=== liam_ is now known as Guest89217 === Spads_ is now known as Spads === CyberJacob is now known as CyberJacob|Away === CyberJacob|Away is now known as CyberJacob === CyberJacob is now known as CyberJacob|Away [03:47] I was wanting to experiment adding egress rules to all units in an environment. I can't do that via constraints or similar juju functionality right? (I'm assuming I'll need to update the default secgroup for the env, which will only work when deploying to openstack/aws, not local) === noodles785 is now known as noodles775 === Spads_ is now known as Spads [04:11] noodles775: you could add a subordinate, I've thought of doing that [04:11] noodles775: having a egress filter subordinate that get attched in dev/qa to create firewall rules manually === kadams54 is now known as kadams54-away [06:46] rick_h_: yeah - I was hoping to do it outside the units themselves (ie. secgroups), but wanted to check whether it's something juju manages (I'm assuming not). Thanks! [06:47] noodles775: I don't think so atm === CyberJacob|Away is now known as CyberJacob === CyberJacob is now known as CyberJacob|Away === natefinch is now known as nate-tablet === roadmr is now known as roadmr_afk === enterprisedc_ is now known as enterprisedc === kadams54 is now known as kadams54-away === roadmr_afk is now known as roadmr === liam_ is now known as Guest73106 [15:30] I think the charm helper docs for charm create are out of sync with the package, but could use a sanity check. here's what results from following the docs, http://paste.ubuntu.com/9125147/ [15:30] expected behavior should be http://pythonhosted.org/charmhelpers/getting-started.html#creating-a-new-charm right? [15:35] skay: too many "create" in your commands [15:36] skay: are you looking at using the services framework as well [15:36] the python-basic template is more likely what you're looking for. [15:36] lazyPower: when I tried to use the -t to specify a template, I got a usage error [15:37] skay you have "charm create create" [15:46] tvansteenburgh: woohoo, I like it when it's a typo instead of a complete crazy misunderstanding [15:46] :) [15:47] lazyPower: python-basic is indeed what I was looking for (it also looks like the docs should indicate python-basic for hte template rather than python?) [15:47] thanks all [15:48] skay: It's a bit confusing I agree - python being an alias for the services framework. [15:49] lazyPower: for feedback: when I use the python template the dir tree does not resemble the dir tree on the docs. [15:50] * tvansteenburgh makes a note to update docs [15:50] thanks for pointing that out skay [15:50] tvansteenburgh: I wasn't sure if I should open an issue in case I had misunderstood something. thanks for helping. [15:51] sure thing [15:54] avoine: adding support for pip_extra_args in order to use --no-index --find-links=/some/path, isn't as straight forward as I thought due to pip not being able to find /some/path. [15:55] avoine: I am not sure exactly what to do, and maybe I need to write a subordinate charm to populate /some/path on the python-django unit? [15:55] because I don't think that functionality belongs in python-django [15:56] I think adding support for pip_extra_args belongs, and python-django would just pass that to pip [15:56] and it would be the responsibility of something else to ensure the path is reachable [16:21] skay: Have you tried to upgrade pip? [16:23] avoine: I did try, that did not resolve the problem. (and also would be a problem if one applies pip_extra_args to all calls, if it is the case that the existing pip doesn't support the extra args, but while I investigated I introduced a temp kluge to ignore it when installing pip) [16:31] skay: just to be sure I understand, your installing a tar archive with your django project + your dependencies and then your pointing pip ... --find-links to your dependencies dir, right? [16:32] also pip don't need to be upgrade on trusty since "Added support for bare paths (including relative paths) as argument to –find-links" was added in version 1.3 and trusty is 1.5.4 [16:32] avoine: not exactly. I tarred up my django project and had dependencies in the local file system. oh derp, I should just tar them up too? [16:32] avoine: I'm using precise at the moment (I thought the gunicorn charm was precise, but if I'm mistaken I can use trusty) [16:34] gunicorn is in trusty too [16:34] skay: how do you add your dependencies, with scp? [16:35] –find-links needs a dir where dependencies are already downloaded [16:35] avoine: no, I ignorantly thought that juju would be able to provide access to them in the local file system [16:35] oh like vagrant does [16:35] ? [16:35] yeah [16:35] no indeed it's not [16:35] I would put them in the tar [16:36] yes, it is clear to me now [16:36] this is simpler. [16:36] thanks for walking me through that [16:36] but I think that's a feature they want to add [16:36] that would be usefull [16:38] lazyPower: mounting local directory inside a local deployement (with nfs for example) is still a feature you wanted to add to juju? [16:39] avoine: we found a work around as documented in the NFS charm [16:39] cool [16:39] skay: ^ [16:39] o/ [16:39] it requires some manual intervention - so adding something without the manual intervention would be nice - but i can see it as being lower priority [16:40] skay: https://jujucharms.com/nfs/trusty/0 - take a look at "On the LXC host" in the readme [16:41] * skay takes a look [16:42] I guess that would be a good use case for a juju plugin [16:52] lazyPower, avoine this sounds like what Juju Actions is intended for [16:53] jw4: i dont think actions will be able to modify the HOST on behalf of the user. [16:53] those particular edits take place on the LXC host so the containers can receive the nfs kernel modules - otherwise apparmor blocks them [16:54] lazyPower: I see - and actions run in the context of the units [16:54] correct [16:55] actions are intended to be helpers that dont necessarily belong in the hooks - like database backups, adding users to an app, etc. [16:55] lazyPower: oh well. I got excited for a minute there === kadams54_ is now known as kadams54-away === utlemming_kitche is now known as utlemming_mobile === kadams54-away is now known as kadams54_ [17:19] I forgot, is there a way I can get the data of a relation outside a hook or with a juju command? [17:31] juju-run on the vm did the trick === kadams54_ is now known as kadams54-away === kadams54-away is now known as kadams54_ [18:41] hi juju team, can anyone help with a juju hang problem? [18:47] mwenning: hey o/ whats going on with juju? [18:49] lazyPower, hi, I'm running juju bootstrap --debug -e maas and it hangs for 10 min and then fails with a timeout [18:50] mwenning: are the units still spinning up? iirc you're working in a pure hardware environment [18:50] I'm pretty sure I didn't set the proxy right, but it won't take the right value anymore. [18:51] juju destroy-environment --debug --force maas also hangs [18:51] stby I'll get a pastebin [18:51] mwenning: is it still giving the 502? [18:51] yes. [18:52] huh [18:52] after I set the proxy correctly it worked for awhile, now it's hung again [18:52] s/hung/hanging [18:54] https://pastebin.canonical.com/120815/ [18:55] lazyPower, jose, it hangs at line 9 for probably 10 min, then dumps the rest of that stuff. [18:56] mwenning: lazyPower will give you a hand with this - cannot access canonical services and I wouldn't risk pasting any private details on a public site [18:56] jose, ok thx [18:58] mwenning: can your workstation reach the node its attempting to bootstrap when you just activate it in the MAAS gui? [18:58] lazyPower, yes, maas works fine, I can start/stop all my nodes [18:59] is `instance-type` ignored in environments.yaml? [18:59] mwenning: and workstation connectivity to the nodes works as well? [18:59] I put m3.mediums but it's asking for m1.smalls [18:59] jcastro: thats a depreciated constraint [18:59] you should add proper constraints [18:59] lazyPower, yes, I can ssh into all the nodes from my maas server [19:00] I thought we were putting it back in? [19:00] or at least defaulting to m3's? [19:00] mwenning: hmm... this is troubling [19:00] mwenning: which juju revision are you on? 1.20.x i assume? [19:01] juju-core 1.20.11 [19:01] ok, latest stable [19:01] hmm this sounds like a reachability issue though [19:02] what's wierd is that it doesn't seem to take anything from environments.yaml or environments/maas.jenv [19:03] what do you mean? [19:03] I tried changing bootstrap-timeout to 120 and it didn't take it. [19:04] I've got http-proxy and https-proxy set in the environments.yaml, that should point to the correct proxy but it doesn't seem to pay attention [19:05] Is there somewhere else that juju is reading that info? [19:06] mwenning: im' asking for a core dev to come take a look with us - 1 moment [19:06] i'm not using any proxies and have limited interaction with using a proxy in juju [19:10] hmm, what the heck? It's trying to access 192.168.1.1/MAAS . [19:10] That's not my MAAS server, it's at 192.168.0.2 [19:10] mwenning: stale maas jenv? [19:10] where is that? [19:11] where is maas.jenv? [19:11] ~/.juju/environments/maas.jenv [19:11] tvansteenburgh: ping, mind a quick pm? [19:11] jose: sure [19:12] actually maas.jenv is gone. [19:15] mwenning: ok so if its not reading a stale jenv file - something else is awry [19:15] can you verify the server config in environments.yaml is the correct ip address? [19:17] I lazyPower , yes. stby I'll paste the whole thing [19:18] https://pastebin.canonical.com/120819/ [19:22] mwenning: do you have another PROXy export set? [19:22] echo $JUJU_PROXY i think [19:23] lazyPower, $JUJU_PROXY shows nothing [19:23] * lazyPower snaps [19:23] that was the last thought i had on what it could be [19:23] i unfortunately dont know why you're seeing that behavior :( [19:24] lazyPower, ok. I've got to drive downtown to a meeting, I'll look at this later [19:25] hey bcsallerm can you add your +1 to this bug? [19:25] https://bugs.launchpad.net/juju-core/+bug/1373516 [19:25] Bug #1373516: Switch default instance type from m1.small to t2.small/m3.medium for EC2 provider

[19:27] hey natefinch, can we get this on someone's radar? ^^^^ [19:27] This will only get worse and worse === CyberJacob|Away is now known as CyberJacob [19:33] lazyPower, if you can think of anywhere that juju might be cacheing that information please let me know, that's all I can figure [19:34] mwenning: my only other suggestion is to do a full env dump and sift through whats in there [19:34] juju reads from env => .jenv => environments.yaml - in that order. [19:35] lazyPower, do you know if it reads http_proxy in the env? [19:35] i believe so [19:36] ok, I think that's new, but I've set it there too. [19:36] jcastro: I think we addressed that... lemme double check, thought I saw that in the 1.21 release notes [19:37] jcastro: hmm. maybe not === kadams54 is now known as kadams54-away [19:39] jcastro: it's definitely on our radar. I'll see if I can get someone to put a little time into fixing it. I think the main problem we had was that there wasn't really a no-brainer replacement. both t2.small and m3.medium had slight problems (I believe t2.small is a burstable instance, and m3.medium is slightly more expensive) [19:39] yeah [19:40] it was just reiterated to us that m1's are going away much faster than we thought [19:40] jcastro: personally, I'd just go with m3.medium.... it's not our fault amazon doesn't have cheap prices :) [19:40] (for reasonable machines) [19:41] jcastro: I'll bring it up right now. I agree it needs to get fixed ASAP. Of course, the problem is that it won't be fixed in older jujus [19:41] I'd rather have people pay more than having it not work because m1's are being culled [19:41] yep [19:41] jcastro: we also should handle that error better by requesting from a different AZ in the region and/or trying a different instance type [19:43] jcastro: who said the m1's are going away faster? Just want to give proper attribution. [19:43] ben and robert [19:43] aka the CPC team [19:43] robert filed the initial bug [19:50] jcastro: just sent a mail to juju-dev. Assuming no one objects, I'll try to make sure someone gets on it ASAP. [19:50] -<3 thanks === kadams54-away is now known as kadams54 === roadmr is now known as roadmr_afk === kadams54 is now known as kadams54-away === roadmr_afk is now known as roadmr === arosales__ is now known as arosales === CyberJacob is now known as CyberJacob|Away