[00:51] <veebers> wallyworld: any joy finding what's making bootstrap suck atm?
[00:51] <wallyworld> it's getting hung setting up the agent config, not sure why yet
[00:52] <veebers> ugh, hopefully it'll become apparent
[01:04] <wallyworld> veebers: ah, in my case it was my own fault. i had introduced a lock conflict, hence the hang
[01:04] <veebers> wallyworld: oh, so no idea why mines taking *ages*. might try a different region for now
[01:07] <wallyworld> it's probaby the apt updates
[01:17] <anastasiamac> veebers: unless it true style this is the case of 'great minds think alike'...
[01:18] <anastasiamac> in*
[01:52] <thumper> veebers: ping
[01:52] <veebers> thumper: pong, what's the haps?
[01:52] <thumper> veebers: http://ci.jujucharms.com/job/github-merge-juju/1072/consoleFull
[01:53]  * veebers looks
[01:53] <thumper> veebers: last two jobs have been killed from the outside
[01:55] <veebers> thumper: it says it was aborted by user 'developer'
[01:55] <thumper> ah
[01:56] <thumper> I love this test failure line:
[01:56] <thumper>     c.Assert(sInfo.Since, gc.NotNil) // TODO(dfc) WTF does this check do ? separation of concerns much
[01:56] <thumper> mine failed with that
[01:57] <thumper> I thought I had fixed the intermittent all watcher tests already
[01:58] <veebers> thumper: when you say "yours failed with that" you mean the 2 jobs you linked me just now?
[01:58] <thumper> no, the one I linked was anastasiamac's
[01:58] <thumper> mine was the job after that
[01:59] <thumper> I saw that failure in someone else's merge job too
[01:59] <thumper> I'll look at it after this interview
[01:59] <veebers> ah ok, it's possibly that anastasiamac aborted those 2 jobs on purpose (well, I hope so otherwise someone else is doing it)
[02:47] <anastasiamac> yes, i have aborted on purpose coz i saw failures
[04:01] <babbageclunk> thumper: have you got a moment?
[04:02] <thumper> babbageclunk: yeah, but I may lose power soon briefly
[04:02] <thumper> electrician in the power box
[04:02] <thumper> whazzup?
[04:02] <babbageclunk> in 1:1?
[04:03] <thumper> sure
[04:27] <veebers> huh, I saw this locally, perhaps it's not connectivity in the labs: grouped write of manifest, lock and vendor: error while writing out vendor tree: failed to write dep tree: failed to export golang.org/x/time: unable to deduce repository and source type for "golang.org/x/time": unable to read metadata: go-import metadata not found
[04:40] <veebers> wallyworld: if I deploy cs:~wallyworld/mariadb-k8s without any --storage args, should it succeed? (without any storage bits etc.) At the mom I'm getting a manifold worker error complaining about not being able to create persistent storage
[04:42] <wallyworld> veebers: no
[04:42] <wallyworld> the charm declares it needs storage in its metadata.yaml
[04:43] <veebers> wallyworld: ok, if failure is expected that is ok. How do I fix the deploy if I missed the --storage bit off the original deploy command?
[04:43] <veebers> oh, should it have errored at the deploy command stage?
[04:43] <wallyworld> delete the app and start again
[04:43] <wallyworld> it depends if the metadata said 1+
[04:44] <wallyworld> if the metadata didn't say one or more then juju won't count that as an error
[04:45] <wallyworld> without storage args you may have expected it to work, just with ephemeral storage. that should probably be fixed. but it really makes no sense to allow that for prod
[04:45] <veebers> wallyworld: ack ok, thanks for clarifying
[04:45] <wallyworld> it's a bit rough atm
[04:45] <wallyworld> needs polish
[04:46] <veebers> wallyworld: ack, we'll get it polished!
[05:10] <thumper> babbageclunk: fixed one, broke 20
[05:11] <thumper> well 5
[05:11] <thumper> FFS
[05:12]  * thumper calls it a day
[08:52] <Ablu> Hi, does anyone have an idea regarding my question from the weekend?:
[08:52] <Ablu> I try to create a "manual" cloud and then add a machine which actually is a docker container. However, after the add-machine exits the machine is stuck in pending and I also do not see any process running within the docker container... Are there any specific requirements regarding how the machine which is about to be added needs to be configured? (https://docs.jujucharms.com/2.4/en/clouds-manual does not help a lot as soon things
[08:52] <Ablu> go wrong unfortunately)
[08:56] <stickupkid> Ablu: what's running in the docker container?
[08:58] <Ablu> stickupkid: only sshd and the shell i use for inspection after the add was successful
[09:17] <manadart> Ablu: Juju needs to run agents in the machine.  It assumes there is an init system running. Any reason not to use LXD?
[09:20] <stickupkid> You could run ubuntu image, but then you'd have to wonder how you could keep it not from restarting, because it's not got any INIT running and the container will "exit"
[09:23] <Ablu> manadart: I want to set specific networking and resource limiting on the container (it is part of a larger emulation framework) and I am looking for ways to do this with juju charms which are used in the software which should be "emulated". But the init system is a good hint. Thanks for that.
[09:23] <Ablu> stickupkid: I currently use a ubuntu + SSHD, but it probably lacks an init system, thus the restart maybe fails.
[09:26] <Ablu> While on it: If you have better suggestions how I might be able to set IP namespaces with custom links, resource restrictions on juju maintained machines (ideally before any code is executed in the machine) feel free to tell me :). The framework I use can already do this for Docker, so if I can get juju to use "machines" which I already started "manually" with the right configuration I could do all that :)
[09:29] <manadart> Ablu: You might be interested in the CAAS support (Kubernetes) that is maturing in edge, see https://discourse.jujucharms.com/t/getting-started/152 and other posts in the category.
[09:42] <Ablu> manadart: hm, that looks like it would start the services in a k8s cluster. Where would I be able to set my own constraints before the container actually goes live?
[09:51] <manadart> Ablu: The ones actively working on the feature can shed more light on that than I.
[09:52] <manadart> They are Australian east coast, so EoD. I suggest asking a question on Discourse. They will attend to it smartly if not around here.
[09:53] <Ablu> manadart: Ok. Thanks for the info! I will try to explain my problem in a bit greater detail and do a post.
[09:53] <manadart> Ablu: Sure thing.
[13:48] <jam> stickupkid: should a negative size be capped at 0 or just treated as error ?
[13:48] <jam> stickupkid: I'm now thinking Error
[13:49] <stickupkid> jam: I would agree, a silent 0 could mask more issues
[14:09] <stickupkid> hml: any chance of a HO?
[14:09] <hml> stickupkid: sure, standup?
[14:09] <hml> stickupkid: if it’s free
[14:09] <stickupkid> yeap
[15:00] <manadart> externalreality_: Pushed the mods we discussed to https://github.com/juju/juju/pull/9157 .
[15:01] <externalreality_> manadart, ack
[15:01] <externalreality_> will take a look, give me 15 to make my way to the buss
[15:02] <manadart> externalreality_: No rush on my account. Wrapping it up for the day.
[15:02] <externalreality_> manadart, ack
[17:23] <hml> anyone up for a 1 char PR?  :-D  https://github.com/juju/juju/pull/9160
[20:32] <veebers> Morning o/