| *** cyndis has quit IRC | 02:30 | |
| *** lachlanmackenzie has quit IRC | 02:36 | |
| *** lachlanmackenzie has joined #baserock | 02:36 | |
| *** gtristan has joined #baserock | 05:37 | |
| *** fay has joined #baserock | 07:21 | |
| *** fay is now known as faybrocklebank | 07:21 | |
| *** cyndis has joined #baserock | 08:10 | |
| *** ctbruce has joined #baserock | 08:28 | |
| *** faybrocklebank has quit IRC | 08:47 | |
| paulsherwood | join #grimoire | 08:58 |
|---|---|---|
| paulsherwood | bah | 08:58 |
| *** bashrc_ has joined #baserock | 09:04 | |
| *** toscalix has joined #baserock | 09:11 | |
| pedroalvarez | I've been wondering if instead of writing our own software to trigger ci builds, we reuse Zuul and turbo-hipster | 09:15 |
| jjardon | pedroalvarez: IIRC, KDE uses that | 09:18 |
| jjardon | https://gerrit.vesnicky.cesnet.cz/r/#/q/status:open | 09:19 |
| pedroalvarez | OpenStack uses Zuul + Jenkins plugin I believe | 09:20 |
| *** CTtpollard has quit IRC | 09:27 | |
| gtristan | Well... aboriginal ybd is ... *finally* off the ground, as of this morning | 09:38 |
| pedroalvarez | woooooooooo | 09:40 |
| *** jonathanmaw has joined #baserock | 09:51 | |
| *** rdale has joined #baserock | 09:53 | |
| SotK | pedroalvarez: +1000 | 09:53 |
| *** rdale has quit IRC | 09:54 | |
| *** CTtpollard has joined #baserock | 10:02 | |
| *** rdale has joined #baserock | 10:02 | |
| *** rdale has quit IRC | 10:07 | |
| *** rdale has joined #baserock | 10:08 | |
| *** ssam2 has joined #baserock | 10:10 | |
| *** ChanServ sets mode: +v ssam2 | 10:10 | |
| *** edcragg has quit IRC | 10:18 | |
| *** locallycompact has joined #baserock | 10:18 | |
| *** fay has joined #baserock | 10:28 | |
| *** fay is now known as faybrocklebank | 10:28 | |
| *** edcragg has joined #baserock | 10:32 | |
| pedroalvarez | locallycompact: https://irclogs.baserock.org/%23baserock.2016-03-01.log.html#t2016-03-01T09:15:39 | 10:55 |
| *** jonathanmaw has quit IRC | 10:56 | |
| *** jonathanmaw has joined #baserock | 10:57 | |
| locallycompact | pedroalvarez, can you elaborate? I've read some of the Zuul and still don't really understand what it is | 10:57 |
| locallycompact | *manual | 10:57 |
| * locallycompact is also going nowhere near Jenkins, ftr | 10:57 | |
| paulsherwood | gtristan: pls paste the log? | 11:04 |
| paulsherwood | w00t btw | 11:04 |
| pedroalvarez | locallycompact: if what you are looking at connecting Gerrit patches to CI jobs, Zuul + turbohipster can help you to fill that gap | 11:04 |
| pedroalvarez | Zuul will read Gerrit stream events, and create Jobs for them in Gearman Job queue | 11:05 |
| ssam2 | SotK more or less completely implemented that a year or two ago | 11:05 |
| ssam2 | patches are in the Baserock mailing list archives | 11:05 |
| pedroalvarez | then using turbohipster you can get jobs from that queue and execute them | 11:06 |
| gtristan | paulsherwood, the log is a work in progress :-/ | 11:06 |
| pedroalvarez | (I'm not sure about how all of this works, tbh) | 11:06 |
| paulsherwood | gtristan: :) | 11:06 |
| * gtristan just got out of meeting and was checking if he could wrangle up an armv7lhf aboriginal quickly | 11:06 | |
| ssam2 | definitions.git branch baserock/baserock/adamcoldrick/zuul-mason-v2 might be useful | 11:06 |
| SotK | there is a more up to date branch somewhere I think, but I might not have ever pushed it :/ | 11:08 |
| ssam2 | and http://git.baserock.org/cgit/baserock/baserock/system-tests.git/log/?h=baserock/mason-v2 is the turbo-hipster plugin part | 11:09 |
| ssam2 | which actually does the interesting work of triggering builds and then triggering tests of those builds | 11:09 |
| ssam2 | although I guess Concourse might take over that part, not sure what you're building | 11:09 |
| locallycompact | I want to keep that option available if at all possible. My plan was to try something like gerrit stream-events > jq filter > generate pipeline > deploy to concourse, that seemed like the path of least resistance. | 11:12 |
| locallycompact | But I don't want to miss a trick with this zuul thing, if it has any | 11:13 |
| locallycompact | But I also want to avoid introducing more python wherever I don't need it | 11:15 |
| ssam2 | zuul basically goes 'gerrit stream-events -> some kind of task runner', if i remember rightly | 11:17 |
| ssam2 | where the task runner has to speak Gearman (which is fairly simple I think) | 11:17 |
| ssam2 | the trick that Zuul does is it handles cases where you have 10 patches to test and some of them depend on each other | 11:17 |
| *** jonathanmaw_ has joined #baserock | 11:18 | |
| *** jonathanmaw has quit IRC | 11:18 | |
| ssam2 | i *think* that rather than trying to testing 1,2,3,4,5,7,8,9 and then 10 (which will be slow!) it tries to merge them all together, and if that succeeds, tests them all, and if that succeeds, they all pass | 11:18 |
| ssam2 | if you don't think you need that, probably handling `gerrit stream-events` makes sense | 11:19 |
| ssam2 | handling it directly, I mean | 11:19 |
| locallycompact | We do need that, I was just going to write that logic. | 11:20 |
| * locallycompact digs a bit | 11:20 | |
| SotK | http://docs.openstack.org/infra/zuul/gating.html#testing-in-parallel explains what Zuul does in depth | 11:20 |
| pedroalvarez | I think having zuul in our infra will be really nice | 11:23 |
| *** jonathanmaw_ has quit IRC | 11:23 | |
| SotK | me too | 11:23 |
| *** jonathanmaw has joined #baserock | 11:23 | |
| *** bashrc_ has quit IRC | 11:35 | |
| *** ssam2 has quit IRC | 11:37 | |
| *** jonathanmaw has quit IRC | 11:39 | |
| *** locallycompact has quit IRC | 11:39 | |
| *** faybrocklebank has quit IRC | 11:39 | |
| *** tiagogomes has quit IRC | 11:39 | |
| *** ctbruce has quit IRC | 11:39 | |
| *** CTtpollard has quit IRC | 11:39 | |
| *** edcragg has quit IRC | 11:39 | |
| *** ctbruce has joined #baserock | 11:40 | |
| *** faybrocklebank has joined #baserock | 11:41 | |
| *** faybrocklebank has quit IRC | 11:43 | |
| *** ctbruce has quit IRC | 11:43 | |
| *** locallycompact has joined #baserock | 11:44 | |
| *** edcragg has joined #baserock | 11:45 | |
| *** paulw has joined #baserock | 11:45 | |
| *** edcragg has quit IRC | 11:47 | |
| *** locallycompact has quit IRC | 11:47 | |
| *** paulw has quit IRC | 11:48 | |
| *** edcragg has joined #baserock | 11:50 | |
| *** paulw has joined #baserock | 11:53 | |
| *** edcragg has quit IRC | 11:55 | |
| *** paulw has quit IRC | 11:58 | |
| *** SotK_ has joined #baserock | 12:15 | |
| *** edcragg has joined #baserock | 12:23 | |
| *** tiagogomes has joined #baserock | 12:24 | |
| *** fay has joined #baserock | 12:24 | |
| *** paulw has joined #baserock | 12:24 | |
| *** ssam2 has joined #baserock | 12:24 | |
| *** ChanServ sets mode: +v ssam2 | 12:24 | |
| *** fay is now known as Guest2006 | 12:25 | |
| *** jonathanmaw has joined #baserock | 12:25 | |
| *** Guest2006 is now known as faybrocklebank | 12:27 | |
| *** CTtpollard has joined #baserock | 12:28 | |
| *** bashrc_ has joined #baserock | 12:31 | |
| gtristan | ewww... baserock doesnt have sed in the form of a tarball | 12:31 |
| *** paulw has quit IRC | 12:31 | |
| *** faybrocklebank has quit IRC | 12:32 | |
| *** fay_ has joined #baserock | 12:32 | |
| gtristan | any prior art/policy about cases where we might actually want both a tarball _and_ a git ? | 12:32 |
| * gtristan will lorry a tarball locally for the moment and use a local git url | 12:33 | |
| paulsherwood | gtristan: http://git.baserock.org/cgit/?q=gcc | 12:33 |
| gtristan | makes sense for that early stage, right | 12:34 |
| * gtristan wonders how he succeeds to build gawk | 12:34 | |
| gtristan | weird, http://git.baserock.org/cgit/delta/gawk.git/commit/?id=dc5af665700d9b04fdf9c18930526d28eef5d5d9 | 12:37 |
| *** locallycompact has joined #baserock | 12:37 | |
| gtristan | looks like gawk maintainers make commits of fully generated tarball scripts | 12:37 |
| * gtristan grunts and carries on... | 12:37 | |
| *** jonathanmaw has quit IRC | 12:40 | |
| *** jonathanmaw has joined #baserock | 12:40 | |
| *** edcragg has quit IRC | 12:42 | |
| *** edcragg has joined #baserock | 12:42 | |
| edcragg | are there any ways to ensure morph uses as many cores as possible? | 12:47 |
| ssam2 | the max-jobs fields controls what it sets for MAKEFLAGS | 12:49 |
| ssam2 | the default value for max-jobs is calculated in utils.py | 12:49 |
| gtristan | Ummm, was there a regression in file url support in ybd ? | 12:49 |
| gtristan | This is valid definition: | 12:49 |
| gtristan | repo: file:///home/tristan/codethink/BASEROCK/REPOS/sed-tarball | 12:49 |
| gtristan | ? | 12:49 |
| paulsherwood | gtristan: actually, i have not tried that for a long time, so maybe | 12:51 |
| paulsherwood | gtristan: can you paste a log? | 12:52 |
| *** bashrc_ has quit IRC | 12:53 | |
| *** edcragg has quit IRC | 12:55 | |
| *** ssam2 has quit IRC | 12:55 | |
| *** edcragg has joined #baserock | 12:55 | |
| gtristan | paulsherwood, relevant bit here: http://paste.baserock.org/lowivisavi | 12:55 |
| *** fay_ has quit IRC | 12:55 | |
| *** jonathanmaw has quit IRC | 12:55 | |
| gtristan | it does clone into a tmp dir fine | 12:56 |
| *** jonathanmaw has joined #baserock | 12:56 | |
| gtristan | and I double checked the sha1 (ref), it's there in the cloned dir | 12:56 |
| paulsherwood | ack | 12:57 |
| paulsherwood | gtristan: and you're using latest tag? | 12:57 |
| paulsherwood | (of ybd) | 12:57 |
| *** tom__ has joined #baserock | 12:58 | |
| *** fay has joined #baserock | 12:58 | |
| *** fay is now known as Guest22386 | 12:59 | |
| *** ssam2 has joined #baserock | 12:59 | |
| *** ChanServ sets mode: +v ssam2 | 12:59 | |
| *** CTtpollard has quit IRC | 13:01 | |
| gtristan | paulsherwood, no last week :-/ | 13:02 |
| * gtristan tries rebasing against new tag | 13:02 | |
| paulsherwood | heh | 13:03 |
| paulsherwood | it *shouldn't* behave differently, but let's be on the same page :) | 13:03 |
| gtristan | eek | 13:05 |
| *** ssam2 has quit IRC | 13:05 | |
| paulsherwood | ? | 13:06 |
| gtristan | http://paste.baserock.org/apafaxibof | 13:06 |
| gtristan | pip install <smth> ? | 13:06 |
| *** tiagogomes has quit IRC | 13:06 | |
| gtristan | osfs | 13:06 |
| paulsherwood | pip install fs | 13:06 |
| paulsherwood | documentation has been updated | 13:06 |
| gtristan | same error | 13:07 |
| *** jonathanmaw has quit IRC | 13:07 | |
| *** tom__ has quit IRC | 13:07 | |
| *** tiagogomes has joined #baserock | 13:07 | |
| *** tom__ has joined #baserock | 13:07 | |
| paulsherwood | you're saying you still get 'ImportError: No module named fs.osfs;' after pip install fs ? | 13:08 |
| gtristan | no no | 13:08 |
| gtristan | same git clone error on the file:///path/to/repo | 13:08 |
| paulsherwood | aha, ok | 13:08 |
| *** Guest22386 has quit IRC | 13:09 | |
| gtristan | paulsherwood, looking at repos.py, says something about hardlinks option | 13:09 |
| gtristan | could it have to do with the fact that my ybd 'gits' is configured in a different partition ? | 13:10 |
| paulsherwood | possibly. | 13:10 |
| paulsherwood | could you try running that git command in situ? | 13:10 |
| paulsherwood | ie git clone --no-hardlinks gitdir checkout | 13:11 |
| paulsherwood | if you want to get exact values of gitdit checkout.... | 13:11 |
| paulsherwood | put the following in the line above: | 13:12 |
| paulsherwood | import pdb ; pdb.set_trace() | 13:12 |
| paulsherwood | and then you can type the variable names to see what they contain (sorry if you already know this) | 13:12 |
| gtristan | hmmm | 13:12 |
| gtristan | fatal: This operation must be run in a work tree | 13:12 |
| gtristan | I chdir'ed into the repo it cloned in and ran: git checkout --force <ref> | 13:13 |
| gtristan | thats what it tells me | 13:13 |
| paulsherwood | hmmm. | 13:13 |
| gtristan | oddly the clone command doesnt say --bare | 13:13 |
| gtristan | but the repo is a bare one | 13:13 |
| paulsherwood | gtristan: to allow progress, maybe just push your sed-tarball repo somewhere and link to that? | 13:13 |
| *** tom__ has quit IRC | 13:13 | |
| * paulsherwood will need to investigate properly later | 13:14 | |
| *** tiagogomes has quit IRC | 13:14 | |
| *** jonathanmaw has joined #baserock | 13:14 | |
| *** edcragg has quit IRC | 13:15 | |
| *** tom__ has joined #baserock | 13:16 | |
| *** tom__ has quit IRC | 13:17 | |
| *** jonathanmaw has quit IRC | 13:23 | |
| *** CTtpollard has joined #baserock | 13:25 | |
| *** fay_ has joined #baserock | 13:25 | |
| *** jonathanmaw has joined #baserock | 13:25 | |
| *** bashrc_ has joined #baserock | 13:26 | |
| *** tiagogomes has joined #baserock | 13:32 | |
| *** paulw has joined #baserock | 13:32 | |
| gtristan | paulsherwood, anyway I'm gonna pack up for now... but what I've found is with GIT_TRACE=1 it shows some useful context | 13:32 |
| gtristan | it tells me that the tmpdir it clones to is in fact the one setup with --mirror | 13:33 |
| gtristan | even though it errors out at the line with --no-hardlinks | 13:33 |
| gtristan | paulsherwood, os.rename() must be the problem | 13:36 |
| gtristan | if I understand correctly, that will not work across file systems | 13:36 |
| gtristan | if os.rename() is what I think it is, anyway | 13:37 |
| *** franred has joined #baserock | 13:37 | |
| *** ctbruce has joined #baserock | 13:41 | |
| *** gtristan has quit IRC | 13:41 | |
| *** bashrc_ has quit IRC | 13:44 | |
| richard_maw | aye, os.rename() doesn't work across fliesystems, there'll be something in shutil vofor muvoving | 13:45 |
| *** fay_ has quit IRC | 13:45 | |
| *** franred has quit IRC | 13:47 | |
| *** jonathanmaw has quit IRC | 13:47 | |
| *** jonathanmaw has joined #baserock | 13:48 | |
| *** ctbruce has quit IRC | 13:49 | |
| *** ctbruce has joined #baserock | 13:50 | |
| *** CTtpollard has quit IRC | 13:52 | |
| *** CTtpollard has joined #baserock | 13:53 | |
| *** fay_ has joined #baserock | 13:53 | |
| *** franred has joined #baserock | 13:54 | |
| locallycompact | So where is the gearman setup in this old zuul branch? | 13:56 |
| *** bashrc_ has joined #baserock | 13:58 | |
| *** edcragg has joined #baserock | 14:00 | |
| *** gtristan has joined #baserock | 14:01 | |
| *** jonathanmaw has quit IRC | 14:05 | |
| *** jonathanmaw has joined #baserock | 14:05 | |
| *** fay_ is now known as faybrocklebank | 14:09 | |
| *** ssam2 has joined #baserock | 14:17 | |
| *** ChanServ sets mode: +v ssam2 | 14:17 | |
| locallycompact | How far did this zuul system get? What was it capable of when the branch was current | 14:30 |
| locallycompact | Because I don't see gearman in g.b.o and this seems to imply it's essential to the whole thing | 14:31 |
| locallycompact | http://docs.openstack.org/infra/zuul/zuul.html | 14:31 |
| ssam2 | it got to working prototype stage | 14:31 |
| ssam2 | gearman is a protocol | 14:31 |
| pedroalvarez | but I believe gerrit was running on ubuntu when we investigated this | 14:31 |
| ssam2 | the way zuul-mason was intended to work was with some Baserock-specific task runners in the system-test.git repo | 14:32 |
| ssam2 | these: http://git.baserock.org/cgit/baserock/baserock/system-tests.git/log/?h=baserock/mason-v2 | 14:32 |
| ssam2 | these were implemented with something called Turbo Hipster, but that's kind of an implementation detail | 14:32 |
| ssam2 | the point is that using the Gearman protocol, Zuul could trigger these task-runners | 14:32 |
| ssam2 | Turbo Hipster is a generic way of writing task runners that speak Gearman, I think | 14:33 |
| ssam2 | https://github.com/openstack/turbo-hipster | 14:33 |
| pedroalvarez | but for gearman you need the actual gearmand server running somewhere | 14:34 |
| locallycompact | I think this is what I meant in the first place ^ ? | 14:34 |
| locallycompact | Where was this thing | 14:34 |
| pedroalvarez | so, I think Gearan job server was just installed in Gerrit, which was ubuntu. But I guess only SotK knows | 14:35 |
| ssam2 | ah, i see | 14:37 |
| ssam2 | actually I don't | 14:38 |
| ssam2 | isn't Zuul the Gearman job server? | 14:38 |
| locallycompact | This thin says in the conf: http://docs.openstack.org/infra/zuul/zuul.html | 14:38 |
| locallycompact | gearman | 14:38 |
| locallycompact | server | 14:38 |
| locallycompact | Hostname or IP address of the Gearman server. server=gearman.example.com | 14:38 |
| ssam2 | oh, yeah | 14:38 |
| ssam2 | ok | 14:38 |
| pedroalvarez | Zuul, I guess is a Gearman client that requests jobs to the gearman server | 14:38 |
| * richard_maw attempted to build gearman in baserock once, to look to see if it could be made to socket activate | 14:39 | |
| richard_maw | got it building and running, but ugh | 14:39 |
| richard_maw | massive pile of C++ and Boost | 14:39 |
| * persia seems to recall SotK getting a zull+gearman based Mason up at some point, although not deployed | 14:40 | |
| locallycompact | richard_maw, what's 'socket activate'? | 14:40 |
| ssam2 | why does gearmand have to be in Baserock? | 14:40 |
| persia | ssam2: At one point, the goal was to have everything used by baserock built by baserock. I don't know that we hold that as dear today. | 14:40 |
| SotK | the mason-v2 branch ran gearman in the same machine as Zuul | 14:41 |
| richard_maw | locallycompact: server doesn't bind the sockets, systemd does, and starts the service when something connects to the socket | 14:41 |
| * paulsherwood still holds it dear | 14:41 | |
| locallycompact | richard_maw, ah | 14:41 |
| SotK | locallycompact: the system in that branch would run zuul, gearman and turbo-hipster | 14:42 |
| locallycompact | where's gearman in g.b.o or definitions? | 14:42 |
| richard_maw | it's not, the last time I looked | 14:42 |
| richard_maw | I hacked about, since I think the version of boost was incompatible | 14:42 |
| locallycompact | richard_maw, SotK just said it was in the system in that branch | 14:43 |
| locallycompact | I still don't understand where turbo hipster fits into this, tw | 14:43 |
| locallycompact | *btw | 14:43 |
| * richard_maw remembers why he was investigating gearman now, he was looking to see if he could configure gearman to start on an ephemeral socket and report which port it bound to, and concluded that making it socket activate would be more useful | 14:43 | |
| richard_maw | locallycompact: I never did either | 14:43 |
| * SotK goes to look | 14:44 | |
| persia | I thought turbo-hipster was just a gearman worker that did certain sorts of DB-related test jobs | 14:46 |
| * SotK can't find it, I don't know how it was working off the top of my head | 14:46 | |
| persia | Was it http://git.baserock.org/cgit/baserock/baserock/definitions.git/log/?h=baserock/adamcoldrick/zuul-mason-v2 ? | 14:50 |
| SotK | that was indeed the working branch | 14:52 |
| SotK | I did some stuff to get it more up to date, but then ran out of time to finish | 14:52 |
| locallycompact | More up to date than what's there? | 14:52 |
| SotK | http://git.baserock.org/cgit/baserock/baserock/definitions.git/commit/?h=baserock/adamcoldrick/mason-2015 | 14:52 |
| locallycompact | ^ is same but better? | 14:53 |
| locallycompact | There's a thing just called 'gear' in there | 14:54 |
| locallycompact | Whats that | 14:54 |
| locallycompact | not the backend | 14:54 |
| locallycompact | So, if this is mason, what are the references to 'mason' in definitions master? | 14:56 |
| richard_maw | masonv1 | 14:56 |
| *** richard_maw has left #baserock | 14:57 | |
| locallycompact | What does it do? | 14:57 |
| SotK | gear is a python api for interacting with gearman iirc | 14:57 |
| SotK | masonv1 runs a build of ci.morph every time there is a new commit to master I think | 14:57 |
| locallycompact | ok | 14:58 |
| persia | (for those counting, mason versions seem to start at 0 today, with mason v0 being the jenkins thing) | 15:03 |
| locallycompact | so how does gearman spin up workers to do things | 15:04 |
| locallycompact | putting aside the mystery of its not being anywhere | 15:04 |
| SotK | elasticity was an unsolved problem with the zuul work when we ran out of time for it | 15:05 |
| locallycompact | ok cool | 15:06 |
| pedroalvarez | I think you can have workers anywere, if they have access to the gearman job server :) | 15:09 |
| persia | Note that some other zuul users (KDE, OpenStack, etc.) have solutions for elasticity. | 15:09 |
| locallycompact | oh I see | 15:09 |
| locallycompact | so gearman isn't spinning things up it's just being resourced for jobs | 15:10 |
| locallycompact | wait | 15:10 |
| locallycompact | ? | 15:10 |
| pedroalvarez | gearman workers will go to gearman job server to request jobs to do | 15:10 |
| locallycompact | but gearman itself doesn't deploy workers for that purpose? | 15:11 |
| pedroalvarez | gearman itself is only a job queue, producers will put jobs on it, and consumers will take them | 15:12 |
| pedroalvarez | just that | 15:12 |
| locallycompact | ok, nice | 15:12 |
| pedroalvarez | Zuul and turbo-hipster will use that to work | 15:12 |
| rjek | There is no turbo-hipster, only Zuul. | 15:13 |
| *** edcragg has quit IRC | 15:13 | |
| locallycompact | There does seem to be a turbo-hipster | 15:13 |
| locallycompact | It's in this system branch | 15:13 |
| persia | It is the primart consumer to use pedroalvarez' description of gearman | 15:13 |
| locallycompact | Right so contacting zuul to say, 'go tell gerrit all is well' is not done by gearman, it's done directly by the worker | 15:15 |
| locallycompact | and gearman gives the worker the information to do that | 15:15 |
| locallycompact | is that right? | 15:16 |
| SotK | I think turbo-hipster sets the job status to complete or something like that | 15:16 |
| locallycompact | The github page says | 15:16 |
| locallycompact | "Each task_plugin is a zuul gearman worker that implements, handles, executes a job, uploads/post-processes the logs and sends back the results to zuul." | 15:16 |
| pedroalvarez | right | 15:20 |
| SotK | yeah, it does that by setting the gearman job status to complete, which zuul then sees (since it created the job) iirc | 15:20 |
| locallycompact | But presumably there are more details involved in providing a +1 to gerrit than 'complete' | 15:21 |
| locallycompact | if it's just a 'complete', and that's just a gearman protocol, what is significant bout turbo-hipster being described specifically as a 'zuul gearman worker' | 15:22 |
| SotK | turbo-hipster understands the jobs that Zuul creates | 15:23 |
| locallycompact | ok, yeah | 15:23 |
| locallycompact | so the basic-est idea for mason was to have zuul, gearmand and turbo all in the same machine | 15:24 |
| locallycompact | with some elasticity solution for workers when the job queue gets slammed | 15:24 |
| locallycompact | to come later | 15:25 |
| SotK | yeah | 15:25 |
| SotK | openstack use nodepool to spin up places to run tests I think | 15:25 |
| locallycompact | more words :) | 15:26 |
| locallycompact | oh ok | 15:26 |
| locallycompact | looks a bit boshy | 15:26 |
| *** edcragg has joined #baserock | 15:28 | |
| locallycompact | so it's conceivable to write a turbo hipster plugin that just calls the concourse binary with the relevant generated pipeline | 15:29 |
| locallycompact | and noms the result somehow and spits it back to gearman, zuul then gerrit | 15:29 |
| SotK | yeah, turbo-hipster plugins can just be a shell script if you want I think | 15:30 |
| locallycompact | this is much clearer, thanks | 15:31 |
| SotK | yw | 15:31 |
| persia | locallycompact: The issue is visualisation: you'll ideally want a way to represent all the incoming candidates to some landing point graphically, along with the success/failure state, etc. | 15:40 |
| locallycompact | Huh? | 15:40 |
| *** zoli_ has quit IRC | 15:42 | |
| locallycompact | The success/failure should be detailed in gerrit, no? | 15:44 |
| *** jonathanmaw has quit IRC | 15:44 | |
| *** zoli_ has joined #baserock | 15:51 | |
| *** jonathanmaw has joined #baserock | 15:53 | |
| *** jonathanmaw has joined #baserock | 15:53 | |
| *** CTtpollard has quit IRC | 15:59 | |
| *** CTtpollard has joined #baserock | 16:05 | |
| persia | Yes, the CI system should submit gerrit reviews, but it's hard to see the total set of what is happening in a project with just that detail. | 16:13 |
| persia | Anyway, just a minor observation when looking at gerrit+zuul+gearman+turbo-hipster from the perspective of concourse. | 16:14 |
| persia | As concourse looks great, but assumes post-merge, etc. | 16:14 |
| locallycompact | concourse isn't critical here | 16:16 |
| *** ctbruce has quit IRC | 16:20 | |
| *** CTtpollard has quit IRC | 16:24 | |
| *** CTtpollard has joined #baserock | 16:24 | |
| SotK | persia: there was a plan early last year to write a UI similar to current mason's for Zuul, and also add a pipeline which builds current master regularly | 16:26 |
| SotK | https://storyboard.baserock.org/?#!/story/8 | 16:27 |
| persia | Strangely, that link redirected me to "https://storyboard.baserock.org/[object URL]#!/page/about" | 16:32 |
| franred | it does work for me | 16:33 |
| SotK | persia: what is "[object URL]"? | 16:33 |
| Zara | works for me | 16:33 |
| persia | SotK: A sequence of characters. Each is distinct for me. I can replicate with a few other things at storyboard.baserock.org: I've not seen this with other storyboards. | 16:34 |
| persia | I get the same thing using another browser, which usually solves this sort of issue for me. | 16:35 |
| SotK | persia: kind of like "?state=zyvwLrETp4kKzXvGHOnk&code=9tyRUQDv5reR8Rh63bhgYUmTB8H9Tc"? | 16:35 |
| persia | (although in the second browser I tried, it is "[object%20URL]" | 16:35 |
| persia | SotK: Yes: at least in my location areas, I don't seem to have any markup indicating it is a substitution. | 16:36 |
| SotK | pedroalvarez: does storyboard.baserock.org have the 401 fix in? | 16:37 |
| pedroalvarez | I think it does | 16:37 |
| SotK | hm | 16:37 |
| SotK | persia: if you log in and then paste the link, does it work? | 16:38 |
| *** franred has quit IRC | 16:39 | |
| persia | Hmm. Seems my password isn't what I thought it was. This will take longer than I expected. | 16:40 |
| * persia begins to wonder how often openid.baserock.org processes the mail queue | 16:45 | |
| jjardon | Im seeing this when building, should I be worried? | 16:49 |
| jjardon | https://www.irccloud.com/pastebin/tbptdFHM/ | 16:49 |
| ssam2 | persia: right away I think | 16:52 |
| ssam2 | might be greylisted? | 16:52 |
| ssam2 | jjardon: that's from the default splitting-rules, I think | 16:53 |
| ssam2 | jjardon: default strip-commands, I mean | 16:53 |
| ssam2 | jjardon: hmm, no, not that either | 16:53 |
| ssam2 | jjardon: it's from ybd running 'ldconfig -r $sandbox' | 16:54 |
| ssam2 | jjardon: there's a .py file inside usr/lib64 for some reason, ldconfig tries to parse it as a shared library, fails, then warns | 16:54 |
| persia | ssam2: Nope: No logs on the relevant mail server of any attempt by anything @ baserock.org | 16:55 |
| * persia tries again | 16:55 | |
| ssam2 | hmm... nothing much in the logs of sendmail since Feb 28th either | 16:55 |
| ssam2 | but it is running | 16:56 |
| persia | Just tried again, and I don't see anything. If you also don't see anything, I wonder if something between the web form and the mail server is unhappy. | 16:56 |
| persia | (I had thought the first error might have been a typo in the email address, but ...) | 16:56 |
| ssam2 | no errors | 16:57 |
| ssam2 | i don't think it logs password reset mails | 16:57 |
| ssam2 | i'll try myself | 16:57 |
| ssam2 | i got a mail right away | 16:58 |
| ssam2 | and sendmail logged it | 16:58 |
| ssam2 | so this is strange | 16:58 |
| ssam2 | Do you see 'Email with password reset instructions has been sent.' from the website? | 16:59 |
| edcragg | if anyone knows about autotools, i'm having this error when trying to build autotools on armv8: http://paste.baserock.org/daluwinuwu | 17:01 |
| ssam2 | what exactly are you building? | 17:04 |
| ssam2 | "autotools" is multiple pieces of software | 17:04 |
| ssam2 | it looks like you have the dependency order wrong, at a guess? | 17:04 |
| ssam2 | hmm... actually aclocal comes with automake, so I guess not | 17:05 |
| ssam2 | actually it seems to be running 'autoreconf' as part of the automake build, and it's that which is failing | 17:06 |
| persia | ssam2: Sorry: yes, I did see that message from the website, twice. | 17:06 |
| ssam2 | perhaps because it's missing aclocal, because aclocal isn't installed yet | 17:06 |
| ssam2 | persia: ok, hmm | 17:07 |
| persia | Could it be my address? Is there some sort of filter that gets unhappy if I use a .jp? | 17:07 |
| edcragg | ssam2: yes, it's the automake build | 17:08 |
| ssam2 | persia: the web application uses django-registration-redux to implement all this | 17:08 |
| *** gary_perkins has quit IRC | 17:08 | |
| persia | ssam2: default or simple backend? | 17:09 |
| edcragg | could this be a problem specific to building on armv8? i expect the automake build hasn't been found to be broken elsewhere | 17:09 |
| ssam2 | edcragg: it seems unlikely | 17:09 |
| ssam2 | persia: default | 17:09 |
| ssam2 | persia: version 1.3 | 17:09 |
| * edcragg curses autotools, which has only ever caused him pain | 17:10 | |
| *** faybrocklebank has quit IRC | 17:10 | |
| *** CTtpollard has quit IRC | 17:11 | |
| *** CTtpollard has joined #baserock | 17:12 | |
| persia | Hrm. I tried a fresh "register" action, hoping that would help, and got a 500. | 17:12 |
| * persia goes back to reading docs | 17:12 | |
| *** gary_perkins has joined #baserock | 17:14 | |
| *** jonathanmaw has quit IRC | 17:15 | |
| *** jonathanmaw has joined #baserock | 17:15 | |
| *** tiagogomes has quit IRC | 17:16 | |
| *** tiagogomes has joined #baserock | 17:16 | |
| *** jonathanmaw has quit IRC | 17:16 | |
| pedroalvarez | right! openssl upgrade time! https://gerrit.baserock.org/#/c/1945 | 17:17 |
| ssam2 | persia: oh, the 500 is new | 17:18 |
| ssam2 | persia: at least those are logged :-) | 17:18 |
| *** jonathanmaw has joined #baserock | 17:18 | |
| persia | Heh, cool. | 17:18 |
| paulsherwood | pedroalvarez: +2 | 17:18 |
| persia | Is there anything else I can do to help troubleshoot? | 17:18 |
| ssam2 | persia: registration turns out to be totally broken (probably since the upgrade to Django 1.9), so thanks for finding that! | 17:19 |
| *** CTtpollard has quit IRC | 17:19 | |
| *** CTtpollard has joined #baserock | 17:19 | |
| ssam2 | persia: not sure what else, I guess the next step is get logging from django-registration-redux, but i'm not sure exactly how. | 17:19 |
| persia | ssam2: Ah, excellent! If you get it working, please let me know, so I can re-register and/or adjust my password. | 17:19 |
| ssam2 | might be just django logging config | 17:20 |
| pedroalvarez | paulsherwood: ta! | 17:20 |
| persia | SotK: Apparently, I can't test the authenticated behaviour at storyboard-baserock.org today: can you test the not-authenticated behaviour? If you can't replicate, I'll try again when I can test both ways. | 17:21 |
| * SotK tests it | 17:22 | |
| SotK | works for me when logged out | 17:22 |
| ssam2 | upgrading to django-registration-redux 1.4 just broke things in a different way | 17:24 |
| ssam2 | this is pretty rubbish | 17:24 |
| ssam2 | i don't have time to dig in to this tonight | 17:24 |
| persia | SotK: In that case, I'll try to replicate some other time, and let you know if I can figure out how that happened. | 17:24 |
| persia | ssam2: No worries. Let me know if you want me to try things again once you do have time. | 17:25 |
| ssam2 | The new failure appears to be because Django's database migrations don't actually work usefully | 17:31 |
| ssam2 | ok, hopefully registration works again, I've manually added the column that the migrations failed to | 17:33 |
| ssam2 | nope... | 17:33 |
| ssam2 | will look at this further tomorrow, bye! | 17:34 |
| *** ssam2 has quit IRC | 17:34 | |
| pedroalvarez | ANNOUNCEMENT: I'll reboot git.baserock.org to finish the upgrade | 17:36 |
| paulsherwood | ack | 17:36 |
| pedroalvarez | it seems to be working :) | 17:40 |
| paulsherwood | how come it's lorrying elixir, but locallycompact's patch is still +1 | 17:41 |
| paulsherwood | but yes, it seems to be working in a lovely way :) | 17:42 |
| pedroalvarez | hm? that patch is to add a definition, not to lorry it | 17:42 |
| paulsherwood | ah ok | 17:42 |
| paulsherwood | how often is http://git.baserock.org/lc-status.html refreshed? | 17:43 |
| pedroalvarez | every miniute I think | 17:43 |
| paulsherwood | not afaict | 17:44 |
| pedroalvarez | it'supposed to update every minute, yes | 17:47 |
| pedroalvarez | paulsherwood: what do you see to think that it's not happening? | 17:47 |
| paulsherwood | nothing, now :) it's refreshed | 17:48 |
| *** jonathanmaw has quit IRC | 18:00 | |
| *** bashrc_ has quit IRC | 18:01 | |
| *** toscalix has quit IRC | 18:27 | |
| *** jjardon_ has joined #baserock | 18:33 | |
| gary_perkins | Hi, would it be too difficult to import fail2ban into BR? Our public BR servers are getting hammered by the Chinese | 18:34 |
| pedroalvarez | gary_perkins: I would be happy to try | 18:36 |
| gary_perkins | pedroalvarez: Awesome! Thanks. It would make us all feel a bit safer :) | 18:37 |
| *** jjardon has quit IRC | 18:37 | |
| *** jjardon_ is now known as jjardon | 18:39 | |
| *** wdutch has quit IRC | 19:25 | |
| *** edcragg has quit IRC | 19:30 | |
| *** wdutch has joined #baserock | 19:31 | |
| *** rdale_ct has joined #baserock | 19:31 | |
| *** rdale has quit IRC | 19:34 | |
| *** rdale has joined #baserock | 19:38 | |
| *** rdale_ct has quit IRC | 19:41 | |
| *** cosm has joined #baserock | 19:51 | |
| *** rdale_ct has joined #baserock | 19:54 | |
| *** rdale has quit IRC | 19:58 | |
| *** brlogger has joined #baserock | 19:59 | |
| *** edcragg has joined #baserock | 20:13 | |
| *** rdale has joined #baserock | 20:19 | |
| *** rdale_ct has quit IRC | 20:22 | |
| *** rdale has quit IRC | 20:25 | |
| *** rdale has joined #baserock | 20:26 | |
| *** locallycompact has quit IRC | 22:46 | |
| *** locallycompact has joined #baserock | 22:48 | |
| *** locallycompact has quit IRC | 23:27 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!