so that people can have one requirements.txt and not have it pull in unneeded thing | 17:07 |
| mordred | tsufiev: (this was a request from ... I forget, someone) | 17:07 |
| *** Ryan_Lane has joined #openstack-infra | 17:07 | |
| *** guohliu has quit IRC | 17:08 | |
| tsufiev | mordred: I understand it could be nice in some circumstances, but it is confusing that pbr does that thing quietly | 17:08 |
| *** Ryan_Lane has quit IRC | 17:08 | |
| mordred | tsufiev: would you like it to print a message about that it's skipping it on purpose? | 17:09 |
| mordred | would that be helpful? | 17:09 |
| tsufiev | mordred: yes, it would be great | 17:09 |
| *** markmcclain has joined #openstack-infra | 17:12 | |
| fungi | mordred: SpamapS: mmm, so i think maybe the "warning: LocalManifestMaker: standard file '-c' not found [...] error: No such file or directory" could be related after all. trying to see where it's picking up a -c | 17:12 |
| *** yassine has quit IRC | 17:12 | |
| SpamapS | fungi: thanks for investigating. | 17:14 |
| clarkb | mattymo: fungi: approved | 17:14 |
| *** esheffield_ has joined #openstack-infra | 17:15 | |
| *** dkliban has quit IRC | 17:15 | |
| *** Ryan_Lane has joined #openstack-infra | 17:15 | |
| openstackgerrit | A change was merged to openstack-infra/config: add fuel-dev channel to gerritbot https://review.openstack.org/53656 | 17:15 |
| mordred | SpamapS, fungi: I cannot reproduce that | 17:15 |
| mordred | virtualenv foo | 17:15 |
| mordred | foo/bin/pip install heat-cfntools | 17:16 |
| mordred | works for me | 17:16 |
| *** DinaBelova has joined #openstack-infra | 17:16 | |
| *** vipul is now known as vipul-away | 17:16 | |
| *** ruhe has joined #openstack-infra | 17:17 | |
| *** hashar has quit IRC | 17:17 | |
| *** ruhe has quit IRC | 17:17 | |
| fungi | mordred: yeah, i wonder whether something on the reddwarf end is contributing to this | 17:17 |
| *** olaph has quit IRC | 17:17 | |
| fungi | s/reddwarf/trove/ | 17:17 |
| openstackgerrit | Sean Dague proposed a change to openstack-infra/devstack-gate: Let us gate grenade forward https://review.openstack.org/53940 | 17:18 |
| tsufiev | mordred: there was a short network failure in our office, didn't i misssed your replies? | 17:19 |
| fungi | tsufiev: http://eavesdrop.openstack.org/irclogs/%23openstack-infra/%23openstack-infra.2013-10-28.log | 17:19 |
| tsufiev | ok, nothing was missed :) | 17:20 |
| sdague | so before I spend too much time on some of the d-g refactoring, it would be good to know if - https://review.openstack.org/#/c/54132/ and https://review.openstack.org/#/c/54133/ are the kinds of cleanups that people want to see (I'm currently sitting my grenade upgrade stuff behind them) | 17:23 |
| *** flaper87 is now known as flaper87|afk | 17:23 | |
| *** krotscheck has joined #openstack-infra | 17:24 | |
| ArxCruz | ALL: Have you guys seen this problem: http://paste.openstack.org/show/50070/ | 17:24 |
| ArxCruz | I have the ip command in /usr/sbin/ip and I can execute as jenkins, stack, and root users | 17:24 |
| ArxCruz | but when it's running through jenkins it's failing | 17:24 |
| *** vipul-away is now known as vipul | 17:25 | |
| openstackgerrit | Monty Taylor proposed a change to openstack-dev/pbr: Make sure that we ioudly skip versioned 2.6 libs https://review.openstack.org/54152 | 17:27 |
| mordred | tsufiev: ^^ | 17:27 |
| sdague | ArxCruz: /usr/sbin might not be in the path by default on fedora | 17:27 |
| ArxCruz | sdague: i think it's in jenkins default path | 17:27 |
| *** markmcclain has quit IRC | 17:27 | |
| sdague | but this is under the stack user | 17:27 |
| *** yassine has joined #openstack-infra | 17:27 | |
| sdague | so the path would be reset | 17:28 |
| *** yassine has quit IRC | 17:30 | |
| *** olaph has joined #openstack-infra | 17:31 | |
| *** markmcclain has joined #openstack-infra | 17:32 | |
| *** Ryan_Lane has quit IRC | 17:35 | |
| *** mgagne has quit IRC | 17:37 | |
| *** mgagne has joined #openstack-infra | 17:37 | |
| *** mgagne has joined #openstack-infra | 17:37 | |
| tsufiev | mordred: that was fast :) | 17:38 |
| *** fbo is now known as fbo_away | 17:39 | |
| ogelbukh | fungi: mordred: clarkb: gentlemen, could you please tell me what's needed to move a project to stackforge from public github repo? | 17:41 |
| *** reed has quit IRC | 17:41 | |
| ogelbukh | i filed a change request already to the config repo: https://review.openstack.org/#/c/53858/ | 17:41 |
| pleia2 | ogelbukh: have you seen http://ci.openstack.org/stackforge.html | 17:42 |
| ogelbukh | sure | 17:42 |
| pleia2 | ok, cool :) | 17:42 |
| ogelbukh | i think i'll need to provide some additional information | 17:42 |
| *** tkammer_ has quit IRC | 17:43 | |
| pleia2 | yeah, you'll need to request a group set up with initial core reviewers | 17:43 |
| fungi | ogelbukh: mostly time. we're a bit backlogged with releases, upcoming summit preparations, conferences and travel. also, stackforge-related work is generally a lower priority for us than work on openstack itself | 17:44 |
| ogelbukh | pleia2: thanks! | 17:44 |
| *** markwash has joined #openstack-infra | 17:44 | |
| fungi | but we do try to get through and review all submitted changes | 17:44 |
| pleia2 | ogelbukh: but that's after your change has merged, so as fungi says - time :) | 17:44 |
| ogelbukh | fungi: fair, thamnk you! | 17:45 |
| ogelbukh | -m | 17:45 |
| clarkb | fungi: ogelbukh: ya this week is going to be hard to get "real" work done | 17:48 |
| *** reed has joined #openstack-infra | 17:48 | |
| fungi | interesting statistic... 189 projects in review.openstack.org. 42% stackforge, 35% openstack, 18% openstack-infra, 5% openstack-dev | 17:49 |
| pleia2 | stackforge sure has grown | 17:49 |
| *** sarob has joined #openstack-infra | 17:51 | |
| ogelbukh | stackforge is representing ecosystem, so it's natural, i guess | 17:51 |
| reed | does seafile support storing files in swift? | 17:53 |
| pleia2 | fungi: worth some informal chatting about at summit re: allocation of resources? | 17:53 |
| fungi | pleia2: probably at least twice... maybe one of those without beer | 17:55 |
| pleia2 | ++ | 17:55 |
| openstackgerrit | Sean Dague proposed a change to openstack-infra/devstack-gate: Let us gate grenade forward https://review.openstack.org/53940 | 18:02 |
| sdague | so close.... | 18:02 |
| sdague | fungi: so I think https://review.openstack.org/53940 is functionally right, we just need to do this lockstep thing with the grenade upgrade | 18:02 |
| sdague | so I need to make it run grizzly as stable on master until we get working havana => master grenade | 18:03 |
| fungi | can that be phased, or will both changes have to be force-merged in tandem? | 18:03 |
| clarkb | sdague: I am reviewing 53490 | 18:04 |
| * fungi is as well | 18:06 | |
| klrmn | lifeless: testresources does not work out-of-the-box with nose, i had to write a nose plugin to force it to use the OptimisingTestSuite. i'll let you know when i publish the plugin…first it has to survive the fire that will be refactoring all my tests to use it | 18:07 |
| lifeless | klrmn: cool | 18:07 |
| clarkb | sdague: reviewed | 18:09 |
| bodepd_ | fungi: thanks fungi. I didn't even realize those branches were there. Looking into it now | 18:09 |
| sdague | clarkb: thanks, good catches | 18:11 |
| sdague | I'm going to let this run to completion to see if the grenade job actually worked this time | 18:11 |
| clarkb | wfm | 18:11 |
| sdague | fungi: yes, so the attempt as phasing it is doing the wrong thing in this change and setting old=grizzly for master | 18:12 |
| sdague | then fixing grenade | 18:12 |
| sdague | then fixing this to old=havana for master | 18:12 |
| sdague | clarkb: my comments on your comments there. On missed variable, and one bad merge :) | 18:13 |
| fungi | completely unrelated, but i confirmed you can get git tab completion to do some nasty local code execution tricks by backticking a command in a tag name, as long as you respect git's limitations on what characters are valid for a tag | 18:15 |
| fungi | it's actually pretty funny | 18:15 |
| *** dcramer_ has joined #openstack-infra | 18:15 | |
| *** ruhe has joined #openstack-infra | 18:16 | |
| clarkb | fungi: your idea of funny is disturbing :) | 18:17 |
| fungi | for example, create a backdoor.sh (doing something like "touch pwnd") script in your repo and make it executable, then tag a commit as `./backdoor.sh` and try tab-completing git branch -D | 18:17 |
| *** adalbas has joined #openstack-infra | 18:17 | |
| fungi | this cropped up from some earlier work i did on trying to inject arbitrary code in tags we reference in our release jobs, which turned out not to be viable but the side effect was that i started getting file-not-found errors when tab-completing refnames in that checkout. only just noticed it | 18:19 |
| reed | I love CLAs corner cases | 18:19 |
| *** yaguang has joined #openstack-infra | 18:19 | |
| *** SergeyLukjanov is now known as _SergeyLukjanov | 18:19 | |
| *** _SergeyLukjanov has quit IRC | 18:20 | |
| *** DinaBelova has quit IRC | 18:21 | |
| *** DinaBelova has joined #openstack-infra | 18:22 | |
| clarkb | fungi: that is pretty funny. are '$'s allowed in tag names? | 18:23 |
| clarkb | something like `echo "Hello, $USER this is git. Nice to meet you."` would be an excellent tag name | 18:23 |
| fungi | clarkb: well, the biggest trick is that spaces aren't allowed | 18:24 |
| clarkb | well we use underscores then! | 18:24 |
| clarkb | and not echo | 18:24 |
| fungi | but you could stuff that echo into a script in the repo and have the tag exec it | 18:24 |
| *** ruhe has quit IRC | 18:25 | |
| *** katyafervent has quit IRC | 18:26 | |
| fungi | oh, also it seems stdout gets swallowed, but things like file interaction still work | 18:26 |
| *** mattymo has quit IRC | 18:26 | |
| *** tsufiev has quit IRC | 18:26 | |
| lifeless | jgriffith: hey | 18:27 |
| lifeless | jgriffith: you should totally hang out in #tripleo | 18:27 |
| lifeless | fungi: hard space! | 18:28 |
| *** Bada has quit IRC | 18:28 | |
| fungi | heh | 18:28 |
| fungi | i think tag names may be ascii only, so utf-8 hard space wouldn't be an option | 18:28 |
| fungi | a very strict printable ascii subset, if memory serves (i looked it up in the manpage a while back) | 18:29 |
| *** dkliban has joined #openstack-infra | 18:30 | |
| *** fallenpegasus has quit IRC | 18:30 | |
| *** SergeyLukjanov has joined #openstack-infra | 18:31 | |
| *** esker has joined #openstack-infra | 18:35 | |
| *** fallenpegasus has joined #openstack-infra | 18:36 | |
| *** changbl has quit IRC | 18:37 | |
| *** ben_duyujie has quit IRC | 18:43 | |
| openstackgerrit | Arx Cruz proposed a change to openstack-infra/config: Adding support to a generic zuul server in JJB for devstack-gate https://review.openstack.org/54160 | 18:45 |
| ArxCruz | fungi: mordred ^ | 18:46 |
| *** dkliban has quit IRC | 18:46 | |
| *** thedodd has joined #openstack-infra | 18:47 | |
| *** fallenpegasus has quit IRC | 18:47 | |
| *** changbl has joined #openstack-infra | 18:50 | |
| *** ArxCruz has quit IRC | 18:51 | |
| *** ArxCruz has joined #openstack-infra | 18:51 | |
| *** yaguang has quit IRC | 18:54 | |
| *** larsbutler has quit IRC | 18:54 | |
| *** jergerber has joined #openstack-infra | 18:56 | |
| *** hogepodge has quit IRC | 18:59 | |
| openstackgerrit | Arx Cruz proposed a change to openstack-infra/config: Adding support to a generic zuul server in JJB for devstack-gate https://review.openstack.org/54160 | 19:00 |
| *** markmcclain has quit IRC | 19:02 | |
| *** dims has quit IRC | 19:07 | |
| openstackgerrit | Anita Kuno proposed a change to openstack-infra/config: Create a seafile instance. https://review.openstack.org/54030 | 19:07 |
| *** dims has joined #openstack-infra | 19:10 | |
| openstackgerrit | Arx Cruz proposed a change to openstack-infra/config: Adding support to a generic zuul server in JJB for devstack-gate https://review.openstack.org/54160 | 19:11 |
| *** rainya has quit IRC | 19:13 | |
| *** melwitt has joined #openstack-infra | 19:14 | |
| *** DinaBelova has quit IRC | 19:15 | |
| openstackgerrit | Arx Cruz proposed a change to openstack-infra/config: Adding support to a generic zuul server in JJB for devstack-gate https://review.openstack.org/54160 | 19:16 |
| ArxCruz | fungi: mordred do I need, in my openstack installation, open the 35357 door in order to jobs run? | 19:17 |
| *** DinaBelova has joined #openstack-infra | 19:19 | |
| *** DinaBelova has quit IRC | 19:22 | |
| *** jcoufal has quit IRC | 19:24 | |
| *** johnthetubaguy has quit IRC | 19:24 | |
| *** jcoufal has joined #openstack-infra | 19:24 | |
| openstackgerrit | Arx Cruz proposed a change to openstack-infra/config: Adding support to a generic zuul server in JJB for devstack-gate https://review.openstack.org/54160 | 19:26 |
| *** danger_fo_away has quit IRC | 19:27 | |
| *** nicedice__ has joined #openstack-infra | 19:28 | |
| sdague | fungi: did you have any other comments on this - https://review.openstack.org/#/c/53940/ ? | 19:29 |
| sdague | before I respin with the fixes clarkb suggested | 19:29 |
| sdague | as I managed to get it to pass grenade... yay! | 19:29 |
| *** dstufft has quit IRC | 19:29 | |
| *** dstufft has joined #openstack-infra | 19:30 | |
| *** LinuxJedi_ has joined #openstack-infra | 19:30 | |
| *** danger_fo_away has joined #openstack-infra | 19:31 | |
| *** nati_ueno has joined #openstack-infra | 19:31 | |
| openstackgerrit | Arx Cruz proposed a change to openstack-infra/config: Adding support to a generic zuul server in JJB for devstack-gate https://review.openstack.org/54160 | 19:31 |
| *** yamahata has quit IRC | 19:31 | |
| *** jeblair has quit IRC | 19:31 | |
| *** LinuxJedi has quit IRC | 19:31 | |
| *** uvirtbot has quit IRC | 19:32 | |
| *** yamahata has joined #openstack-infra | 19:32 | |
| *** jeblair has joined #openstack-infra | 19:33 | |
| *** davidhadas has quit IRC | 19:34 | |
| *** markwash_ has joined #openstack-infra | 19:34 | |
| fungi | sdague: no, seems safe enough, those couple items aside | 19:34 |
| *** mtreinish has quit IRC | 19:35 | |
| openstackgerrit | Joe Gordon proposed a change to openstack-dev/hacking: Continue reserving H6xx namespace for calling methods https://review.openstack.org/54165 | 19:35 |
| fungi | side note, we have rather many branch variables in devstack-gate scripts. wish there were better ways to distinguish them' | 19:36 |
| jog0 | clarkb, sdague: ^^ last patch before ready to cut hacking 0.8.0 | 19:36 |
| *** markwash has quit IRC | 19:36 | |
| *** markwash_ is now known as markwash | 19:36 | |
| *** mtreinish has joined #openstack-infra | 19:40 | |
| fungi | clarkb: good news on the git tag tab completion command injection front... can't seem to reproduce running debian's 1.8.4~rc3-1 package | 19:48 |
| sdague | fungi: yes, for sure, something maybe to think about in the grenade session | 19:48 |
| fungi | hopefully the old bash tab completion file was screwy | 19:49 |
| openstackgerrit | Sean Dague proposed a change to openstack-infra/devstack-gate: Let us gate grenade forward https://review.openstack.org/53940 | 19:53 |
| fungi | yeah, i can reproduce it on precise :( | 19:55 |
| sdague | ok, so that should be good to go. fungi / clarkb you guys ok with the refactor patches under it? | 19:59 |
| sdague | or should I rebase that on master? | 19:59 |
| openstackgerrit | Anita Kuno proposed a change to openstack-infra/config: Create a seafile instance. https://review.openstack.org/54030 | 20:01 |
| *** mrodden1 has quit IRC | 20:01 | |
| *** mrodden has joined #openstack-infra | 20:01 | |
| *** emagana has joined #openstack-infra | 20:02 | |
| wenlock | quick config question for gerrit group settings. is there a reason why Release Managers group is visible to all but Stable Maintainers gorup is not? | 20:03 |
| *** LinuxJedi_ is now known as LinuxJedi | 20:03 | |
| *** sarob has quit IRC | 20:03 | |
| wenlock | same question for External Testing Tools group... why is that visible and not Continuous Integration Tools group? | 20:05 |
| *** sarob has joined #openstack-infra | 20:05 | |
| fungi | wenlock: probably just oversight on our part. about the only things i feel we should keep non-visible are the cla-related groups, but it's probably something we ought to discuss with the other infra team members | 20:07 |
| wenlock | fungi, thanks... i was just trying to understand it... i get some groups are more sensative than others | 20:08 |
| wenlock | i was thinking the logic might have been reversed though | 20:08 |
| fungi | in particular we started making the .*-core groups visible so people with reviews knew who to reach out to if necessary | 20:09 |
| wenlock | ie; protect the managers more than the maintainers.... or same for the ci accounts, protect the names since they arn't real people | 20:09 |
| fungi | but ultimately i think there's very little sensitive about the vast majority of our groups | 20:09 |
| wenlock | yeah, *-core groups being visible makes perfect sense | 20:09 |
| wenlock | great, thanks | 20:10 |
| fungi | it's mostly that gerrit defaults to making group details and membership invisible unless you're a member of that group, but we override that for all of our automatically-generated groups these days | 20:10 |
| *** hashar has joined #openstack-infra | 20:11 | |
| wenlock | yeah... it was bugging me because when you add a group via ssh, if you try automating the make visible flag, there is no way to do it via api that i can see. if you insert the entry into mysql, then gerrit will want a restart to make it visible. | 20:12 |
| *** markmcclain has joined #openstack-infra | 20:12 | |
| openstackgerrit | Anita Kuno proposed a change to openstack-infra/config: Create a seafile instance. https://review.openstack.org/54030 | 20:13 |
| *** dprince has quit IRC | 20:15 | |
| *** julim has quit IRC | 20:15 | |
| *** changbl has quit IRC | 20:21 | |
| fungi | wenlock: you want 'gerrit create-group --visible-to-all mynewgroup' | 20:21 |
| wenlock | fungi, awesome! | 20:21 |
| *** jcoufal has quit IRC | 20:22 | |
| wenlock | didn't even see that as an option... | 20:22 |
| fungi | wenlock: http://git.openstack.org/cgit/openstack-infra/gerritlib/tree/gerritlib/gerrit.py#n116 | 20:22 |
| fungi | that's what we're calling from the manage-projects script in jeepyb | 20:22 |
| wenlock | perfect | 20:23 |
| sdague | jog0: +2 on your previous review request | 20:24 |
| *** mrodden has quit IRC | 20:25 | |
| *** vipul has quit IRC | 20:26 | |
| *** vipul has joined #openstack-infra | 20:27 | |
| wenlock | fungi, just tried it. that works | 20:28 |
| wenlock | no server restart needed :D | 20:28 |
| fungi | crap, so the puppet refactoring for mysql in gerrit which i approved over the weekend seems to be having a puppet problem on review.o.o which did not manifest on review-dev | 20:30 |
| fungi | fix on the way | 20:31 |
| wenlock | hmm, i had looked at that over the weekend as well... it looked like an ok change. | 20:32 |
| fungi | for those playing along at home, the error is... "Could not retrieve catalog from remote server: Error 400 on SERVER: Invalid parameter mysql_password at /opt/config/production/modules/openstack_project/manifests/review.pp:171 on node review.openstack.org" | 20:33 |
| *** dcramer_ has quit IRC | 20:33 | |
| *** sarob has quit IRC | 20:33 | |
| *** sarob has joined #openstack-infra | 20:34 | |
| wenlock | ahh yes, we had forked that file, forgot to post on the forum that it broke us :P | 20:34 |
| wenlock | i mean irc | 20:34 |
| *** changbl has joined #openstack-infra | 20:36 | |
| clarkb | gerrit defaults to non visible groups. I think we just negelected to make them visible | 20:37 |
| *** ericloo has joined #openstack-infra | 20:37 | |
| wenlock | clarkb, so they should all be visible? I mean the groups we create by hand per this site: http://ci.openstack.org/gerrit.html (groups section) | 20:38 |
| clarkb | wenlock: I think so, there might be a small argument that Project Bootstrappers shouldn't be | 20:38 |
| *** sarob has quit IRC | 20:39 | |
| *** denis_makogon has joined #openstack-infra | 20:39 | |
| wenlock | clarkb: ok, cool. i was improving my changes for gerrit server setup to bootstrap gerrit without doing all the manual setup | 20:39 |
| *** changbl has quit IRC | 20:40 | |
| *** openstackgerrit has quit IRC | 20:41 | |
| *** openstackgerrit has joined #openstack-infra | 20:41 | |
| *** sarob has joined #openstack-infra | 20:42 | |
| clarkb | sdague: I think we may need to sort out when to setup the workspace | 20:44 |
| clarkb | sdague: I think it needs to happen before the reexec in order to make sure the new d-g is abailable to reexec | 20:44 |
| clarkb | sdague: what if you put the branch selection and setup of new workspace before reexec? | 20:45 |
| clarkb | I think that gives us maximum coverage of the reexec testing | 20:45 |
| *** davidhadas has joined #openstack-infra | 20:45 | |
| openstackgerrit | Jeremy Stanley proposed a change to openstack-infra/config: Don't pass mysql passwords to project gerrit class https://review.openstack.org/54177 | 20:46 |
| fungi | clarkb: that could stand some ninja ^ | 20:46 |
| clarkb | fungi: looking | 20:47 |
| fungi | it was correct in review_dev.pp but not in review.pp, so testing on review-dev didn't expose it | 20:47 |
| clarkb | fungi: how does review.o.o know its mysql info? | 20:47 |
| fungi | clarkb: up in the gerrit::mysql block | 20:48 |
| clarkb | fungi: gotcha | 20:48 |
| fungi | it's a brown bag fix from something which got missed in 6a2e31d | 20:48 |
| clarkb | fungi: lgtm, feel free to push through | 20:49 |
| openstackgerrit | A change was merged to openstack-infra/config: Don't pass mysql passwords to project gerrit class https://review.openstack.org/54177 | 20:52 |
| fungi | fwiw, that ^ got "tested" in production (applying it un-broke puppet on review.o.o) | 20:52 |
| clarkb | k | 20:52 |
| sdague | clarkb: ok, I thought you wanted me to move it up. I can move it back down to after the workspaces | 20:52 |
| clarkb | I figured you had done that | 20:52 |
| clarkb | sdague: I think we need something in the middle | 20:52 |
| clarkb | sdague: we want the section that reexecs to happen as early as possible, but it must happen after the new workspace is configured | 20:53 |
| fungi | basically anything which runs before the reexec is not self-tested | 20:54 |
| sdague | ok | 20:54 |
| *** luhrs1 has joined #openstack-infra | 20:55 | |
| clarkb | there may be some way to just deal with d-g repo first then check reexec then setup workspaces proper | 20:55 |
| clarkb | do something like gerrit git prep just for d-g | 20:55 |
| *** moted_ has joined #openstack-infra | 20:56 | |
| *** hashar_ has joined #openstack-infra | 20:56 | |
| ArxCruz | fungi: https://review.openstack.org/#/c/54160/ i believe the failure in gate-config-layout it's because I move the devstack-gate jobs is it possible to you verify ? I remove all the blank and tabs spaces :) | 20:56 |
| fungi | ArxCruz: in a bit. i have to cook dinner but i'll try to look while i do so | 20:57 |
| *** ^d has quit IRC | 20:57 | |
| *** hashar has quit IRC | 20:57 | |
| *** moted has quit IRC | 20:57 | |
| *** hashar_ is now known as hashar | 20:57 | |
| *** moted_ is now known as moted | 20:57 | |
| *** che-arne has quit IRC | 20:57 | |
| ArxCruz | fungi: take your time :) | 20:58 |
| *** aspiers has joined #openstack-infra | 20:59 | |
| sdague | clarkb: so, in patchset 9 - https://review.openstack.org/#/c/53940/9/devstack-vm-gate-wrap.sh I had this down after the workspace setup | 21:00 |
| sdague | just go back to that? | 21:00 |
| *** ^d has joined #openstack-infra | 21:01 | |
| *** changbl has joined #openstack-infra | 21:02 | |
| clarkb | ArxCruz: fungi: why do shell scripts in jenkins jobs need to know where ZUUL is located? I think it might be better to supply that info from zuul itself | 21:02 |
| clarkb | sdague: we can go back to that, it would nice if there is something in between that we can do | 21:02 |
| *** Ajaeger1 has joined #openstack-infra | 21:03 | |
| sdague | ok, let's start there, and we can make it better later | 21:03 |
| clarkb | sdague: do the bare minimum to sort out branches, then setup workspace/new, check reexec then setup workspace old | 21:03 |
| ArxCruz | clarkb: oh... | 21:03 |
| sdague | clarkb: so what's the concern on setting up workspace old before it? | 21:03 |
| clarkb | ArxCruz: it might need a patch to zuul to do that, but I think since zuul is driving the whole thing it can just inject that into the build parameter list | 21:03 |
| ArxCruz | clarkb: thanks for make a lot of ctrl+c ctrl+v useless | 21:03 |
| sdague | just that it's slow? | 21:03 |
| ArxCruz | :) | 21:03 |
| *** mrodden has joined #openstack-infra | 21:03 | |
| clarkb | sdague: no, changes to d-g won't be self tested on their ability to setup workspace old | 21:04 |
| ArxCruz | clarkb: yes, zuul is better, I didn't though that | 21:04 |
| fungi | ArxCruz: yes, i think you're right about the failing layout and comparison jobs... they'd need extra intelligence to deal with one of the config files being a puppet template | 21:04 |
| fungi | ArxCruz: clarkb: and yes, if we can supply that from zuul directly, all the better | 21:04 |
| ArxCruz | clarkb: but zuul send ZUUL_URL info ? | 21:04 |
| fungi | didn't even think of that earlier when discussing it with ArxCruz and mordred | 21:05 |
| clarkb | ArxCruz: yeah, zuul is triggering these jobs and injecting environment variables, rather than pass that data through in weird third party places, just do it directly | 21:05 |
| Ajaeger1 | infra team, we're splitting up the install guide and I would love some good review on this patch https://review.openstack.org/54076, please | 21:05 |
| clarkb | then you can point at different zuuls without any configuration changes | 21:05 |
| sdague | clarkb: ok, so what we really need is devstack-gate setting itself up independently, ok, let me slice this again. | 21:06 |
| clarkb | sdague: ya that would be the proper fix | 21:06 |
| clarkb | sdague: fwiw if that is terrible I am fine doing what patchset 9 did for now | 21:06 |
| ArxCruz | clarkb: so, if pass ZUUL_URL in zuul, it will export the variable correctly without the need of changes in jjb ? | 21:07 |
| sdague | so, honestly, I think the proper fix requires a lot more refactoring of setup_workspace so that it can take a project as a param | 21:07 |
| sdague | so I think, given that we wedged ourself this morning on a novaclient change, I'd like to get the patch 9 version moving forward now | 21:07 |
| clarkb | ArxCruz: correct if you look at https://jenkins02.openstack.org/job/check-tempest-devstack-vm-full/3598/parameters/? you see there are parameters that are being directly passed into the job from zuul | 21:08 |
| sdague | because we have real possible breaks happening. Then we do this right later. | 21:08 |
| *** beagles has quit IRC | 21:08 | |
| *** thomasem has quit IRC | 21:08 | |
| ArxCruz | clarkb: okay, cool | 21:08 |
| clarkb | sdague: sounds good to me, sorry I didn't fully grasp why it was moved initially. Want to push up a thing with the other fix in patchset 10 that is otherwise patchset 9? | 21:08 |
| ArxCruz | fungi: so, can I just abandon https://review.openstack.org/#/c/54160/ ? | 21:09 |
| *** yamahata has quit IRC | 21:09 | |
| fungi | ArxCruz: yeah, there is definitely a better, more elegant way | 21:09 |
| *** yamahata has joined #openstack-infra | 21:09 | |
| ArxCruz | :) | 21:09 |
| ArxCruz | okay, hacking zuul now... | 21:10 |
| openstackgerrit | Sean Dague proposed a change to openstack-infra/devstack-gate: Let us gate grenade forward https://review.openstack.org/53940 | 21:11 |
| sdague | clarkb: yeh, I split it a little bit to do the old setup after | 21:11 |
| sdague | and left a big comment | 21:11 |
| sdague | clarkb: also, per previous question, that's sitting on top of 2 refactoring clean ups. Would be good to know if you are good with those as well, as I think it would be good to get this whole series in soon, so we can actually land the havana => master grenade changes, then get that switched on | 21:13 |
| *** beagles has joined #openstack-infra | 21:13 | |
| *** misterpink has joined #openstack-infra | 21:14 | |
| *** misterpink has quit IRC | 21:15 | |
| *** misterpink has joined #openstack-infra | 21:18 | |
| *** fbo_away is now known as fbo | 21:19 | |
| zul | i think we should turn on the gate for py3 for cinderclient/novaclient | 21:19 |
| clarkb | zul: It should be on for novaclient | 21:21 |
| zul | sweet | 21:21 |
| clarkb | zul: but we should do cinderclient too if it is ready | 21:24 |
| *** dcramer_ has joined #openstack-infra | 21:25 | |
| zul | clarkb: its ready ill do it tonight | 21:25 |
| *** mrodden1 has joined #openstack-infra | 21:28 | |
| *** jhesketh has joined #openstack-infra | 21:30 | |
| *** mrodden has quit IRC | 21:31 | |
| *** alcabrera has quit IRC | 21:32 | |
| *** dkranz has quit IRC | 21:34 | |
| fungi | jgriffith: python-cinderclient works and passes tests on python 3.3 now? | 21:41 |
| fungi | zul: that's awesome news, btw | 21:41 |
| clarkb | jog0: sdague: do we expect py33 and pypi tests to not pass for hacking https://review.openstack.org/#/c/54165/1 I ask because novaclient and soon cinderclient are gating on py33 and while flake8 isn't run on them with py33 it isn't far fetched that someone may try that | 21:41 |
| *** ruhe has joined #openstack-infra | 21:41 | |
| fungi | i thought they did pass at one point and then regressed? | 21:42 |
| clarkb | yeah I think that may be what happened, just checking that it is ok | 21:42 |
| *** mriedem has quit IRC | 21:43 | |
| * fungi could be misremembering | 21:43 | |
| *** ruhe has quit IRC | 21:43 | |
| fungi | it's past beer-o'clock here now, so my memory is increasing in fuzziness | 21:43 |
| fungi | well, technically saké-o'clocky but who's counting? | 21:44 |
| *** justinabrahms has quit IRC | 21:44 | |
| *** thomasem has joined #openstack-infra | 21:44 | |
| *** thomasem has quit IRC | 21:45 | |
| *** luhrs1 has quit IRC | 21:45 | |
| *** thomasem has joined #openstack-infra | 21:45 | |
| *** justinlilly has joined #openstack-infra | 21:46 | |
| *** che-arne has joined #openstack-infra | 21:46 | |
| *** justinlilly is now known as justinabrahms | 21:47 | |
| *** rakhmerov has quit IRC | 21:47 | |
| *** justinabrahms has quit IRC | 21:52 | |
| *** pcm_ has quit IRC | 21:52 | |
| openstackgerrit | Arx Cruz proposed a change to openstack-infra/zuul: Add the possibility to pass the zuul url https://review.openstack.org/54196 | 21:53 |
| Ajaeger1 | fungi, I can only offer virtual beers and will throw a virtual round for reviews ;) - but it's really good German beer ;) | 21:53 |
| ArxCruz | clarkb: fungi I think I did it right now ^ | 21:53 |
| *** slong has joined #openstack-infra | 21:53 | |
| fungi | Ajaeger1: if it comes in liter steins, i'll start looking for plane tickets ;) | 21:54 |
| fungi | ArxCruz: lookin' | 21:54 |
| Ajaeger1 | :) | 21:54 |
| *** dcramer_ has quit IRC | 21:54 | |
| SpamapS | fungi: did anything ever come of the pbr bug noted this morning? | 21:55 |
| *** weshay has quit IRC | 21:56 | |
| clarkb | ArxCruz: that change looks good to me other than the pep8 failure | 21:57 |
| fungi | SpamapS: we... think? it's trove-setup-specific? | 21:57 |
| *** Ajaeger1 has quit IRC | 21:57 | |
| fungi | SpamapS: were you able to recreate it by installing via pip? | 21:57 |
| ArxCruz | clarkb: it's really annoying these blank lines | 21:59 |
| fungi | neither mordred nor i had any luck coming up with a trivial test case to cause that error (note: redstack mysql kickstart doesn't count as "trivial") | 21:59 |
| *** esker has quit IRC | 21:59 | |
| anteaya | I feel soiled having to type the code I have to type. like being forced 2 tp lik ts | 21:59 |
| anteaya | *shudder* | 21:59 |
| openstackgerrit | Arx Cruz proposed a change to openstack-infra/zuul: Add the possibility to pass the zuul url https://review.openstack.org/54196 | 21:59 |
| clarkb | anteaya: huh? did you end up in AIM land? | 22:00 |
| anteaya | ew | 22:00 |
| fungi | i thought her keyboard started rebelling | 22:00 |
| anteaya | reading how to puppet downloading and unpacking a tarball | 22:00 |
| anteaya | sometime (rarely, but sometimes) I wish I drank | 22:00 |
| anteaya | the last time was when I had to take a live squirrel out of the wall | 22:01 |
| anteaya | this is another one | 22:01 |
| fungi | anteaya: sometimes the computers do respond more favorably to beer | 22:01 |
| anteaya | I'll pour beer on my keyboard | 22:01 |
| SpamapS | fungi: https://bugs.launchpad.net/ubuntu/+source/python-pbr/+bug/1245676 | 22:02 |
| fungi | it looked like seafile had instructions for building/installing from source, but it did require some additional dependencies be installed on the system | 22:02 |
| anteaya | would that be easier than the tarball? | 22:02 |
| anteaya | I would still have to download and unpack something, get it in the right directory | 22:03 |
| anteaya | use the exec {} resource | 22:03 |
| anteaya | would I not? | 22:03 |
| fungi | SpamapS: that looks a lot more specific--thanks | 22:03 |
| fungi | anteaya: in that situation we at least have examples of doing that via the vcsrepo module | 22:04 |
| SpamapS | fungi: np, thank andrew_plunk though. :) | 22:04 |
| anteaya | fungi I saw vcsrepo usage for git cloning | 22:04 |
| fungi | SpamapS: well, i'm thanking you for bringing it to my attention sooner than i got back around to that maildir | 22:04 |
| anteaya | does it work with tarballs as well? | 22:04 |
| anteaya | I'll look | 22:04 |
| *** ^d has quit IRC | 22:04 | |
| *** andrew_plunk has joined #openstack-infra | 22:04 | |
| anteaya | all hail andrew_plunk | 22:05 |
| SpamapS | fungi: thank you fungi, thank you #openstack-infra.. thank you, thank you, thank you-ooo-ooo</alanis> | 22:05 |
| andrew_plunk | ahaha hello | 22:05 |
| *** ^d has joined #openstack-infra | 22:05 | |
| fungi | anteaya: you'd use it to clone their git repository and then follow their installation from source instructions (but write a script to encode them) | 22:05 |
| clarkb | anteaya: there are a couple of archive puppet modules on puppet forge that look like they may make the situation more san | 22:05 |
| anteaya | okay, let me try that | 22:05 |
| clarkb | *sane | 22:05 |
| fungi | andrew_plunk: many thanks for responsible and detailed bug reporting! | 22:05 |
| anteaya | clarkb: the puppet forge modules are named archive? | 22:05 |
| andrew_plunk | np fungi: I was just trying to follow the instructions on launchpad | 22:06 |
| anteaya | all hail andrew_plunk's responsible and detailed bug reporting | 22:06 |
| anteaya | I haven't read it yet myself, I'm trusting fungi's judgment here | 22:06 |
| andrew_plunk | ahaha | 22:06 |
| *** ^d has quit IRC | 22:07 | |
| * anteaya reads the andrew_plunk bug report | 22:07 | |
| * anteaya nods agreement with fungi's original assessment | 22:08 | |
| fungi | i'm reading through it now, but it has a full transcript of pip install failure for pbr | 22:08 |
| *** thomasem has quit IRC | 22:08 | |
| fungi | looks vaguely like a pypi problem... | 22:08 |
| * anteaya goes back to puppet examples | 22:09 | |
| anteaya | sigh | 22:09 |
| ArxCruz | clarkb: fungi jenkins gave the +1 so if you guys don't mind, go there and gave your blessings :D | 22:09 |
| fungi | ArxCruz: did that work on your systems? (just curious) | 22:10 |
| ArxCruz | fungi: I'm still waiting some job who uses that :/ | 22:10 |
| ArxCruz | unless there's a way to force the job | 22:10 |
| *** ^d has joined #openstack-infra | 22:12 | |
| *** ^d has joined #openstack-infra | 22:12 | |
| sdague | ok, time to head home for the night - d-g changes: https://review.openstack.org/#/c/54132/, https://review.openstack.org/#/c/54133/, https://review.openstack.org/#/c/53940/ are needed to get grenade actually working on havana (and set up the forward grenade upgrades) | 22:13 |
| sdague | the above would have prevented the python-novaclient wedge this morning (which I assume is going to be reproposed as soon as someone realizes we reverted them) | 22:13 |
| fungi | heh. it'll make an excellent test ;) | 22:14 |
| *** senk1 has joined #openstack-infra | 22:14 | |
| sdague | https://review.openstack.org/#/c/53739/ - is d-g protection from tempest being able to wedge the world if we aren't up to date on requirements | 22:14 |
| *** lnxnut_ has joined #openstack-infra | 22:14 | |
| *** andrew_plunk has quit IRC | 22:14 | |
| sdague | fungi: yep, that was going to be my first test after that's merged, revert the revert | 22:14 |
| anteaya | clarkb: what are the names of the puppet modules to which you refer? | 22:14 |
| anteaya | so far I haven't seen them myself | 22:15 |
| *** dkliban has joined #openstack-infra | 22:16 | |
| jog0 | clarkb: fungi is correct, pypi passed but there was a regression | 22:17 |
| jog0 | clarkb: I agree both should pass, just waiting for someone to pick it up | 22:18 |
| *** lnxnut has quit IRC | 22:18 | |
| *** rakhmerov has joined #openstack-infra | 22:18 | |
| *** rcleere has quit IRC | 22:18 | |
| *** senk1 has quit IRC | 22:19 | |
| clarkb | anteaya: something/archive. it looks like there are several of them | 22:20 |
| clarkb | probably all forks of each other | 22:20 |
| anteaya | k thanks | 22:20 |
| anteaya | looks worth a look: http://forge.puppetlabs.com/smarchive/archive | 22:21 |
| anteaya | appears to be worth a look | 22:22 |
| anteaya | or look, look, look | 22:22 |
| anteaya | it is moved to here: https://github.com/gini/puppet-archive | 22:24 |
| anteaya | so would the module be called smarachive/archive still or gini/archive, I wonder? | 22:25 |
| anteaya | gini | 22:25 |
| *** rakhmerov has quit IRC | 22:25 | |
| *** senk has joined #openstack-infra | 22:26 | |
| openstackgerrit | Clark Boylan proposed a change to openstack-infra/elastic-recheck: Better end of console log indexing detection. https://review.openstack.org/53487 | 22:26 |
| *** UtahDave has quit IRC | 22:26 | |
| jog0 | clarkb: I will respin 54165 because of the H903 typo | 22:27 |
| clarkb | jog0: ok | 22:27 |
| clarkb | 53487 has been rebased on sdague's e-r changes. I still haven't quite tested it, but left a comment in the review about testing (want ot make sure I set up the functional tests properly) | 22:28 |
| *** beagles has quit IRC | 22:28 | |
| *** senk1 has joined #openstack-infra | 22:29 | |
| lifeless | mordred: so, changelog and pbr | 22:30 |
| *** senk has quit IRC | 22:30 | |
| openstackgerrit | Joe Gordon proposed a change to openstack-dev/hacking: Continue reserving H6xx namespace for calling methods https://review.openstack.org/54165 | 22:30 |
| *** jerryz has joined #openstack-infra | 22:31 | |
| *** prad_ has quit IRC | 22:33 | |
| openstackgerrit | A change was merged to openstack-infra/jenkins-job-builder: Make --delete-old more robust https://review.openstack.org/52262 | 22:38 |
| *** hashar has quit IRC | 22:38 | |
| *** datsun180b has quit IRC | 22:41 | |
| *** dims has quit IRC | 22:43 | |
| *** adalbas has quit IRC | 22:44 | |
| harlowja | btw, bumped https://pypi.python.org/pypi/gerrit-view with czuul , comments/feedback welcome | 22:45 |
| harlowja | https://github.com/harlowja/gerrit_view/#czuul (screenshot there) | 22:45 |
| clarkb | fungi: don't let me forget but we should sort out the puppet running on puppetmaster stuff this week as it may make applying keys or removing them summit time interesting | 22:49 |
| fungi | clarkb: very good point (i have a pending change i need to push for that myself tonight too) | 22:51 |
| fungi | harlowja: i did actually start playing around with it last night, but no feedback yet | 22:52 |
| harlowja | np | 22:52 |
| harlowja | its pretty basic, ha | 22:52 |
| fungi | i tend to work in traditional 80x25 terminals (yay greenscreen nostalgia), so some of it seemed a bit cramped | 22:52 |
| clarkb | fungi: yeah I need to push one too | 22:52 |
| harlowja | hmmm, ya the 25 part is tough | 22:53 |
| harlowja | only so much u can do with 25 :-P | 22:53 |
| * fungi does a lot in 25 lines x a dozen terminals on his monitor, but yeah, understood | 22:53 | |
| harlowja | actually just checked, mine is 156x28, so i guess not that much diff | 22:54 |
| harlowja | ha | 22:54 |
| *** lnxnut_ has quit IRC | 22:54 | |
| *** lnxnut has joined #openstack-infra | 22:55 | |
| sdague | clarkb: on this - https://review.openstack.org/#/c/53487/2 - that's matching success runs as well now right? that intentional? | 22:55 |
| *** dims has joined #openstack-infra | 22:55 | |
| clarkb | sdague: no, I should add a filter on build_status, thanks | 22:56 |
| sdague | yep, that should do it | 22:56 |
| clarkb | sdague: basically need something that happens at the end of every console log and the line where it copies itself works | 22:56 |
| clarkb | but I should filter on build_status | 22:56 |
| sdague | yep | 22:57 |
| sdague | heh, I should go -1 it and leave the feedback there as well :) | 22:58 |
| *** lnxnut has quit IRC | 22:59 | |
| *** DennyZhang has joined #openstack-infra | 22:59 | |
| openstackgerrit | Clark Boylan proposed a change to openstack-infra/elastic-recheck: Better end of console log indexing detection. https://review.openstack.org/53487 | 22:59 |
| clarkb | sdague: ^ that should fix it | 23:00 |
| clarkb | thanks | 23:00 |
| *** sarob_ has joined #openstack-infra | 23:02 | |
| sdague | +2 | 23:02 |
| *** sarob_ has quit IRC | 23:02 | |
| *** DennyZhang has quit IRC | 23:02 | |
| *** senk1 has quit IRC | 23:03 | |
| *** sarob_ has joined #openstack-infra | 23:03 | |
| *** sarob has quit IRC | 23:06 | |
| *** sarob_ has quit IRC | 23:07 | |
| lifeless | clarkb: your https://review.openstack.org/#/c/52689/ has inventory :) | 23:10 |
| clarkb | lifeless: thanks | 23:10 |
| *** dkliban has quit IRC | 23:11 | |
| openstackgerrit | Adrian Otto proposed a change to openstack-infra/config: Adding project Solum to StackForge https://review.openstack.org/53151 | 23:12 |
| clarkb | lifeless: fwiw the file specific comments are common around these parts | 23:13 |
| openstackgerrit | Anita Kuno proposed a change to openstack-infra/config: Create a seafile instance. https://review.openstack.org/54030 | 23:13 |
| lifeless | clarkb: -infra ? | 23:14 |
| clarkb | lifeless: yes | 23:14 |
| lifeless | clarkb: huh, well up to you guys :) | 23:14 |
| clarkb | the other thing should defnitely be fixed though | 23:15 |
| lifeless | clarkb: with the focus on one-concept changes and why-over-what, I don't think anyone else feels the need for per-file-comments. | 23:15 |
| lifeless | but shrug | 23:15 |
| lifeless | clarkb: also a test, I realised I hadn't asked for one. | 23:15 |
| clarkb | lifeless: that captures why and what | 23:15 |
| lifeless | clarkb: right, I know you did :) | 23:16 |
| lifeless | clarkb: if you haven't I would have also whined about hat | 23:16 |
| lifeless | s/hat/that/ | 23:16 |
| clarkb | also test, good point (I whipped the change up after dealing with the issue as a way to remind myself to make it so, I should add a test now) | 23:17 |
| *** ^d has quit IRC | 23:17 | |
| jerryz | clarkb: i see in the code that e-r queries elastic search based on change id and patch id. but if i use recheck to trigger a second build if the first fails due to a known bug, e-r will comment with two bugs if the second build fails due to a different known bug. | 23:21 |
| *** oubiwann has quit IRC | 23:22 | |
| jerryz | clarkb: is this by design or could build uuid be part of the query? | 23:22 |
| *** rakhmerov has joined #openstack-infra | 23:23 | |
| clarkb | jerryz: build_uuid should be part of the query | 23:23 |
| clarkb | jerryz: or log_url which includes part of the build_uuid | 23:23 |
| openstackgerrit | Anita Kuno proposed a change to openstack-infra/config: Create a seafile instance. https://review.openstack.org/54030 | 23:24 |
| reed | guys, is the sched for the DEsign Summit sessions ready? | 23:24 |
| openstackgerrit | Adrian Otto proposed a change to openstack-infra/config: Adding project Solum to StackForge https://review.openstack.org/53151 | 23:24 |
| jerryz | clarkb: def classify(self, change_number, patch_number, comment): | 23:25 |
| clarkb | reed: I think the schedule for infra is ready, not sure about the other projects | 23:25 |
| clarkb | jerryz: comment is the magic bit that includes the uuid | 23:25 |
| *** hashar has joined #openstack-infra | 23:25 | |
| anteaya | hey hot and rainy in hk next week | 23:26 |
| reed | clarkb, thanks | 23:26 |
| clarkb | woot | 23:26 |
| pleia2 | reed: I am both amused and depressed by your sched tweets today (I feel your pain) | 23:27 |
| pleia2 | reed: if you do find an ical integration mechanism, you'll be my bff | 23:27 |
| jerryz | clarkb: i don't quit understand. uuid is a part of the comment for jenkins log_url | 23:27 |
| *** rakhmerov has quit IRC | 23:27 | |
| jerryz | clarkb: but not part of the elastic-recheck comment | 23:27 |
| reed | pleia2, unfortunately I found no way without loosing the possibility to sync the actual source of data | 23:28 |
| clarkb | jerryz: if you look at the mechanism around that comment variable it is checking that the log_url left in the gerrit comment by zuul/jenkins matches the log_url in elasticsearch | 23:28 |
| pleia2 | reed: sigh | 23:28 |
| openstackgerrit | Anita Kuno proposed a change to openstack-infra/config: Create a seafile instance. https://review.openstack.org/54030 | 23:28 |
| reed | pleia2, even with how Sched suggests to operate (add the personal calendar to google), google pulls only once a day | 23:28 |
| *** cyeoh_ is now known as cyeoh | 23:29 | |
| reed | my owncloud server cannot import an .ics | 23:29 |
| pleia2 | ah | 23:29 |
| anteaya | what version of owncloud are you on, reed? | 23:29 |
| reed | I can put it in Thunderbird but then it sits there... useless, as I need it on my phone | 23:29 |
| reed | anteaya, let me check | 23:29 |
| anteaya | k | 23:29 |
| reed | anteaya, 5 something | 23:30 |
| anteaya | okay | 23:31 |
| anteaya | yeah I was on 5 something too | 23:31 |
| anteaya | 5.0.12 is latest stable | 23:31 |
| anteaya | then 6 alpha | 23:31 |
| anteaya | I was on 5.04 | 23:31 |
| reed | honestly, I am starting to think that it messes up with the calendars | 23:31 |
| anteaya | 5.0.4 | 23:31 |
| reed | hard to diagnose though as my brain may be as buggy | 23:31 |
| clarkb | lifeless: I think I may need to bug jeblair about testing that change. It looks like TestScheduler overrides zuul.merger's init mechanism | 23:31 |
| anteaya | we I'm pounding away on seafile now, we had better all like seafile | 23:32 |
| clarkb | lifeless: not sure if that should be torn out or if we need to do something more silly | 23:32 |
| pleia2 | I'll probably just end up using the sched inteface again on my tablet, hope for the best with things on my other calendar (whatever, people know I'm traveling, I can miss things :)) | 23:32 |
| clarkb | lifeless: but I definitely think we should have a test (as my discovery shows) | 23:32 |
| reed | anteaya, trying to update... it fails with no discernible error | 23:32 |
| clarkb | reed: clearly the solution is printouts | 23:33 |
| anteaya | yeah, I couldn't update owncloud from the admin page on owncloud either | 23:33 |
| clarkb | reed: I am probably going to print out a schedule just beacuse | 23:33 |
| *** dcramer_ has joined #openstack-infra | 23:33 | |
| jerryz | clarkb: this is what i see in the debug log about the query: Making a request equivalent to this: curl -XGET 'http://10.145.81.231:9200/_search?size=10' -d '{"sort": {"@timestamp": {"order": "desc"}}, "query": {"query_string": {"query": "( @message:\" 503\" OR message:\" 503\" ) AND filename:\"logs/syslog.txt\" AND syslog_program:\"proxy-server\"\n AND build_change:\"34951\" AND build_patchset:\"1\""}}}' | 23:33 |
| reed | clarkb, why didn't I think of that? :) | 23:33 |
| anteaya | never updated, no output | 23:33 |
| *** b3nt_pin has joined #openstack-infra | 23:33 | |
| reed | clarkb, unfortunately not a solution for me: I need to be able to see in *one* place all of my appointments | 23:33 |
| clarkb | jerryz: right, the matching happens in e-r (it can probably be done in elasticsearch proper but isn't today) | 23:33 |
| reed | sessions I *have to* be in and personal meetings | 23:34 |
| clarkb | reed: ah | 23:34 |
| reed | if things are not on my calendar they don't happen | 23:34 |
| clarkb | yeah if you have more than just sessions to juggle then printouts will be hard | 23:34 |
| reed | and given that walking from one side to another in HK may take as long as 15 minutes... | 23:34 |
| fungi | reed: clarkb: my solution is a text file with a list of where i intend to go when ;) | 23:34 |
| reed | I need to have a very *solid* system or I'll miss lots of meetings | 23:35 |
| fungi | granted, my text file doesn't beep at me to let me know when something is coming up | 23:35 |
| clarkb | fungi: you'll be surprised at how hard it is to edit a simple txt file on chromeos. It is the biggest flaw so far | 23:35 |
| reed | fungi, a paper agenda would work too... but it doesn't ring to remind me to get the hell out of the meeting and run to a session | 23:35 |
| reed | or viceversa :) | 23:35 |
| reed | i don't like to trust my brain | 23:36 |
| clarkb | like seriously how hard can it be to expose the vi(m) that must be on the system | 23:36 |
| reed | clarkb, are you taking a disposable chrome os? | 23:36 |
| clarkb | reed: yup | 23:37 |
| reed | damn, you paranoid | 23:37 |
| fungi | that would risk exposing ways to risk rooting and installing of non-googOS[TM] | 23:37 |
| reed | now you make me feel like I'll be out there like naked in the snow | 23:37 |
| clarkb | well "disposable" I am not super paranoid about it but need a light machine with good battery life that isn't completely insecure | 23:37 |
| fungi | clarkb: i'm bringing tamper-evident evidence bags, evidence-tape and foil stickers, in case you want to partake (i'll have plenty to go around) | 23:38 |
| reed | clarkb, did you have to sign your soul to google to use it? | 23:38 |
| clarkb | reed: yes | 23:38 |
| *** SergeyLukjanov is now known as _SergeyLukjanov | 23:38 | |
| reed | clarkb, good :) | 23:38 |
| * reed feels better now | 23:38 | |
| clarkb | none of my other machiens have secure boot | 23:38 |
| fungi | good for when you don't trust the hotel room safe and want to leave your disposable hardware while you dine | 23:38 |
| clarkb | which is sort of necessary to trust disk encryption | 23:38 |
| reed | fungi, geez | 23:38 |
| *** _SergeyLukjanov has quit IRC | 23:38 | |
| reed | they can tamper your laptops once they stole it :) | 23:39 |
| clarkb | fungi: nice, I may have to partake | 23:39 |
| *** dcramer_ has quit IRC | 23:39 | |
| fungi | reed: if they steal it, good riddance. i just want to know if they try to put it back after | 23:39 |
| anteaya | my syntax is okay with puppet lint and that is all I can do tonight, I still haven't run it, if anyone has a boatload of spare time and can't wait to give me the benefit of their time, I welcome your comments or I'll just keep on it tomorrow | 23:39 |
| reed | :) | 23:39 |
| anteaya | https://review.openstack.org/#/c/54030/18 | 23:39 |
| clarkb | reed: it has a decent IPS display which makes me want an IPS display on everything now | 23:40 |
| reed | seafile seems interesting... does it store files in swift or just git? | 23:40 |
| clarkb | and ~6 hour battery life which is about 4 hours better than the alternatives | 23:40 |
| anteaya | storage on the my beta is mysql | 23:40 |
| *** mrodden1 has quit IRC | 23:40 | |
| reed | wikipedia:IPS_display | 23:40 |
| anteaya | we are going for trove once it is up | 23:40 |
| reed | anteaya, it stores files in mysql? | 23:41 |
| anteaya | it doesn't use swift | 23:41 |
| anteaya | mysql was the only db I set up for the beta installs | 23:41 |
| *** jergerber has quit IRC | 23:42 | |
| reed | i became a big fan of sshuttle over the weekend: looks like it works on my wife's mac os too | 23:42 |
| reed | so very simple ssh sessions in different datacenters will give me VPN tunnels easily | 23:43 |
| *** nati_uen_ has joined #openstack-infra | 23:43 | |
| *** jhesketh_ has joined #openstack-infra | 23:43 | |
| reed | on different ports, too... very easy, no need to mess up with openvpn or ipsec | 23:43 |
| *** jhesketh has quit IRC | 23:45 | |
| clarkb | reed: fwiw I would not use a chromebook for daily use | 23:45 |
| *** b3nt_pin is now known as beagles | 23:45 | |
| clarkb | but as a travel device it seems decent | 23:45 |
| *** nati_ueno has quit IRC | 23:46 | |
| *** beagles has quit IRC | 23:47 | |
| *** yamahata has quit IRC | 23:47 | |
| *** denis_makogon has quit IRC | 23:47 | |
| fungi | i like traveling with a cheap netbook. bonus is i'm not stuck with teh googolz | 23:47 |
| *** b3nt_pin has joined #openstack-infra | 23:48 | |
| *** b3nt_pin is now known as beagles | 23:48 | |
| *** SergeyLukjanov has joined #openstack-infra | 23:48 | |
| *** weshay has joined #openstack-infra | 23:51 | |
| *** dcramer_ has joined #openstack-infra | 23:52 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!