=== shardy_afk is now known as shardy === shardy is now known as shardy_afk === shardy_afk is now known as shardy === shardy is now known as shardy_afk === shardy_afk is now known as shardy [16:00] hey cloud-init, meeting will probably start a couple mins late [16:01] ok [16:02] blackboxsw, how do i make the bot start logging ? [16:03] o/ [16:03] # meetingology startmeeting [16:03] https://wiki.ubuntu.com/meetingology [16:03] # #startmeeting cloud-init status meeting [16:04] #startmeeting cloud-init status meeting [16:04] Meeting started Mon Sep 18 16:04:28 2017 UTC. The chair is smoser. Information about MeetBot at http://wiki.ubuntu.com/meetingology. [16:04] Available commands: action commands idea info link nick [16:05] hiya o/ [16:05] o/ [16:05] #link https://public.etherpad-mozilla.org/p/cloud-init-meeting [16:05] hello [16:06] so that contains a agenda that we'd been following [16:06] #link https://lists.launchpad.net/cloud-init/msg00100.html [16:07] sorry for slow going here. === smoser changed the topic of #cloud-init to: Release Preparation [16:08] shoot [16:08] #topic Release Preparation [16:08] #subtopic Release Preparation [16:08] hm.. [16:09] anyway [16:09] https://lists.launchpad.net/cloud-init/msg00100.html [16:09] that is the primariy point of this meeting for today. [16:09] The goal is to make a release named 17.1 on Thursday of this week. [16:09] strange [16:09] We have landed several branches in the last few days, and have 2 more to land at least. [16:10] from http://bit.ly/ci-reviews [16:10] meetingology │ Available commands: action commands idea info link nick [16:10] something to look into. :) [16:10] #link http://bit.ly/ci-reviews [16:10] the two we consider needs to be integrated are [16:10] https://code.launchpad.net/~chad.smith/cloud-init/+git/cloud-init/+merge/330875 [16:10] https://code.launchpad.net/~smoser/cloud-init/+git/cloud-init/+merge/330880 [16:11] and Chad's (the first) is the one we're looking at now. [16:11] or that is the one that is non-trivial [16:11] does anyone here wnat to raise other bugs or MP that they believe should be pulled ? [16:12] ... [16:12] ajorg, i think its best to hold off on some of yours until after release. [16:13] anyone have anything else ? [16:13] msaika (vmware) pinged on a branch a couple days ago. but I think there is a little of discussion/re-work that's needed there before we land [16:13] smoser: I assume you mean the instance identity one, in which case I agree. [16:13] (in either case I'd agree) [16:13] and simpletable [16:13] yeah [16:14] ok. so that is where we are. we are still targettin a release on Thursday. [16:14] other than that, i thinkw e move on to open discussion or office hours [16:14] #topic Open Discussion [16:14] anyone have anything for this ? [16:14] Customer has run into an issue where we run out of threads during the user script phase, should generally set TasksMax in the service file, i.e. is that of interest upstream? [16:15] we are still testing, but that appears to be the solution for this particular problem [16:16] robjo, i dont have an immediate objection to that idea. [16:16] could you open a bug ? [16:16] sure [16:17] i dont think that we'd want to pull that change in right now, but it otherwise seems reasonable at frist b lush. [16:18] anything else ? [16:18] Agreed, do't think this is sufficiently urgent for Thursday's release, it can wait until the next one [16:19] were there other bugs that we should consider high enough prio to deserve landing a fix in the release? [16:20] ajorg, the only 2 that i'm aware of are linked in those mp. i can dig the numbers [16:21] * bug 1717598 [16:21] bug 1717598 in cloud-init (Ubuntu) "Traceback when passing user-data on GCE" [High,In progress] https://launchpad.net/bugs/1717598 [16:21] * bug 1717627 [16:21] bug 1717627 in cloud-init "permission denied when executing dhclient in Ec2 datasource" [High,In progress] https://launchpad.net/bugs/1717627 [16:21] ok [16:22] Anything else? other wise i'll call this meeting done and hang around for office hours for the next 30 m inutes at least. [16:22] thanks for attending / feedback, robjo and ajorg [16:23] #link https://launchpad.net/bugs/1717598 [16:23] Ubuntu bug 1717598 in cloud-init (Ubuntu) "Traceback when passing user-data on GCE" [High,In progress] [16:24] #linkhttps://launchpad.net/bugs/1717627 [16:24] Ubuntu bug 1717627 in cloud-init "permission denied when executing dhclient in Ec2 datasource" [High,In progress] [16:24] just in case [16:24] duno whats up with the bot.... again [16:27] ok. well, lets move on [16:27] #endmeeting [16:27] Meeting ended Mon Sep 18 16:27:32 2017 UTC. [16:27] Minutes: http://ubottu.com/meetingology/logs/cloud-init/2017/cloud-init.2017-09-18-16.04.moin.txt [16:27] office hours now if you have anything [16:28] blackboxsw: bot looks ok, but... it doesn't understand topics [16:28] blackboxsw: it never echos on links or infos, that is normal [16:28] I'm playing with making simpletable output look like prettytable output, even though I think simpletable is prettier [16:29] published minutes will show up at https://cloud-init.github.io/from now on [16:29] heh ajorg nice [16:30] ajorg: lol [16:33] smoser: I did want you to take another look at the instance identity thing. I think you had your resolution comments backwards [16:36] ajorg, ok. [16:36] i have sense looked at that and the dns is wierd. [16:36] $ host why-does-this-exist.ec2.archive.ubuntu.com [16:36] why-does-this-exist.ec2.archive.ubuntu.com has address 91.189.92.201 [16:36] why-does-this-exist.ec2.archive.ubuntu.com has IPv6 address 2001:67c:1360:8c01::19 [16:37] i think you showed (and i see elsewhere) [16:37] $ host why-does-this-exist.ec2.archive.ubuntu.com [16:37] Host why-does-this-exist.ec2.archive.ubuntu.com not found: 3(NXDOMAIN) [16:37] meaning it's different depending on where we're calling from? [16:38] yeah. somewhat. but mostly i was just giving you bad information [16:38] i had a canonical VPN up. [16:38] and was getting a dns response over that . i just figured this out... shut that down and i get the does not exist [16:39] smoser: you route everything through the VPN right ? [16:39] not everything. [16:39] but apparently dns was going there for at lease *canonical [16:39] er.. *ubuntu [16:39] strange, I know I'm on a diff setup; but same vpn, and I've never had that host resolve; it always failed to resolve for me (with vpn up) [16:40] ah. so in The Real World it probably works the way I think it does? [16:40] ajorg, right. [16:40] at least for now :) [16:41] rharper, i'm also using artful. so its possible network manager tied better into systemd-networkd or osmething and got *.ubuntu.com passed to vpn dns [16:41] i dont know [16:41] yeah; on an internal system, I can see the wildcard working as you saw [16:44] ajorg, so we have more thinking to do there. but i unfortunately can't prioritize said thinking right now. [16:45] okay. It can wait. I do hope you'll get back on it soon, because it's one of the more important patches we're carrying. [17:09] well, that wasn't too hard: http://paste.ubuntu.com/25566740/ [17:11] I'll write a unit test or three this week and resubmit that with PrettyTable ripped out completely. [17:11] Do I recall correctly that we're encouraging putting tests next to modules now? [17:16] ajorg: yes please [17:17] thanks, will do [17:17] if there isn't a tests/ subdir next to your module just add one. post this release I think we'll overhaul the tests in one mega branch and put them all in the 'proper' location [17:21] cool [18:01] smoser: commented on https://code.launchpad.net/~smoser/cloud-init/+git/cloud-init/+merge/330880 [18:01] approve with a nit [18:04] * blackboxsw thought about adding a more significant unit test for main, but I'd ultimately like that to go away anyway with a standardized datasource.crawl_metadata() method (and cloud-init devel crawl subcommand) [18:04] so I figure why spend much time on covering that main() method w/ tests [18:04] since I'd like to see it dropped in a couple weeks [18:05] the mains are debug only anyway [18:05] devloper tools [18:05] i took your suggested test cahnge [18:05] and committed and pushed [18:05] runnign tox here [18:22] +1 [18:41] Hi [18:42] anyone here? [18:43] * dpb1 hides [18:55] dpb1: huh. maybe you shouldn't have hid [18:55] LOL [18:56] smoser: Re https://code.launchpad.net/~paul-meyer/cloud-init/+git/cloud-init/+merge/330925 [18:57] From history we know that some infrastructure updates take at least 5 minutes and may induce transitional failures in the wire server... [18:57] Of course, that's unacceptable in the provisioning path, which we'll work on separately... [18:58] I'd like to set the timeout at 900 to be on the safe side of 300+... [18:58] Is that within the realm of possibilities? [19:03] hi, can anyone take a look at my review request. https://code.launchpad.net/~msaikia/cloud-init/+git/cloud-init/+merge/330105 [19:03] hey folks [19:03] can I get help for cloudbase-init here? [19:04] I'm trying to kick off a jenkins jnlp windows slave, but can't seem to get the user-data I specify with Jenkins' Cloud plugin to execute [19:15] cj: do you have access to the logs to see if your user-data had any issues? [19:18] I think I do. I'm pretty sure I specified where the logs should live... [19:19] "msiexec /i C:\\packer\\CloudbaseInitSetup_Stable_x64.msi /qn /l*v c:\\packer\\cbi-log.txt" [19:19] from line 72 of https://gerrit.iotivity.org/gerrit/#/c/21639/41/packer/templates/baseline-win.json [19:19] or is this just the installation log? [19:20] cj: tbh not sure, as I am not 100% certain 1) how much different cloudbase init is from cloud-init and 2) how things get run on windows ;) [19:27] paulmey, i'm willing to defer to you. [19:29] Alrighty, mp updated [19:30] cj: well [19:31] blackboxsw, subp_blob_in_tempfile will need updating [19:31] cj: it's not in tree, and not supported by people here directly. [19:31] to /var/tmp/ [19:32] smoser: yeah I wasn't sure if we wanted that in this branch or not. We could just change temp_utils.tempdir() [19:32] then we don't have to change it in all call-sites in the future [19:33] or put a define somewhere. [19:33] TEMP_EXE_DIR [19:33] or something [19:33] and then use that from both? [19:36] why would we not want to use it in all cases? [19:36] _ROOT_TMPDIR == '/var/tmp' [19:36] one place to fix, instead of the caller having to know that they want exec perms [19:36] cj: I'd say look here: https://ask.cloudbase.it/questions/, it seems to be their preferred method of Q/A [19:37] blackboxsw, i'm not opposed... but it seems good to identify the places that do need exe [19:37] but yeah, you're right in that we'd need also a non root value [19:37] ? [19:37] oops typo [19:38] smoser: is it worth a param to temp_dir(exec=True) ? [19:38] which then gets sorted [19:38] another option would be to have a kwarg 'needs_exe' or something. [19:38] yea [19:39] I'm all for needs_exe, it'll make it explicit for the caller. and they won't have to deal w/ imports of TEMP_EXE_DIR etc [19:39] if you can do that in short term, next hour or so i'm good with that . but for the moment i'm also ok with everything being /var/tmp. assuming we're reasonably confident that /var/tmp isnt going to be cleaned on boot [19:40] thanks, dpb1 [19:45] smoser: shall we put unpriveleged users of temp_dir(needs_exec=True) in /var/tmp/cloud-init or just root [19:46] blackboxsw, i think its fine to put unpriv users tehre. we do set the perms so they can , i think. on creation [19:46] os.chmod(tdir, 0o1777) [19:46] yeah [19:47] ok will do [19:47] I only need another 20mins I think on this. [19:47] will have a fix [19:48] adding a couple unit tests now [19:48] k. thanks. [20:29] blackboxsw, what do you think about https://code.launchpad.net/~paul-meyer/cloud-init/+git/cloud-init/+merge/330925 [20:32] rharper: remind me again where vendor data comes from? [20:32] the cloud metadata service [20:32] via the cloud-specific data source? [20:32] cloud-init fetches 'user-data' key from the metadata service, it will/can also look for 'vendor-data' ; yeah inside the DataSource [20:33] ok, I don't think Azure currently has that... [20:33] but it would be good to think about that... [20:34] azure does not provide vendor data at the moment. [20:34] yeah [20:34] I'll put it on my wish list [20:38] smoser: I guess we're dropping the maxwait default in the Azure data source, so it's a required param now per paul's branch. I only see two callers of that function, so sure. I'm +1 https://code.launchpad.net/~paul-meyer/cloud-init/+git/cloud-init/+merge/330925 [20:40] It'd be nice if we had a way to introspect the walinux agent health on this instead waiting. But, maybe that's the approach we take for now and we improve upon it as time permits [20:40] it wouldn't look very healthy... [20:40] lol [20:40] the health check would block forever ? [20:40] showing 503/404 for requests to the wireserver/metadata server [20:41] or just timing out there... [20:41] I suppose that's what paulmey means by "I'm working on a fix for this issue." :) [20:41] the issue is on the host [20:41] :-) sure [20:42] I should probably have said "I have other teams working on a fix for this issue" [20:42] hah [20:43] silent provisioning failures lead to customer support calls [20:43] long provisioning times and failures just yield bad rep [20:43] s/bad rep/bad bug reports/ [20:43] or both [20:43] yeah [20:44] bad rep leads to mean tweets [20:44] mean tweets lead to executive calls [20:44] and mean tweets lead to smoser losing sleep [20:44] lol [20:45] so paulmey if we block waiting on certs and does walinux agent ever fix itself in these cases? [20:46] Yes, it should. Eventually the wire server should be up again. [20:46] hrm, we could add a smoser reads mean tweets to the cloud-init meeting agenda [20:46] This situation mainly occurs when the wireserver itself is being updated [20:47] rharper, once for uds or uos (online) we had the security team read mean tweets [20:47] hehe [20:50] https://youtu.be/4fB94_DndL8 [21:02] blackboxsw, https://code.launchpad.net/~chad.smith/cloud-init/+git/cloud-init/+merge/330875 [21:02] any status ? [21:02] i've got to be running. [21:02] what i'll do right now is upload trunk to artful [21:02] that will get us the gce fix [21:02] and then i'll come back in ~ 4 hours and if you have merged paulmey and your branch i'll grab and upload those too [21:06] smoser: yep just finished a patch [21:06] took a while to setup the unit tests [21:07] wrap_and_call issues on my sid [21:33] finally pushed to the right branch https://git.launchpad.net/~chad.smith/cloud-init/commit/?id=a3a33b43c3c62b82c30050a78c69a11a5eca8c40 [21:34] will await CI and run an AWS deploy test on this latest [22:10] paulmey: merged your branch and updated the commit message. it should be in our next release (which we are cutting this week) [22:10] https://code.launchpad.net/~paul-meyer/cloud-init/+git/cloud-init/+merge/330925 [22:10] thanks again [22:11] blackboxsw: thanks!