=== kadams54-away is now known as kadams54 === kadams54 is now known as kadams54-away [01:48] Any way to change the default memory size of KVM guests when they're made? [01:48] Everything seems to be being created with 512M or RAM no matter what I do [01:50] X-Rob: what's creating your kvm guests? [01:51] sarnold: 'juju add-machine kvm:10 --constraints mem=8G' [01:51] which various places seem to imply should work [01:51] X-Rob: interesting, I hadn't seen the kvm:10 thing before :) thanks [01:52] https://bugs.launchpad.net/juju-core/+bug/1399613 [01:52] Bug #1399613: juju-core not using constraints when creating KVM unit on maas machine [01:52] X-Rob: hmm... almost certainly a bug [01:52] X-Rob: in that the requested constraints for the container aren't passed down to the host [01:52] That says it's my issue, but, it's linked to a totally different bug [01:52] um.. [01:53] I'm pretty sure that memory constraints were tested when I wrote it... [01:53] thumper: Well, any debugging I can do? [01:53] X-Rob: I agree that it should work [01:54] X-Rob: that bug looks like it describes your problem exactly, but the duplicate feels like a mistake to me [01:55] Ahha [01:55] OK [01:55] sarnold: yeah that was supported before lxc:# :) [01:55] lazyPower: it was?? why does'nt anyone tell me these things :) [01:55] sarnold: because we stink at writing changelogs? [01:56] looks like sinzui's not online, bugger, he might have had a good reason for duping them.. [01:56] lazyPower: yeah, I might really enjoy using a kvm: option.. :) [01:56] I can't seem to find where, if any, logs of this stuff would be [01:56] sarnold: whoa whoa buddy - lets not get carried away here. LXC is the hotness [01:57] X-Rob: I think those bits are in the machine-0 log, but i might be mistaken. They coudl be going to stdout/err and by proxy discarded. [01:57] LXC -is- the hotness, except, it doesn't work well with pacemaker and floating IP addresses. [01:57] lazyPower: nah, nothing in machine-0 [01:57] X-Rob: ah i was tongue in cheek'ing @ sarnold. [01:57] X-Rob: is your environment running? [01:58] thumper: eeyup [01:58] lazyPower: heh, I spin up a dozen kvms all the time, {i386,amd64}x{lucid,precise,trusty,utopic,vivid,wet}, etc... [01:58] lazyPower: containers are neat enough but not sufficient for those kinds of tests :) [01:59] thumper: are we capturing the lxc container / kvm creation logs? Its sounding like we arent. [01:59] in which case i'm more than happy to file a bug on it [01:59] X-Rob, lazyPower: ok, this is going to give you a lot of info... [02:00] you want to rack up the logging on a certain subset [02:00] so [02:00] Woo, I love logs. [02:00] get the existing logging by running 'juju get-env logging-config' [02:00] tell me what you have [02:00] =WARNING;unit=DEBUG [02:00] ^^ same [02:00] ok, do this: [02:01] juju set-env 'logging-config=juju.container.kvm=TRACE' [02:01] oh i love it [02:01] thumper: <3 can we put that in the docs? [02:01] this will capture all command line calls to the libvirt and their output [02:01] lazyPower: yes [02:01] WOO [02:02] yay docs! :) [02:02] lazyPower: juju.container.lxc does the same for the lxc commands [02:02] X-Rob: this will tell us whether the 8G mem constraint is being passed to libvirt or not [02:02] if it is, it is a libvirt problem [02:02] if it isn't, it is a juju problem [02:02] thumper: OK, lemme go do this thing. [02:03] https://www.irccloud.com/pastebin/z0Bi2JHZ [02:03] now.. for the logs.. [02:04] ... um. OK, WHICH logs? on 9? [02:04] yep, on 9 [02:04] 2015-01-06 02:03:27 TRACE juju.container.kvm libvirt.go:35 uvt-kvm [create --log-console-output --user-data /var/lib/juju/containers/juju-machine-9-kvm-7/cloud-init --bridge juju-br0 --memory 512 --cpu 1 --disk 8 juju-machine-9-kvm-7 release=trusty arch=amd64] [02:04] thumper: juju's fault. [02:05] X-Rob: add that to the bug please [02:05] I'll try to get someone to look at it [02:05] * thumper complains that it used to work [02:05] actually [02:05] not sure if I tested that particular use case [02:06] I tested kvm machines created for the local provider using memory constraints [02:06] probably never worked... [02:06] oops [02:06] my bad [02:06] thumper: the bits to reset the logging to default? [02:06] juju set-env 'logging-config==WARN;unit=DEBUG' ? [02:07] juju set-env 'logging-config==WARN' [02:07] stellar [02:07] Thank you [02:07] lazyPower: the unit=DEBUG is added unless you explicitly turn it down [02:07] this means that we always see hook output [02:07] right [02:07] you can turn it off by doing this: [02:07] juju set-env 'logging-config==WARN;unit=WARN' [02:07] that way your logs won't contain the hook output [02:08] lazyPower: levels are: TRACE, DEBUG, INFO, WARNING, ERROR [02:08] highly recommended that you have at least WARNING [02:08] and anything under WARNING is likely to be developer related debugging and logging [02:08] so not necessarily intelligible [02:09] the logging-config can be as long as you like [02:09] I like logs. [02:09] juju help logging [02:09] but.. not much use with this [02:09] LIKE A BOSS [02:09] https://github.com/juju/docs/pull/216 [02:10] OMGDOCSWTF [02:10] lazyPower: <3 [02:10] X-Rob: i'm like Tron - I fight for the users. [02:10] lazyPower: .. where do those logs go? [02:11] sarnold: in thumpers email, where they belong :P [02:11] all-machines.log [02:11] lazyPower: lol [02:11] which is looked at for 'juju debug-log' [02:11] Updating [02:12] https://i.imgur.com/ZPHgGeu.jpg [02:12] That's my test environment, by the way [02:13] X-Rob: is that sixteen "blades" with two storage boxes? [02:13] sarnold: yeah. [02:13] sarnold: whats your github user id? you're not sarnold @_@ [02:13] X-Rob: and is that a volleyball in the back? :) [02:13] sarnold: that's actually my gym 8) [02:13] lazyPower: "setharnold", I laughed at github and didn't move on it quickly enough... [02:14] https://i.imgur.com/gn1aoRw.jpg [02:14] "the whole point of git is that it is decentralized, who would want to put it all on a central server anyway?" ... heh [02:14] my exuse is that I don't use it anyway, and it's got aircond, AND I happen to have 2 15amp 240v circuits in there [02:14] X-Rob: sweet, complete with minisplit aircon! :) [02:15] (<-- au, we have 240v natively) [02:15] nice [02:15] Ebay, $2k [02:15] X-Rob: what are you testing for exactly? [02:16] X-Rob: wow, I thought the "au tax" would make that rig five times that... [02:16] thumper: Well, the long term result is this stuff is going to be an openstack build for my minecraft server (I run mcau.org) [02:16] sarnold: I did have to jump in my van and drive for 16 hours (each way) with 1 hour notice to my wife to go pick it up. The guy wanted it gone NOW [02:17] X-Rob: ooof :) [02:17] thumper: the short term is me learning all about openstack, because fuck vmware. I'm a VCP, but I now (officially) hate them. [02:17] man australia is a big place. [02:17] VCP? [02:18] my kids are now all into minecraft :) [02:18] https://www.google.com.au/maps/dir/Gladstone+QLD/Caringbah+South+NSW/@-28.8680946,146.9265686,6z/data=!3m1!4b1!4m13!4m12!1m5!1m1!1s0x6bc27489ada17a29:0x500eef17f210e60!2m2!1d151.2597998!2d-23.8487083!1m5!1m1!1s0x6b12c78d09dddc5b:0x5017d6816334960!2m2!1d151.1213599!2d-34.0551444 [02:18] thumper: vmware certified professional [02:18] I did the course and stuff [02:19] thumper: well, mcau will be MAAS+JUJU+OpenStack [02:19] nifty [02:22] sarnold: https://plus.google.com/100016143571682046224/posts/ZHG9ybVN3pR [02:22] lots of activity on the post here - seems like one person was already +1 on this idea [02:22] X-Rob: long-ass drive, but for that kind of hardware at that price, probably makes sense... [02:23] sarnold: yep, it was. It's kinda old, but, good enough for lots of small minecraft servers. [02:23] thumper: did you see the minecraft post I made lastyear? Let me refresh you :D https://plus.google.com/100016143571682046224/posts/PaVGh51FYCR [02:24] lazyPower: heh, I gave a +1 earlier.. is it mine? :) [02:24] sarnold: actually no - an indie security researcher [02:24] X-Rob: I'm eyeing some older server gear for home use myself, I'm more interested in a pile of storage than a bunch of vms though.. [02:24] aha [02:26] sarnold: it's really easy to get a pile of storage these days. ceph + a couple of cheap-arse motherboards with 6 or 8 SATA sockets. [02:28] X-Rob: I've thought a bit about going that route but think I'd rather get something like a supermicro or norco with 15~20 drive bays and do zfs instead... [02:29] X-Rob: a pal keeps suggesting getting a drive shelf instead and hook it up with external sas, but I think I'd rather avoid expanders .. [02:30] .. of course, I also worry about the noise of going that route. a few desktops running ceph would probably be quieter. [02:31] Ick, I';m with you there, external sas is nothing but heartache [02:31] sarnold: my original ceph cluster was two milk crates with motheboard|cardboard|pizza spacers|cardboard|sata disks [02:32] and I cabled tied the psus to the outside [02:32] hahaha [02:32] PROFESSIONAL AWW YES [02:32] seems legit [02:32] the pizza expanders made it even more legit [02:32] s/expanders/spacers/ [02:32] yeah I lost it at the pizza spacers :) [02:32] heheh [02:33] They were needed, they gave the cardboard better structural integrity so it didn't collapse all over the motherboard [02:33] plus also, kept the pizz from getting the cardboard messy in the first place, amirite? :) [02:35] heh. They weren't pizza boxes. That would have been the best idea, now I think about it [02:55] Hi, I have used juju to deploy some charms, my understanding is juju would cache charms, where can I find the cached charms? [02:56] catbus1: iirc, in ~/.juju/ there's a directory of cached charms.. [02:57] sarnold: I do see them in ~/.juju/, there is environments/, environments.yaml and ssh/. [02:58] I don't see a list of charms I deployed in those 2 directories. [02:58] catbus1: look under environments/, it's probably in a subdirectory named after the environment [02:58] sarnold: there is just maas.jenv in environments/. [02:59] catbus1: is this a local environment, or a cloud environment? [02:59] catbus1: if its ac loud environment,t hose charms are then cached on the stateserver [02:59] lazyPower: a local environment. [02:59] catbus1: then sarnold is correct, you should have a local charm cache in ~/.juju/local/ [03:00] thumper: has anymore thoughts/discussions happened wrt making the stateserver an honest container on the local provider? [03:00] lazyPower: yeah... I'm considering doing Fun Fridays :-) [03:00] lazyPower: got to get through MESS first [03:00] or at least happy it is going in the righ tway [03:01] lazyPower: I know what I want to do... [03:01] oh agreed, i wasnt pressuring by any means, just curious if we're still headed down that path in the future. [03:01] lazyPower: and it doesn't seem too hard [03:01] oh yes [03:01] I want it [03:01] and I want it soon [03:01] because to be honest, it sounds really compelling. And i'd love to see that be the case. [03:01] so I can test MESS and HA using a local environment [03:01] MESS? [03:01] haha [03:01] <3 [03:01] multi environment state server [03:01] OH! [03:01] mark hates it [03:01] the name that is [03:01] but I think it is cute [03:01] I'm just making a mess [03:02] i just laughed so hard i got light headed [03:02] haha [03:02] hehe [03:02] lazyPower: so, in something completely unrelated === negronjl is now known as negronjl-afk [03:02] * lazyPower smells a work item coming out of this aside [03:02] lazyPower: I want python-django to support django 1.7, virtual environments, and python 3 [03:02] called it [03:03] I've hacked my local python-django so I can deploy my django 1.7 app [03:03] and I wanted to use python 3 [03:03] but gave up [03:03] so I'm using 2.7 and django 1.7 [03:03] that should be fairly simple. Python3 exists in the repositories [03:03] you should be able to warden all of that with venv's [03:03] yeah, but python-django doesn't use virtual envs [03:03] lazyPower: catbus was incorrect, we're using juju via MAAS [03:04] bladernr`: ok - that makes more sense why it wasn't showing up in ~/.juju/local [03:04] not the local provider. so you're saing the charms would be cached on the bootstrap node? [03:04] bladernr`: the cached charms will be on your state server on the MAAS machine. [03:04] correct [03:04] thumper: are those charms getting stuffed in gridfs or are they on disk on the state server? this is where i get fuzzy [03:04] um... [03:04] deployed charms? [03:04] and wait what? django doesn't support venvs? [03:04] lazyPower: nope [03:05] hang on [03:05] as in the framework or the charm? [03:05] django does [03:05] the charm doesn [03:05] tt [03:05] ah ok, the charm could be modified i would think to be a simple task [03:05] inspect for a .venv file/dir, if so - route throught hat. using something like virtualenvwrapper [03:05] lazyPower: if you make it work, I'll test it [03:05] thumper: if you file bugs, i'll put it on the list [03:05] ha [03:05] fair call [03:06] forewarning though - i'm not a django dev - so i might make a bigger mess than MESS [03:06] so i imagine i'll be pestering you for pairing [03:06] lazyPower: thanks, /var/lib/juju/charmcache finally found it [03:06] bladernr`: awesome! glad you got it sorted :) [03:09] lazyPower: perhaps we can scratch each other's back... [03:09] lazyPower: I may submit a patch or two to python-django [03:09] thumper: go on... [03:09] mostly to clean it up [03:09] I'd like that, and i'm sure the maintainer would too [03:09] it is a bit messy, [03:09] yeah, last time i poked around in there i had @_@ these eyes [03:09] its one of those inhereted charms from when the store was the wild west [03:10] but if I just make multiple small patches [03:10] we're cleaning quite a bit of it up though - did you see the email that hit the last pre-break about the 30 or some odd charms slated for moving to ~unmaintained? [03:10] I'm a little frustrated that it doesn't support django 1.7 even though it has been released for 10 months [03:10] lazyPower: I don't follow the list too closely for those things [03:11] tsk tsk, but you could say the same to me for not diving through the -dev list for all your features that are landing. [03:11] okay, belated new years resolution - i'll follow juju-hackers 10% more closely than i currently do. [03:11] which will put my current level of monitoring at 10% [03:13] heh [03:13] Actually, lazyPower, 10% more of 0% is still 0%. [03:13] lazyPower: and one for me, you'll get a patch for python-django in the next few weeks [03:13] 1.1 * 0.0 == 0.0 ... success! [03:13] snap, sarnold [03:13] X-Rob: your message got to me first though, you beat me fair and square :) [03:14] So the company I work for has been bought by another company and they want me to use exchange. [03:14] sarnold X-Rob: touche - however 1.1 * 1 = 1.1 [03:14] * X-Rob grumbles. [03:15] reading juju-hackers list is kind of like checking the fridge, looking around for something because you know you'r ehungry - closing the door and coming back 15 minutes later to repeat the process. [03:15] you shuffle around the condiments a bit, figure there's nothing in there to munch on and order a pizza [03:15] X-Rob: suuuuuck [03:15] completely missing the cheese cake in the back [05:51] lazyPower: just saw you were talking about the python-django charm... and missed hte conversation. [05:52] lazyPower: will the pure-python branch be merged soon? the ci tests and stuff got cleaned up [05:52] I have some patches I'd like submit once that goes [12:52] lazyPower: what's the deal with this? https://bugs.launchpad.net/charms/+bug/1373006 [12:52] Bug #1373006: Hortonworks hdp 2.1 Storm+ZK bundle === fuzzy_ is now known as Fuzai === roadmr is now known as roadmr_afk === kadams54 is now known as kadams54-away === kadams54-away is now known as kadams54 === roadmr_afk is now known as roadmr === kadams54 is now known as kadams54-away === kadams54-away is now known as kadams54 [15:39] any chance of getting pure-python python-django MR pushed soon? I would like to start putting together some MR and that has a lot of changes in it [15:40] I'll be in merge hell if I don't wait [15:41] skay: I'm not sure the state of the review, but you're welcome to do a review of it yourself if it's waiting for more eyes [15:42] marcoceppi: I am relatively new to juju and not sure my review would be very insightful [15:43] skay: well, we always welcome community reviewers. If you're looking to get mroe in to charming it's a great way to look at what other people are tyring to do. If you find something comment on it! If you don't that's fine too, it'll still go through the review of a charmer before making it's way to the store [15:43] it's been reviewed by some charmers already and has passed hte ci tests [15:44] the commiter made fixes based on the review cmments [15:44] sounds like it's just stuck waiting finial approval then [15:44] I'll see if I can take a look at it today, the queue is mighty big these days :\ [15:44] is there a way for me to tell? [15:45] at this point, I would like to provide suggestions by submitting some small merge requests rather than holding up the MR [15:45] http://review.juju.solutions/ [15:45] is where it is in the queue, it's pretty close to the top now [15:46] thanks [15:46] yeah, this is just waiting for a merge, I'll take a few mins to finalize and merge it today [15:46] \o/ [15:48] btw, I'm getting an error when I try to login to the site [15:49] skay: what's the error? [15:50] marcoceppi: The server encountered an unexpected internal server error [15:50] odd... I was able to log in just fine [15:50] it hates me [15:50] * skay sobs [15:51] myself as well, I wonder if this is because of some weird issue with code [15:51] skay: did you tell it your name? (the openid interstitial that lets you choose what to share) [15:51] I was curious if I could log in and trigger a new jenkins job [15:51] skay: ah, you can't, that's reserved only for charmers [15:51] roadmr: yep, after it fafiled I went and gave it all the permissions [15:51] I noticed nothing was ticked, so I ticked everything before going on [15:51] skay: ok, so that's not it [15:51] the internal server error is just the reviewq not having much good code around login since it was kind of a last min thought [15:52] skay: try again? [15:53] yeah, I see why it's failing [15:53] it's a known issue [15:53] okiedokie. not urgent [15:54] skay: once you get a review in the review queue log in will start working for you === negronjl-afk is now known as negronjl === designate is now known as designated [17:15] hey tvansteenburgh, have a min for a PM? [17:15] jose: yup [17:19] Anyone know if a lint with warnings will stop a charm getting promulgated into the store? [17:21] wesleymason: a warning, yes. though depends on the warning [17:21] which are you getting? === kadams54 is now known as kadams54-away === roadmr is now known as roadmr_afk === kadams54-away is now known as kadams54 [18:03] cory_fu, what's the closest thing you have to a homepage for DHX? [18:03] jcastro: I guess the blog post: http://blog.juju.solutions/cloud/juju/2014/11/26/debug-hooks-ext-plugin.html [18:04] Why do you ask? === kadams54 is now known as kadams54-away [18:04] I am giving a presentation to the SA guys in like 2 weeks [18:04] and I want to go over DHX [18:04] Ah, cool [18:05] because it seems they would benefit a bunch from knowing about it === roadmr_afk is now known as roadmr === makyo_ is now known as Makyo [19:17] hey, someone is giving a juju talk at lcaconf. my spouse is on the video team for the conf [19:18] I wish I could be there. I want to go to lcaconf one day [19:18] this year's is in Auckland, which would be so cool to visit [19:53] lca is pretty awesome === kadams54 is now known as kadams54-away === kadams54-away is now known as kadams54 [21:19] marcoceppi: start/stop hooks missing, it's a subordinate with a binary called on demand, so there's no service to start/stop..I know I can just add blank hooks, but feels messy given they're unnecessary. === kadams54 is now known as kadams54-away [21:20] that hasnt been a warning in s long time wesleymason [21:20] wesleymason: do you have the latest charm-tools from ppa:juju/stable ? [21:22] marcoceppi: nah, from universe [21:23] wesleymason: yeah, sadly that version of charm-tools is a whole rewrite behind [21:23] we'll have it updated in vivid and in releases going forward [21:23] * wesleymason ignores the warnings then [21:23] until then ppa:juju/stable is the best place to get the most recent version [21:23] marcoceppi: ta [21:24] wesleymason: you can also install it in a python virtualenv from pip if you're so inclined [21:24] marcoceppi: avoiding using latest until juju is upgraded in prod by IS [21:24] wesleymason: ack, I can make a charm-tools only ppa if that helps [21:25] marcoceppi: tbf I could just pin it, might just install in the virtualenv anyway as using one for tests === kadams54-away is now known as kadams54 [21:25] wesleymason: virtualenv works pretty well and is the least amount of effort betweent the two of us [21:28] marcoceppi: ta muchly [21:28] np === kadams54 is now known as kadams54-away === kadams54 is now known as kadams54-away === kadams54-away is now known as kadams54 [23:19] lazyPower: https://bugs.launchpad.net/charms/+source/haproxy/+bug/1407815 - I finally tracked down the cause of that relation-changed bug I mentioned yesterday [23:19] Bug #1407815: adding a 2nd website unit fails to call reverseproxy-relation-changed [23:19] noise][: looking [23:20] lazyPower: now I need to go look through the code to figure out __why__ [23:20] noise][: weird, byproduct of a debug setting? [23:20] right?? [23:20] that smells === kadams54 is now known as kadams54-away [23:20] nearly drove me insane [23:20] * lazyPower identifys a code smell [23:20] noise][: well i'm glad you've got more info to go on. I'm subbing to this bug to monitor your progress [23:20] if you need any help feel free to reach out. I'm not an expert with the charm but i can lend a second pair of eyes [23:21] i was just about to chuck my computer in the bin! [23:21] lazyPower: cool, shouldn't be too hard to find the root cause now