14:00:55 #startmeeting fedora-qadevel 14:00:55 Meeting started Mon Jul 11 14:00:55 2016 UTC. The chair is tflink. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:00:55 Useful Commands: #action #agreed #halp #info #idea #link #topic. 14:00:55 The meeting name has been set to 'fedora-qadevel' 14:00:56 #topic Roll Call 14:01:03 * kparal is here 14:01:06 * mkrizek is here 14:01:41 * lbrabec is here 14:02:30 welcome, welcome 14:02:37 * omiday is here 14:03:08 let's get this party started, shall we? 14:03:09 jskladan is on PTO, jsedlak said he would not be here 14:03:41 ok, i knew that not everyone would be here 14:03:50 #topic Announcements and Information 14:04:05 s/everyone/all of the regulars/ 14:04:10 #info python-resultsdb_api and libtaskotron packages are in Fedora - mkrizek 14:04:10 #info resultsdb, resultsdb_frontend, execdb and taskotron-trigger are up for review, waiting for reviewers - mkrizek 14:04:11 #info qa namespace renamed to dist - mkrizek 14:04:11 #info task-rpmgrill is now reviewed, ready to be deployed - kparal 14:04:11 #info errors are now correctly raised when files can't be downloaded using libtaskotron - kparal 14:04:13 #info improved docker testing docs - kparal 14:04:31 any questions/comments? 14:04:52 not here 14:05:39 nope 14:06:55 one question I do have for packagers: would anyone be willing to do review swaps if we have trouble finding folks who are willing to just do the reviews? 14:07:29 sure 14:08:06 thanks 14:08:43 moving on 14:09:06 #topic Docker Testing 14:09:23 This is the specific deliverable that we've committed to for F25 14:09:49 * tflink thinks that it's mostly ready, wanted to double check to make sure he isn't forgetting something 14:11:38 it's hard to image how people will be using it in the real world 14:11:51 but I think the current docs cover the basic use case 14:11:52 agreed 14:12:03 on the hard to imagine how folks will use it 14:12:10 we might need to do some improvement wrt reporting 14:12:22 i suspect that there will be some enhancement requests 14:12:55 also, we will need support in trigger, etc. but the ball is not in our court for that 14:13:17 kparal: do you have any specific ideas on what might be needed for reporting? 14:13:54 for example, the taskotron_result tool can now generate a series of results, but cannot generate an overall result 14:14:11 which might be handy for bash script which have no specific requirements, just any fail -> overall fail 14:14:38 that sounds like something which wouldn't be too hard to do 14:14:42 nope 14:15:25 but again, not sure we should implement before someone actually wants to use it that way 14:15:49 #action tflink to follow up with maxamillion on the fedmsgs coming out of OSBS to figure out what we should listen to for triggering jobs 14:16:03 #chair kparal mkrizek lbrabec 14:16:03 Current chairs: kparal lbrabec mkrizek tflink 14:16:13 osbs? 14:16:14 * tflink always forgets to chair other folks 14:16:24 the system that is producing the layered docker images 14:16:49 open shift build service, I think 14:17:00 but I'm not terribly confident about what OSBS stands for :) 14:17:20 the name is too similar to suse's OBS :) 14:17:22 ok, thanks 14:17:36 anyone can google but I'll save folks typing cycles: https://github.com/projectatomic/osbs-client 14:17:52 i was remembering correctly 14:18:16 omiday: thanks 14:18:29 doing a quick search probably should have been obvious 14:18:40 * tflink blames monday morning with insufficient coffee :) 14:18:45 tflink: no problem, you are welcome 14:19:15 * omiday can't help much with anything else thus far anyways 14:20:17 it sounds like we're pretty much on track, though 14:20:37 * tflink will verify in more detail outside of the meeting 14:21:08 #topic Disk Space for Logs/Artifacts 14:21:33 we're hitting alarms on disk usage for taskotron01.qa again 14:22:05 the cron job will clear the alarm for a matter of hours, so we're bouncing off the 80% usage warning level 14:22:42 * kparal is getting the alarms through fmn 14:22:43 I'm not sure that decreasing the amount of time we keep logs/artifacts around again is going to be the best solution 14:23:53 but increasing the disk size would involve rebuilding 14:24:07 are logs compressed? 14:24:20 no, but they probably should be 14:24:48 doesn't ext4 have now optional compression of selected directories? 14:24:49 since we could configure apache to handle gzipped files transparently, IIRC 14:25:01 it would appear uncompressed for the world 14:25:17 we can start with that to see how much it helps 14:25:30 it's all plaintext, I suspect it would help a lot 14:25:46 80% :) 14:25:49 do we have ext4 or xfs? I forget 14:25:55 ext4, I think 14:26:11 new fs features make me a little nervous, though 14:26:15 what is the box using to rotate logs? logrotate does compress 14:26:32 * tflink would prefer to handle the uncompression at the application level 14:26:52 we're not really rotating logs, at least not how I'd describe rotating 14:27:12 I'd use fs compression just because you don't need to handle it at the app level :) 14:27:18 we have a cron job that runs through that part of the file system and deletes anything which is older than a setpoint 14:27:39 I'm not sure it works as I say, we can search for that. but I wouldn't worry about ext4 features, once they're claimed stable 14:28:07 yeah, I'm not saying that we shouldn't - just voicing a concern 14:28:20 * tflink remembers testing fancy storage features in a previous life 14:29:03 what was the reason behind choosing the custom cron job rather than logrotate? 14:29:09 mkrizek: do you have the spare cycles to look into this? 14:29:18 tflink: yep, will do 14:29:27 omiday: didn't realize that logrotate could handle that kind of job 14:29:52 IIRC, the cronjob is using tmpwatch 14:29:52 I'm not familiar with the application so I may be off 14:30:07 but writing a logrotate config for a set of logs is trivial 14:30:31 doesn't logrotate require rather specific paths? 14:30:31 tflink: it uses find now 14:30:48 yes it does - wildcards too 14:31:09 oh yeah, forgot about that. did we ever figure out why tmpwatch wasn't working? 14:31:38 tflink: nope :/ 14:31:46 with the variability in paths for the logs/artifacts, I'd be worried about writing a config file that's expecting too much regularity 14:31:57 ok, just curious 14:32:11 umm this is ithe releng meeting slot now 14:32:22 it is? 14:32:25 yes 14:32:30 i thought you all met in an hour 14:32:42 we moved three weeks ago 14:32:51 and we haven't met in about 3 weeks 14:33:14 there has never been a meeting here before ours for hours 14:33:22 we've been meeting for years 14:33:27 :) 14:33:38 notat this time 14:33:42 in this room 14:33:47 yes, at this time in this room 14:33:50 dgilmore: we did :) 14:34:02 it's even in fedocal 14:34:11 +-1 hour due to DST 14:34:41 I moved our meeting in fedocal and it gave no conflicts 14:34:54 I guess we will move this week 14:35:01 it shows both of us in here from 14:30 to 15:00 14:35:38 I think fedocal doesn't inform about conflicts 14:35:43 it's just a calendar 14:37:44 getting back to logs, 14:37:52 its a calendars job to inform of conflicts 14:38:31 anyway sorry 14:38:46 no worries, it happens 14:39:04 if logrotate can do things that I'm not aware of, I'd be game for switching 14:39:27 if it can do the compression and removal of old dirs/logs, I mean 14:40:01 anything else on this topic? 14:40:31 logrotate won't do deep searching but a 'find' cron can compress then 14:41:03 although I see where kparal is coming from now with fs compression 14:42:03 we could also do the whole fs and skip the cronjobs, if we go that direction 14:42:18 the logs are on an LVM by themselves 14:43:45 anyhow, moving on 14:43:54 #topic Tasking 14:44:06 is anyone looking for tasks? 14:44:24 * mkrizek has stuff to do for now 14:44:43 I'd like to start working on something yes 14:44:43 * kparal still has some past patches to look through 14:45:04 * tflink is also really behind on reviews :( 14:45:35 omiday: do you have the time to chat in #fedora-qa after the meeting ends? 14:45:51 my day job starts at 9 Mountain 14:46:02 any time after 5pm Mountain or weekends 14:47:03 omiday: I'm just not sure what exactly you'd be looking to do and/or what your skills are 14:47:42 omiday: can you ping me some time when you're free, then? I might be a bit slower to respond after 5pm mountain but should still be around-ish 14:48:39 I will - unfortunately my employer doesn't support my fedora volunteering duriing work hours so that's all I can do 14:49:17 no worries, I've been in that situation before - help is appreciated, though :) 14:49:21 I do work from home so getting home will be fast ;) 14:50:06 if there's nothing else on this topic, I'll move on to open floor 14:50:38 #topic Open Floor 14:50:47 Any other topics/items that folks wanted to bring up? 14:51:33 otherwise, I'll set the fuse 14:51:57 nothing here 14:52:08 nothing from me either 14:52:46 ok, thanks for coming, everyone 14:53:02 * tflink will send out minutes shortly 14:53:06 #endmeeting