09:00:45 #startmeeting 09:00:45 Meeting started Wed Aug 6 09:00:45 2014 UTC. The chair is flock-t343. Information about MeetBot at http://wiki.debian.org/MeetBot. 09:00:45 Useful Commands: #action #agreed #halp #info #idea #link #topic. 09:01:04 #meetingname Taskotron and me 09:01:04 The meeting name has been set to 'taskotron_and_me' 09:01:14 #meetingtopic Taskotron and Me 09:03:23 #topic defnitions 09:03:55 Check: confirmation; exploration; doesn't necessarily return PASS/FAIL; you decide how to respond to result 09:04:21 Test: process of exploration, discover and learning; 09:04:42 Checking is subset of checking, so while all checks are tests, not all tests are checks; tests cannot e automated, tasks can 09:04:49 #topic history 09:05:26 why do we have so many things which would be beneficial to run at scale, but have no way to actually do; 09:05:36 that's what taskotron is trying to address 09:05:57 What is Taskotron; automate tasks on deterministic schedule 09:06:25 similiar to autoqa but ... better! handling/scheduling execution but not the details of the execution 09:07:32 ... kick off tasks based on fedmsgs; or manually in local case; prepare envioronment for task execution; delegate task execution to specialzed code (or runners); report task results to appropriate placess 09:08:12 difference from autoqa? 09:08:30 role of main devs; fedorans can maintain their own tasks; don't need us to do it 09:08:51 ie, scales better; devel's focus on framework; people maintain their own tasks 09:09:15 less confusing naming; libtaskotron is the SW; taskotron is the system 09:10:11 more flexible; eg, disposable test clients; letting users write their own tasks 09:11:03 Q: disposal test clients examples? long living virtual machines are doing execution; eg, if you want to do destructive testing; taskotron helps spin up fresh environment, run, collect logs, take down, etc 09:12:27 Design principals 09:12:37 minimize coupling; results db, for example 09:13:18 pretty much everything is Python; REST and JSON b/w components 09:13:34 tasks are python dependent; only core code 09:13:55 tasks can be isolated in a sigle git repo 09:14:37 Q: what about tests which are only valid for certain package 09:14:55 A: not available yet; in planning though; one test repo per package, for example 09:15:22 How does it work? 09:16:14 locally; works out of the box. You should NEVER need to setup a full taskotron stack to run a local task 09:17:49 task output should (currently) be in TAP 09:18:26 in Production; reporting enabled; pulling from remote git repo instead of locally 09:21:31 Q: what is TAP 09:21:51 A: Test Anything Protocol (from Perl); text format designed for reporting results 09:22:21 TAP 13 is better defined; one of the (if not the only) plain text standard for reporting results 09:22:42 search TAP perl 13 to find it 09:23:30 Q: will dist-git be a backend? 09:23:45 A: good idea; but not currently a plan 09:25:39 Q: can we run tasks on multiple packages; eg, integration tests 09:26:41 A: could be possible but not possible now; taskotron is tied per build (mostly) 09:26:59 Q: image builds in koji support? or just pkg builds 09:27:03 A: pkg builds only 09:27:21 eg cloud images; yea, great question and we're having that discussion 09:28:28 scheduling; simple; triggered from fedmsgs 09:42:10 sorry; internet died 09:42:12 back 09:42:20 in Q&A now 09:42:34 A: when will disposable client support be available 09:42:40 s/A/Q/ 09:43:05 A: still determining whether to use OpenStack or libvirt; this should make some progress this week, actually 13:00:00 Fedora future devices starts in 1m 13:00:59 flock-t343: Error: Can't start another meeting, one is in progress. 13:01:05 #endmeeting