| Zara | hm, webclient hasn't built since 14th, wonder if it's the worker problem again. | 00:04 |
|---|---|---|
| Zara | night night | 00:04 |
| anteaya | night | 00:07 |
| *** anteaya has quit IRC | 03:39 | |
| *** mrmartin has joined #storyboard | 04:59 | |
| *** mrmartin has quit IRC | 04:59 | |
| *** jmccrory has quit IRC | 07:22 | |
| *** jmccrory has joined #storyboard | 07:24 | |
| SotK | eek, cacti suggests that may be the case | 07:32 |
| SotK | I guess I'll be spending today trying to reproduce that again :) | 07:32 |
| *** bethwhite has joined #storyboard | 08:06 | |
| Zara | :D | 09:22 |
| SotK | I had a better idea that involved thinking about what actually happens instead of bashing my head against it until it breaks | 09:22 |
| Zara | hah, I always guess the wrong thing, so prefer head-bashing | 09:26 |
| pedroalvarez | SotK: have you figured out why? | 09:26 |
| Zara | (welp, prefer asking around first) | 09:26 |
| pedroalvarez | SotK: my guess was that the sysvinit daemons were sh**e | 09:27 |
| SotK | that was my old guess, but I can't see anything in https://git.openstack.org/cgit/openstack-infra/puppet-storyboard/tree/templates/storyboard-workers.sh.erb which would leave things alive | 09:29 |
| SotK | now I suspect that storyboard-worker-daemon sometimes doesn't kill its child processes (I'm pretty sure I've seen it happen when running manually too) | 09:30 |
| * SotK is investigating that thought | 09:30 | |
| Zara | I got an error about not killing something right last time I ctrl+c'd it. wonder if that was related or normal | 09:31 |
| pedroalvarez | SotK: here I can see that the script ensures that they are running: https://git.openstack.org/cgit/openstack-infra/puppet-storyboard/tree/manifests/workers.pp | 09:31 |
| pedroalvarez | warning: if they are running it doesn't restart them | 09:32 |
| pedroalvarez | SotK: so... does your ubuntu setup report that the services are running? | 09:32 |
| pedroalvarez | maybe puppet fails to see that they are running, and then, it spawns more and more | 09:32 |
| SotK | hmm, I thought I saw it trying to restart them last time I looked at this | 09:35 |
| * SotK goes to look at that too | 09:35 | |
| pedroalvarez | IIRC I'm using upstart on my setup, it was more reliable | 09:37 |
| Zara | this is what happens when I ctrl+c, if interesting to anyone: http://paste.openstack.org/show/494755/ | 09:37 |
| pedroalvarez | but looking forward to move to ubuntu 16.04 with systemd :P | 09:37 |
| * SotK too | 09:38 | |
| Zara | I'm wary of merging more api things until the webclient catches up | 10:34 |
| Zara | well, I can test them with the current built version + master, but that'll take longer | 10:34 |
| Zara | guess it depends on how long before we can kill processes | 10:35 |
| Zara | in the meantime... | 10:38 |
| Zara | does anyone who commits code have an opinion on https://review.openstack.org/#/c/304539/ ? (qualifying that as, in this case, I think this one only affects people in the process of writing code on the webclient) | 10:39 |
| Zara | it's up to the crowd | 10:40 |
| Zara | pedroalvarez, betherly, anyone-I've-missed ^ | 10:40 |
| pedroalvarez | I don't really understand any of that.. ::/ | 10:41 |
| pedroalvarez | ouch, these new eyes hurt | 10:41 |
| Zara | :0 | 10:41 |
| pedroalvarez | I mean, I only modify something, and then send the patch.. | 10:41 |
| Zara | ah, okay, so atm storyboard tests your webclient code everytime you save a file | 10:42 |
| pedroalvarez | "everytime you save a file" ? | 10:42 |
| Zara | and lets you know if you have pep8 problems, undefined variables etc | 10:42 |
| Zara | this is annoying if you know your code isn't ready yet | 10:42 |
| pedroalvarez | ahm.. | 10:43 |
| Zara | but it's nice if you don't want to do it manually all the time | 10:43 |
| Zara | I like to be poked to fix things as I go along, others find it more annoying than me | 10:43 |
| pedroalvarez | so, the patch removes this from running automatically? | 10:43 |
| Zara | yeah | 10:43 |
| pedroalvarez | ah | 10:43 |
| pedroalvarez | I have no opinions, never used it, and I don't think I ever will | 10:44 |
| Zara | ahahaha | 10:44 |
| Zara | okay | 10:44 |
| pedroalvarez | I think I have good pep8 skillzzzzzzz | 10:44 |
| pedroalvarez | but sounds useful to have, tbh | 10:45 |
| Zara | :) I like it, personally, but I figure affected people should vote on it, since I can still run the tests manually if most people are bothered by it | 10:45 |
| Zara | I have been informed it's not pep8, since that's python-specific, so just style | 10:46 |
| Zara | thanks, secret-python-info-daemon | 10:47 |
| Zara | I have 4 hairs on my chin! I knew, if I waited long enough, I'd grow a UNIX beard. :D | 10:58 |
| openstackgerrit | Zara proposed openstack-infra/storyboard-webclient: Make it clear when user reaches end of events timeline https://review.openstack.org/305483 | 11:45 |
| * Zara now looks into why she made some table columns weird lengths | 11:58 | |
| Zara | (well, I imagine it was 'it looks better that way' but I'm wondering why sensible lengths didn't work.) | 11:58 |
| Zara | hm, so the other table rows are somehow 7 atm | 12:06 |
| Zara | maybe the 'delete' button is stuck on the end somewhere | 12:07 |
| Zara | ohh, it's proooobably that the task notes button added an extra column | 12:08 |
| Zara | which gets applied in the inline tasks bit | 12:08 |
| Zara | though I get that bit as still totalling 12, so huh | 12:09 |
| Zara | unless it was 11 before on purpose | 12:09 |
| * SotK gets confused | 12:10 | |
| Zara | ahh, sorry, i was talking about https://review.openstack.org/#/c/294221/2 | 12:10 |
| Zara | I had a spare moment to look back at my own patches | 12:10 |
| Zara | they've been pretty neglected | 12:11 |
| Zara | so yeah, bootstrap grid, 12 columns, my columns seemed to be odd values | 12:11 |
| Zara | but it looks weird when they seem more sensible, so now I'm trying to find out what's going on | 12:11 |
| Zara | and thinking out loud because I do that \o/ | 12:11 |
| SotK | bootstrap columns != colspan columns | 12:12 |
| Zara | oh, welp. | 12:13 |
| * Zara = html pro | 12:13 | |
| SotK | colspan references the number of actual columns in the table, which is the highest number of <td> elements in a row | 12:13 |
| Zara | (makes sense since that table was part of a bigger page... think I thought of it as grids within grids... fun working out what odd leaps my brain's done without telling me.) | 12:14 |
| SotK | (<td> implies <td colspan="1">) | 12:15 |
| SotK | each row's sum of colspans should be the same, or mess ensues, as I suspect you have discovered | 12:15 |
| * SotK finishes his understanding of colspan | 12:15 | |
| Zara | thanks | 12:16 |
| Zara | guess I want it the same width as task_list_item, then | 12:18 |
| SotK | yeahm which appears to be 7 | 12:19 |
| *** alexismonville has joined #storyboard | 12:48 | |
| * SotK reproduces the workers bug | 12:49 | |
| alexismonville | hi :) I didn't found on the schedule if there was storyboard meetings planned for the next summit? could someone help me to find the information? | 12:49 |
| openstackgerrit | Merged openstack-infra/storyboard-webclient: Make it clear when user reaches end of events timeline https://review.openstack.org/305483 | 12:51 |
| Zara | alexismonville: hi! at the moment, storyboard's going to be discussed at the community task tracking session (I'll find the link, hang on) | 12:54 |
| Zara | other than that, we don't have a room booked, because we're a smallish team, but there are plans to find a corner with wifi and hack on things. and if more people are interested in dicusssing things then I think we can organise an informal gathering of some kind... | 12:56 |
| * SotK agrees | 12:56 | |
| SotK | (with that thought) | 12:56 |
| Zara | my personal preference is to go to Barton Springs pool, but I'm flexible on that (;_;) | 12:57 |
| * SotK wonders if the water has wifi xD | 12:57 | |
| Zara | ah, link is here: https://www.openstack.org/summit/austin-2016/summit-schedule/events/9171?goback=1 | 13:00 |
| *** alexismonville has quit IRC | 13:17 | |
| *** alexismonville has joined #storyboard | 13:17 | |
| pedroalvarez | SotK: any luck with the workers? | 13:19 |
| SotK | not really | 13:19 |
| SotK | I can reproduce by doing `sudo service storyboard-workers restart` | 13:19 |
| SotK | repeatedly, since it doesn't fail everytime | 13:19 |
| pedroalvarez | does `service stroyboard-wrorkers status` report they are running? | 13:20 |
| Zara | at least it's reproducible, I guess :/ | 13:20 |
| Zara | worst-case, we kill all the workers manually again. | 13:21 |
| SotK | pedroalvarez: it reports "not supported" | 13:21 |
| SotK | I tried adding `--retry 1` to the start-stop-daemon stop line, that made it successfully kill the parent process, but still left the 5 children hanging around | 13:22 |
| pedroalvarez | heh | 13:22 |
| pedroalvarez | ok, so at least we can reproduce it | 13:22 |
| pedroalvarez | I'm not sure about upstart being better, but it might handle all this better | 13:23 |
| * SotK is pining for systemd, which he actually has some understanding of :) | 13:24 | |
| persia | There's the infra session on Xenial: maybe storyboard should be an early candidate for migration? | 13:27 |
| Zara | it could be interesting, especially as I think our production is on precise | 13:28 |
| Zara | but our tests are on trusty | 13:29 |
| persia | Spinning up a dev instance might me a reasonable was to at least move to trusty. | 13:32 |
| * pedroalvarez uses trusty, but not with those puppet scripts | 13:33 | |
| Zara | agree that spinning up a dev instance sounds like a good way | 13:34 |
| Zara | (oh yeah, while researching texas wildlife (just checking I wouldn't get eaten, and no, this wasn't on company time), I ended up learning about varieties of armadillos more generally, which led to discovering that this exists and is not sushi: https://www.google.co.uk/search?q=pink+fairy+armadillo&source=lnms&tbm=isch&sa=X&ved=0ahUKEwjq4eGMqp3MAhXJWxoKHaT2CN0Q_AUIBygB&biw=1366&bih=604 . and it doesn't eve | 13:35 |
| Zara | n live in texas, but thought people here might be interested anyway.) | 13:35 |
| SotK | Zara: are we likely to be eaten? | 13:35 |
| Zara | SotK: no. though apparently there's the odd mountain lion in a nearbyish park | 13:35 |
| Zara | and a fair amount of snakes. but we probably won't be there, unless we get very lost | 13:36 |
| Zara | http://www.fws.gov/uploadedFiles/BCNWR_Mammals_200809_A.pdf is the mammals list for that park | 13:37 |
| Zara | and http://www.fws.gov/uploadedFiles/BCNWR_Herps_200809_A.pdf for reptiles. the snakes at the end are the dangerous ones | 13:37 |
| Zara | according to the googles | 13:37 |
| Zara | assuming the wildlife isn't intersted in openstack, it'll probably stay in the park, but be on your guard. | 13:39 |
| SotK | ok, so it seems that on unsuccessful restart attempts, the stop process gets stuck here: https://git.openstack.org/cgit/openstack-infra/storyboard/tree/storyboard/plugin/event_worker.py#n156 | 13:56 |
| SotK | so neither the parent process or any of the children die | 13:56 |
| SotK | (after receiving SIGTERM) | 13:57 |
| SotK | then new workers are started (since the SIGTERM was sent successfully I guess?) | 13:57 |
| * SotK goes to refresh his brain on threading in python | 13:59 | |
| Zara | nice work :) | 14:00 |
| Zara | hm, seems task notes don't show newlines properly for logged out users | 14:09 |
| Zara | prrrrrobably because they're not rendering markdown yet | 14:09 |
| Zara | (that's for story view, idk about board view) | 14:09 |
| Zara | oh hey, storyboard meeting in 45 mins | 14:14 |
| Zara | ugh, anyway, back from distraction, I think the lengths should be 12 to be equal, but I'm not sure they need to use the total columns? the 'this story has no tasks' row only seems to use 6. | 14:27 |
| Zara | I am too warm for this | 14:28 |
| Zara | it's only april ;_; I think I can only work in march and september, outside of that it's either too dark or too hot. | 14:29 |
| Zara | argh colspan vs col-xs-foo | 14:30 |
| Zara | that was my confusion | 14:30 |
| Zara | dsfdjslkfdjs, anyway. | 14:38 |
| persia | Ah, yes, totally different. As far as I can tell, the table in 294221 is entirely innocent of bootstrap layout | 14:38 |
| * Zara changes everything to 7 | 14:38 | |
| persia | Why seven? | 14:39 |
| Zara | because that's the number of <td>s in the biggest row | 14:39 |
| persia | Master has a head of 3+3, followed by body entries of 6. | 14:39 |
| persia | Which row has 7 in master? | 14:39 |
| * persia only saw 7 in the candidate | 14:39 | |
| Zara | (task_list_item) | 14:39 |
| Zara | it has 7 <td>s, where colspan=1 is implied | 14:40 |
| Zara | and it's called inline as a row | 14:40 |
| Zara | (idk if you should use the word 'call' for that, but I don't know what's better) | 14:40 |
| Zara | so the bit starting at line 302 | 14:41 |
| persia | I usually use "include" for that, but yeah, I see now. | 14:42 |
| persia | So the current header is broken anyway :) | 14:42 |
| Zara | yup, probably (definitely) my fault when I added task notes | 14:42 |
| Zara | since it only clicked today | 14:43 |
| Zara | didn't realise that col-foo and colspan were totally different | 14:43 |
| Zara | yayyyyy~ | 14:43 |
| persia | git blame blames someone else, but yes, around task notes. | 14:43 |
| * SotK apologises if it was him | 14:44 | |
| persia | Right. Initially broken with change Ib81fc44e84424a5d3a281b7230217d8a8c017c1f, and then modified (and left broken) in I56b2fa4489424b81385a45a9c069bdd33a2c9203 | 14:44 |
| * persia needs to figure out how to translate Change-Id: entries into gerrit URLs trivially | 14:45 | |
| Zara | ugh, let me know if you find out | 14:45 |
| Zara | I feel like I've done it before but maybe it was a manual thing | 14:45 |
| Zara | ah, searching it does work | 14:46 |
| Zara | I come up as the owner :P | 14:46 |
| persia | https://review.openstack.org/#q,Ib81fc44e84424a5d3a281b7230217d8a8c017c1f,n,z | 14:46 |
| persia | https://review.openstack.org/#q,I56b2fa4489424b81385a45a9c069bdd33a2c9203,n,z | 14:46 |
| openstackgerrit | Zara proposed openstack-infra/storyboard-webclient: Make it easier to find 'add task' button https://review.openstack.org/294221 | 14:46 |
| persia | Anyway, despite my review being entirely wrong, I'm glad we found out why :) | 14:47 |
| Zara | hehe | 14:47 |
| Zara | yeah, I've learnt some more html stuff that'll be useful | 14:47 |
| Zara | 6 minutes | 14:54 |
| Zara | 2 minutes | 14:58 |
| Zara | hm, task templates should probably put old data before new, when reporting a change | 14:58 |
| Zara | tbf those templates are generally terrible | 14:58 |
| Zara | (for email notifications for task changes) | 14:59 |
| Zara | (so I can say that because I made the things) | 14:59 |
| Zara | meeting time! | 15:00 |
| persia | It is currently 10am in Austin | 15:03 |
| Zara | persia is the talking clock :0 | 16:00 |
| persia | Of course, it's 11:00 now :) | 16:01 |
| Zara | please update us at 12:00 | 16:01 |
| *** bethwhite has quit IRC | 16:03 | |
| persia | In terms of update process: I expect the webclient will be redeployed to production as part of the post-merge job currently queued. Will the API server automatically update somehow, or does that need a change pushed as well? | 16:14 |
| SotK | I think post-merge jobs just cause `puppet apply` to run on storyboard.o.o | 16:15 |
| SotK | so I'd expect it to update also | 16:15 |
| * SotK may be mistaken | 16:15 | |
| *** fay_ has quit IRC | 16:17 | |
| persia | The post-merge for storyboard-webclient will update storyboard as a side effect? | 16:20 |
| SotK | maybe, I can't remember or find where I got that impression now | 16:29 |
| persia | https://review.openstack.org/gitweb?p=openstack-infra/storyboard-webclient.git;a=commitdiff;h=fae86518e14a6f86f6e73985d52b1234c1666ba6 is in post now, so I presume that we'll know for sure when that lands | 16:32 |
| persia | (but some jobs have been in post for >50 hours, so it might be a bit) | 16:32 |
| openstackgerrit | Adam Coldrick proposed openstack-infra/storyboard: Don't try to cancel the timer thread if it isn't alive https://review.openstack.org/308471 | 16:47 |
| SotK | that seems to stop `sudo service storyboard-workers restart` leaving things alive | 16:49 |
| persia | Why does that work? I would expect that cancel() didn't work because the task was not yet start()ed (because schedule). The docs say cancel() only works when the timer is in "waiting" (so not yet start()ed, which I would think would mean not is_alive()). Am I missing something? | 16:54 |
| SotK | my understanding is that "waiting" is the time between start() and the provided time elapsing | 16:56 |
| SotK | is_alive() is True from when start() is called to when the action has completed | 16:57 |
| SotK | s/action/provided function/ | 16:58 |
| SotK | as for why it works, I'm not fully sure, because I couldn't determine why cancel() was hanging | 16:58 |
| SotK | I suspect the PerpetualTimer class is broken in more ways than this, since the Timer object always appears to be in the "stopped" state in my experimentations | 16:59 |
| persia | Perhaps, if it is not is_alive(), can exception is raised, which is caught elsewhere, and so it does it anyway. | 16:59 |
| persia | s/can/an/ | 16:59 |
| persia | Well, nothing seems wrong with the code, except that the documentation doesn't express the concepts very well, so I'm not confident why it needs the extra conditional. | 17:00 |
| persia | I suppose once this lands, it needs another cycle of "kill all the workers, restart the service"? | 17:00 |
| SotK | probably, unless the update after landing this doesn't trigger the error | 17:01 |
| *** yarkot has joined #storyboard | 17:32 | |
| openstackgerrit | Adam Coldrick proposed openstack-infra/storyboard: Stop overriding the signal module in terminate() https://review.openstack.org/308492 | 17:34 |
| persia | Oh, heh, that looks more sensible :) | 17:57 |
| persia | Err, except now there seems to be an unused variable. | 17:59 |
| * persia will wait for an explanation before having a more detailed opinion. | 17:59 | |
| *** anteaya has joined #storyboard | 18:57 | |
| SotK | neither of the parameters are used | 18:57 |
| SotK | I don't think this was actually contributing to the bug (fixing it didn't fix the bug) but it makes the traceback Zara pasted this morning less horrible | 18:57 |
| anteaya | Zara: I added some suggestions: https://review.openstack.org/#/c/294221/3 | 18:58 |
| anteaya | Zara: I love a thesaurus | 18:58 |
| persia | SotK: Ah, and we need to define sig and frame to meet some API convention | 18:58 |
| persia | ? | 18:58 |
| anteaya | Zara: can they create a tag with that button? would Create Tag be a reasonable button title? | 18:59 |
| anteaya | or would that confuse people who want to use an existing tag | 19:00 |
| persia | To me, "create tag" implies a new, different, tag being added to the system, and may become confusing if there is ever support for "official tags", as is found in some other tooling (e.g. LP) | 19:00 |
| SotK | persia: it needs to take two arguments, yes (see https://docs.python.org/2/library/signal.html#signal.signal) | 19:01 |
| * SotK gets the same implication from "create tag" as persia does | 19:01 | |
| persia | "affix" and "attach" seem right, but hearken to traditional semantics, which some folk only familiar with the "tag cloud" or "hashtag" concepts may find alien. | 19:01 |
| persia | Aha! Thanks for the docs pointer. | 19:03 |
| anteaya | am offline until Friday | 19:06 |
| *** anteaya has quit IRC | 19:07 | |
| *** mrmartin has joined #storyboard | 19:52 | |
| *** mrmartin has quit IRC | 21:07 | |
| *** b3rnard0 is now known as b3rnard0_away | 22:06 | |
| *** krotscheck is now known as krotscheck_dcm | 22:57 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!