00:15:20  <jbergstroem>rvagg: i'll replace the smartos15 release slave now
00:16:30  <rvagg>OSTYPE, initially came from Windows and the Linux version came from somewhere else (I don't recall where, I just standardised on it). I'm happy to use a system property for it but I don't know one
00:16:40  <rvagg>i.e.: I made it up (mostly)
00:16:50  <rvagg>jbergstroem: what changes are you making to smartos15?
00:17:00  <jbergstroem>rvagg: downgrading to base14
00:17:19  <rvagg>ok, nightlies are re-enabled so we can test succes with that
00:17:25  <jbergstroem>great.
00:17:29  <rvagg>also need a 5.2.0 out this week, I guess I'll start that process if nobody else has
00:17:35  <jbergstroem>i'll copy the labels over
00:18:07  <jbergstroem>do i change the smartos15-release label over stick with it for now to avoid breakage?
00:18:57  * trottjoined
00:19:06  <jbergstroem>(kept it, feel free to update as you go)
00:21:39  <jbergstroem>btw whined to gh support about them removing annotations from the http://github.com/username.key thing - if we used it we could pull the release/test/infra keys as part of the setup
00:25:45  * trottquit (Quit: leaving)
00:31:37  <rvagg>good one, I bet they strip annotations when uploading keys tho
00:32:03  <jbergstroem>someone being "smart"
00:32:17  <rvagg>re smartos15-release, we should update everything to be smartos14-release right? I think that's just the iojs+release job
00:32:30  <jbergstroem>yes that's it
00:32:33  <jbergstroem>i just don't know all the consumers
00:33:01  <jbergstroem>and when the day of reckoning comes (ansible refactor), it will generate the labels in jenkins for us
00:33:22  <jbergstroem>so it'll probably be smartos14-x64-release
00:33:29  <jbergstroem>(in this case)
00:36:16  <jbergstroem>ok, up and running: https://ci.nodejs.org/computer/release-joyent-smartos14-x64-1/
00:36:22  <jbergstroem>i'll flag the other as offline forn ow
00:49:38  <jbergstroem>i've also renamed the other release machine (smartos13)
00:49:42  <jbergstroem>we should be done with smartos now
00:49:56  <jbergstroem>needs some minor improvements and updates, but yeah.
00:50:15  <jbergstroem>fyi - all tests machines with the new naming scheme also has the new test/release/infra keys.
01:40:24  * chorrelljoined
01:52:20  * bnoordhuisquit (Ping timeout: 256 seconds)
02:22:44  * chorrellquit (Quit: My Mac has gone to sleep. ZZZzzz…)
02:59:05  * bnoordhuisjoined
03:03:49  * bnoordhuisquit (Ping timeout: 250 seconds)
04:05:40  <jbergstroem>rvagg: i think platform.system on windows returns "windows"
04:15:04  <jbergstroem>Windows even
04:27:12  <rvagg>jbergstroem: yeah, but echo %OSTYPE% is where it comes from
04:27:40  <rvagg>i.e. it's already defined on Windows, that's where the pattern started .. and I just continued it, helpfully or not, perhaps we should just be using the labels in jenkins for that
04:28:04  <jbergstroem>my plan was to get it from ansible so we could put it in as part of templating
04:28:14  <jbergstroem>but it seems like ansible_system is something different
04:28:20  <jbergstroem>was hoping just to get platform.system
04:29:18  <jbergstroem>seems to be this on windows: $osversion.Platform.ToString()
04:30:32  <jbergstroem>sometimes it's just amazing how complicated things can get in ansible
04:30:39  <jbergstroem>and then, most of the time, it's like a breeze
08:56:48  <jbergstroem>wtf /usr/lib and /lib is differnt on ubuntu? what kind of craziness is that
09:09:05  <jbergstroem>rvagg: could we get commit logs in here from the nodejs/build repo?
09:09:20  <jbergstroem>i think its one of those hooks for a repo
09:09:26  <rvagg>yessir
09:09:30  <rvagg>I'll do itn ow
09:11:04  <rvagg>moved from #io.js to here although the test doesn't seem to have worked ...
09:11:54  <jbergstroem>i've added two new ubuntu14 running at softlayer as well as renamed the 15 one
09:12:00  * node-ghjoined
09:12:00  * node-ghpart
09:12:06  <jbergstroem>they're actually running upstart and systemd, but i don't feel safe enough to commit that quite yet
09:12:21  <rvagg>noice
09:12:47  <jbergstroem>in some scripts we increase file descriptors. is that for node or libuv?
09:12:59  <rvagg>both I think
09:13:16  <rvagg>tho I haven't seen a problem recently, I know we ignore soft limits in node since 0.12 so maybe that's why
09:13:20  <jbergstroem>some of our deployments doesn't have it and i don't think i've seen failed tests recently
09:13:34  <jbergstroem>just good to know whether we should be consistent about it
09:13:36  <jbergstroem>(and remove it)
09:14:08  <rvagg>can probably remove it, it'd be nice not to have to have it because we're closer to the average dev machine
09:14:32  <jbergstroem>exactly
09:14:35  <jbergstroem>and smaller scripts
09:14:44  <jbergstroem>i'm currently working on reordering scripts so we don't bundle by os
09:15:04  <jbergstroem>but to do so i have to make a bit of branching to figure out which init script lives at which host based on gather_fact
09:15:09  <jbergstroem>*gather_facts
09:16:08  <jbergstroem>and the less exceptions we have (like above) the better
09:16:38  <jbergstroem>i think rasbian is the only os that isn't covered by init scripts now
09:30:34  <jbergstroem>joaocgreis: how far away are we from having ssh access to windows? is that even a thing? just checked that pr again
09:34:59  <joaocgreis>jbergstroem: not a thing. I was trying everything to solve the Jenkins disconnect problem, but windows ssh is not easy to do well, if it's even possible
09:35:16  <jbergstroem>joaocgreis: ok, but whats the thing with powershell scripting and ansible?
09:35:23  <jbergstroem>joaocgreis: i'd lvo to see windows scriptable
09:35:27  <jbergstroem>*love
09:36:24  <joaocgreis>jbergstroem: https://github.com/nodejs/build/pull/215 I used that because it is supported by ansible
09:38:14  <jbergstroem>rvagg: do we still need PRE_1_PATH in init scripts?
09:38:23  <jbergstroem>joaocgreis: yeah that's what i saw and liked
09:38:54  <jbergstroem>joaocgreis: if you have a sec, updating passwords and the repo would be awesome for windows machines. that way we can give the test group access
09:39:08  <jbergstroem>joaocgreis: i can start if you want -- how do i change the password? control center someething?
09:39:50  <joaocgreis>jbergstroem: I'm doing it :)
09:43:50  <jbergstroem>joaocgreis: cool, you can copy the template i made for infra
09:44:23  <jbergstroem>https://gist.github.com/jbergstroem/e6597aae4beecb17b762
09:53:35  <rvagg>jbergstroem: https://cloudup.com/c4V4iF5rggu from https://ci.nodejs.org/job/iojs+release/configure
09:56:08  <jbergstroem>gotcha
10:04:58  <joaocgreis>jbergstroem: catching up with the changes in secrets, nice work there
10:05:16  <jbergstroem>joaocgreis: thanks
10:21:15  * thealphanerdquit (Ping timeout: 260 seconds)
10:23:42  * thealphanerdjoined
10:25:34  * node-ghjoined
10:25:34  * node-ghpart
10:38:13  * bnoordhuisjoined
10:41:57  * node-ghjoined
10:41:57  * node-ghpart
11:05:49  <joaocgreis>I will now remove iojs-rackspace-win2012r2-msvs2015-1 as we had talked about before. Any objections? rvagg jbergstroem
11:06:19  <jbergstroem>joaocgreis: assume i forgot all about the discussion and enlighten me again :-D
11:09:07  <joaocgreis>yep, long time ago! That machine was created by rvagg to test vs2015. Now we have vs2015 deployed, running on other machines, so that one is just sitting there looking pretty. I want to move the resources to the win2008 slaves that I'm about to recreate, hoping to speed up the test build by at least a couple of minutes.
11:09:25  <rvagg> right, kill it
11:16:15  <rvagg>bashtastic https://github.com/nodejs/node/pull/4194
11:16:51  <rvagg>a 1/2-finished project I just picked up again and completed! I have too many unfinished things laying around
11:17:51  <jbergstroem>what is this sed dark magic
11:20:48  <rvagg>it's pretty horrid but I've avoided cross-platform hiccups so it's lowest-common-denominator stuff
11:26:21  <jbergstroem>great! jenkins decided to freeze
11:29:17  <jbergstroem>all threads locked; majority here: - jenkins.model.lazy.AbstractLazyLoadRunMap.load(java.io.File, jenkins.model.lazy.AbstractLazyLoadRunMap$Index) @bci=0, line=465 (Compiled frame)
11:30:08  <rvagg>aieee
11:30:12  <jbergstroem>and restart worked
11:30:16  <jbergstroem>great!
11:33:29  <jbergstroem>hm
11:33:47  <jbergstroem>our smartos 32-bit machine was running 64-bit builds up to a few builds ago
11:33:58  <jbergstroem>problem was that we put destcpu in the init scripts
11:40:31  <rvagg>release machine?
11:40:33  <rvagg>derp
11:55:31  <jbergstroem>no, we only have 64-bit release machines
11:55:39  <jbergstroem>but pass that stuff trhough job
12:38:04  * chorrelljoined
12:39:25  * chorrellquit (Client Quit)
14:01:39  * chorrelljoined
15:54:24  * chorrellquit (Quit: Textual IRC Client: www.textualapp.com)
15:59:00  * jgijoined
16:01:50  * jgiquit (Client Quit)
16:57:59  * bnoordhuisquit (Ping timeout: 250 seconds)
17:28:38  * jgijoined
17:55:21  * bnoordhuisjoined
17:56:22  * jgiquit (Quit: jgi)
17:59:08  * jgijoined
18:16:08  * jgiquit (Quit: jgi)
18:25:23  * jgijoined
18:29:05  * jgiquit (Client Quit)
18:48:44  * mariviosojoined
18:48:57  <marivioso>hola
18:50:48  * mariviosoquit (Remote host closed the connection)
18:52:33  * jgijoined
18:55:44  * trottjoined
18:57:26  * jgiquit (Ping timeout: 256 seconds)
19:23:34  <trott>FYI: ubuntu1504-64 build keeps failing on CI.
19:24:06  <trott>armv8-ubuntu1404 too
19:28:03  <rvagg>trott: got some links to failures for us?
19:41:53  * jgijoined
19:42:04  * jgiquit (Client Quit)
19:48:50  <trott>https://ci.nodejs.org/job/node-test-commit-linux/1457/nodes=ubuntu1504-64/
19:49:24  <trott>https://ci.nodejs.org/job/node-test-commit-arm/1423/nodes=armv8-ubuntu1404/console
19:49:32  <trott>rvagg: Those are the two most recent ones I've seen.
19:50:38  <rvagg>grr, that stupid Java error on the armv8 machine ...
19:51:14  <rvagg>jbergstroem: that 15.04 error is weird, any ideas what might be up there?
19:51:48  <jbergstroem>ill check
19:52:02  <jbergstroem>likely related to switching systemd
19:52:18  <jbergstroem>+engrish
19:53:04  <rvagg>blaming systemd is enough for me, that'll contribute to my general systemd rage
19:53:21  <rvagg>trott: I've restarted the armv8 jenkins slave processes, cross fingers and hope that does the trick
19:53:39  <rvagg>trott: thanks for reporting! you're the test master!
19:54:35  <jbergstroem>rvagg: should be fixed now
19:54:53  <jbergstroem>ubuntu is really weird
19:56:23  <jbergstroem>trott: ^ sorry
19:57:41  <trott>On the stress test front, when I select win-vs2015, it seems to be stuck on "still in the queue: Waiting for next available executor on win-vs2015" forever. Not sure how to check to see if it's spurious or not, but I think it is.
20:00:51  <rvagg>whoa .. windows is not happy
20:01:28  <rvagg>joaocgreis: see trott's message above re win-vs2015
20:02:01  <joaocgreis>I'm recreating servers and moving them to test in secrets
20:02:06  <joaocgreis>let me check
20:02:24  * jgijoined
20:04:43  <joaocgreis>trott: it's waiting for build 158, that is running in test-rackspace-win2012r2-x64-2 . Since I have the other one down, that is the only machine for the vs2015 label
20:05:28  <joaocgreis>CI is going to be a bit slower today until I finish this, sorry
20:05:52  <joaocgreis>I'm trying to always have at least 1 machine for each label
20:07:53  <trott>Oh, I see. So if I try to run one with win2012r2 label and one with win-vs2015 label, I'm really just running the same test twice because they're both going to grab the same host?
20:11:03  <joaocgreis>that's true for win-vs2015, but win2012r2 also has the msft machines (I believe it chooses from the group of machines that it has with round robin)
20:11:34  <jbergstroem>trott: fyi all machines prefixed test- should be accessable with credentials in the test secret subfolder (and joao is adding the windows machines)
20:12:12  <joaocgreis>take a look at https://ci.nodejs.org/label/win-vs2015/ and https://ci.nodejs.org/label/win2012r2/ , those are the labels with a list of machines each one has
20:12:13  * jgiquit (Quit: jgi)
20:14:19  <trott>Great, thanks!
20:29:00  <jbergstroem>rvagg: did a nightly run? (and did the new smartos14 release work?)
20:29:50  <jbergstroem>nope
20:29:57  <jbergstroem>lacks ssh config
20:30:31  <rvagg>good to know, I'd like to push 5.2.0 toda
20:31:40  <jbergstroem>fixed
20:31:53  <jbergstroem>that thing needs to be improved
20:36:06  * trottquit (Ping timeout: 240 seconds)
20:37:22  <jbergstroem>btw guys, if you reprovision please remember to update to new access keys and naming convention. i use the naming convention to know whats been "taken care of"
20:43:45  <rvagg>jbergstroem: Will says we can kill iojs-docker-build-1`
20:44:00  <jbergstroem>rvagg: consider it gone
21:04:15  <jbergstroem>rvagg: so, passing tmp test dir - do I exit on missing or mkdirp?
21:04:29  <jbergstroem>im somewhere in the middle at this point
21:08:18  <jbergstroem>so, jenkins. whats up now? :( Warning: you have no plugins providing access control for builds, so falling back to legacy behavior of permitting any downstream builds to be triggered
21:09:42  <joaocgreis>jbergstroem: I'll try to take another look at your pr asap, sorry for not getting to it sooner
21:09:50  <jbergstroem>joaocgreis: no rush
21:10:03  <jbergstroem>joaocgreis: at the moment i think we're reduced to above question
21:10:31  <jbergstroem>did someething just happen do the test-pr job? https://ci.nodejs.org/job/node-test-pull-request/961/console
21:10:41  <jbergstroem>oh, never mind
21:11:21  <joaocgreis>for me it makes sense to mkdirp, the user has set a variable, probably wants it to exist. But otherwise the error is clear, so no harm
21:28:04  <jbergstroem>ok, herd looks back to normal now
21:28:12  <jbergstroem>joaocgreis: ok, i'll take the mkdirp approach then.
21:28:13  <rvagg>jbergstroem: re mkdirp I'm in the middle too, I just feel that the plain mkdirSync() is wrong, it's either throw or mkdirp
21:28:33  <jbergstroem>rvagg: yep you're right
21:28:42  <rvagg>jbergstroem: mkdirp seems too complicated to add tbh so I'd lean toward throwing I think
21:28:47  <jbergstroem>after this we can remove all node_common_pipe and look at running tests in parallel at bots
21:28:59  <rvagg>can't execSync(mkdir -p) either cause Windows, have to program it up
21:29:23  <jbergstroem>rvagg: haven't tried it but i think it'll be angry if i pass a known folder
21:29:36  <jbergstroem>..and jenkins times out :(
21:41:25  <jbergstroem>just failed to restart it a few times
21:41:33  <jbergstroem>someone needs coffee more than me.
21:42:16  <jbergstroem>some day i will just show you a fully working buildbot setup and everyone will forget about jenkins overnight :-D
22:00:44  * jgijoined
22:01:01  * trottjoined
22:09:15  <bnoordhuis>https://ci.nodejs.org/job/node-test-commit/1392/console <- ci trouble?
22:09:32  <bnoordhuis>ah no, sorry. merge conflict
22:10:56  * jgiquit (Quit: jgi)
22:45:24  <jbergstroem>so, jenkins is freezing again. wtf is up
22:46:17  <jbergstroem>i'll talk to the tap author
23:05:46  * jgijoined
23:18:58  * jgiquit (Quit: jgi)
23:19:19  * trottquit (Ping timeout: 250 seconds)
23:27:09  * trottjoined
23:32:52  * trottquit (Ping timeout: 272 seconds)
23:36:24  * trottjoined