[00:00] juju sync-tools on maas 502 bad gateway error | http://askubuntu.com/q/469869 === CyberJacob is now known as CyberJacob|Away [04:46] o/ stub [04:47] stub: I'm looking at the postgresql charm [04:47] and considering two main things: [04:47] * what is the minimum reasonable things I need to configure for ec2 so the db exists beyond the machine [04:47] * how do I restore a backup that has been taken [04:48] * thumper is looking at python-django, postgresql, and gunicorn [04:48] probably going to use a subordinate charm for django for the app... [04:59] thumper: two (or more) units means you have a replica of all your data. pg_restore is the command to restore a database, and the charm doesn't help you do it. [04:59] hey guys, do you think a charm for an archive mirror would be useful? I know it wouldn't be able to run in a t1.micro aws instance, for example, but I think it would be a good idea to have one for, as an example, private clouds [05:00] The backup design and scripts predates me - I need to go over it, adding in PITR options. [05:13] hey stub, did you place test_hooks.py in hooks to work around stupid path munging? [05:13] (wrt the postgresql charm) [05:13] lazyPower: No, it ended up in there because that is where the person who wrote it put it :) [05:14] stub: have a moment to lend your eyeballs to my path munging woes? I'm trying really hard in an area i just dont understand. [05:14] lazyPower: I tend to be an 'accept improvements, and polish them later' type gateway. [05:14] sure [05:14] http://paste.ubuntu.com/7495968/ [05:14] wait, thats an outdated pastebin [05:15] http://paste.ubuntu.com/7496027/ - there's my dir tree [05:17] and the only reason i'm using nose, is because i've used it in context of a python project and it worked pretty well. it just 'knew' what was up. It was a similar layout - instead of hooks/ it was module_name/* -- now the issue. When I run nosetests against tests, it fails under charmhelpers, importing stuff - but only in the context of testing [05:17] when i'm running the hooks, they have no such import errors, so i can only assume its nose's path munging [05:18] 10_test_common.py? [05:18] yeah, it gets to import hooks/common - from there it starts tree loading teh dependency chain and blows up [05:18] 1 sec, let me push this branch to bzr and get you some output for context [05:19] lp:~lazypower/+junk/bind [05:20] http://paste.ubuntu.com/7496033/ [05:21] At the top of your '10_test_common.py', you probably want something like 'import sys; import os.path; sys.path.append(os.path.join(os.path.dirname(__file__), os.pardir, 'hooks')) [05:21] stub: i have hooks, and hooks/charmhelpers being added implicitly in the test file [05:21] With an absolute path, or a relative path? [05:22] If you use a relative path, the current working directory will hurt you [05:22] it'll resolve to an abspath [05:22] ah i stripped that in the version i committed [05:22] what i had was; http://paste.ubuntu.com/7496035/ [05:22] same error [05:22] You really want to calculate the path from __file__, as that is the only fixed absolute path you can start from IIRC [05:24] 'from hooks import' means you need the parent directory of 'hooks' in the path. You are putting the hooks directory itself in the path. [05:24] ooohhhh [05:25] * lazyPower looks shifty [05:25] i knew that [05:25] well that'll resolve something later down the road, still failing. I think I'm going to follow the pattern of the pgsql charm and just move it to hooks [05:26] i dont know what I did wrong, but there's something terribly wrong here. [05:26] try: [imports] except ImportException: print repr(os.getcwd()); print repr(sys.path); raise [05:32] trying to get it to trigger, Its raising OSError which i've changed, but still not catching in the try/except block. Which means its probably raising from before this? [05:39] stub: thanks for the help, its been food for thought. Doesn't seem to be working though [05:41] lazyPower: Yup. I'd need to see the traceback [05:41] I can have another look in about an hour [05:41] stub: i never got the path to print. whats up in the +junk repo is pretty much the same status of the code save for the try/except block. [05:41] I'm calling it a night.. its extremely late [05:42] thanks again for taking a look o/ === CyberJacob|Away is now known as CyberJacob === CyberJacob is now known as CyberJacob|Away === isviridov is now known as isviridov|away [09:39] gnuoy, looking at tribaals resync branches now [09:40] ak [09:43] lazyPower: btw. Make sure your hooks directory has a __init__.py or Python won't think it is a package and 'import hooks' will fail. [09:48] Tribaal, is there a bug for the apt race thing you fixed via charm-helpers? [10:00] axw: around? [10:00] axw: it should work with just juju-mongodb as it does in Trusty. [10:01] axw: we need to push this back in an SRU to Trusty, so I'm not sure we can just bring in a new dependency like that. [10:01] axw: any idea why the requirement changed? === vladk|offline is now known as vladk [10:12] gnuoy, I made the publish target in Makefile run lint and test first [10:12] if they fail you can't publish [10:12] gnuoy, does that make sense? should help stop lint and test failures getting promoted [10:13] completely [10:17] gnuoy, tribaal's branches all landed [10:17] jamespage, landed to stable ? [10:18] gnuoy, yes [10:18] gnuoy, I've not enacted 'next' yet [10:18] :-) [10:18] kk [10:35] gnuoy, nsx fixups queued for review - https://code.launchpad.net/~openstack-charmers/+activereviews [10:36] gnuoy, if you like I can show you how to deploy that on serverstack [10:36] that would be great [10:39] marcoceppi, just so you know I've branched the ganglia, ganglia-node and ntp charms precise->trusty under charmers - but I've not promulagted [10:39] needed charm store accessible trusty versions [10:45] rbasak: it works in Trusty without mongodb-server? ok then - I can take another look in my morning [10:45] rbasak: why does it need an SRU in Trusty, if it's only an issue in Utopic? [10:45] axw: I believe so, yes - we have a dep8 test that passed when juju-core was last uploaded to Trusty, and it worked then. [10:45] mk [10:45] axw: there are other fixes that we can't fix in Trusty, because the SRU process expects them to be fixed in Utopic first. [10:46] ah I see [10:46] Really, juju needs to be fixed so that a copy-forward to the next development release Just Works without needing any code changes. [10:46] I'm addressing that separately. [10:46] rbasak: understood. if I get a chance I'll look tonight, otherwise in my morning. if it's urgent, better get onto someone else in juju-dev [10:47] axw: OK, thanks. This has been delayed by weeks now, so I don't think another day will really make any difference. [10:47] okey dokey [10:47] * axw bbl [13:29] jamespage: are they trusty ready? do they need to be promulgated? [13:29] typically branches shouldn't live under ~charmers that aren't promulgated [13:39] marcoceppi, meh - probably [13:39] I've been using them on trusty for ages === sebas538_ is now known as sebas5384_ === Ursinha is now known as Ursinha-afk === Ursinha-afk is now known as Ursinha [14:58] stub: so, i had a nose plugin that was causing 90% of my headache. [14:58] * lazyPower burnt it with fire === isviridov|away is now known as isviridov === ted is now known as tedg === roadmr is now known as roadmr_afk [16:42] hey guys [16:42] something strange happened here: I cannot access juju-gui with the password on local.jenv [16:42] it was working until yesterday. Today I tried to login, using the password on the file, but it says "Unknown user or password." [16:42] anyone has any idea of what it might be? Or even where can I get the "real" passowrd, since the one on local.jenv is not working? [16:46] worse. "juju stat" is giving me this: http://pastebin.ubuntu.com/7498090/ [16:46] And no, I can't use "juju bootstrap" to create a new environment since I'm on a production machine [16:51] hackedbellini: did you change the password in the jenv file from when the environment was created? [16:51] hackedbellini: and that looks like a local mongodb timeout? [16:52] hackedbellini: and the gui talks over the api to juju so if there's issues there it could be more a juju failure than the login [17:14] rick_h_: no, I didn't change the password [17:14] and yes, I'm googling a little and found that it has something to do with mongodb. Do you have any tips on how can I debug this? [17:16] hackedbellini: can you start by explaining a bit on the environment here. It seems odd that it's timing out mongodb on localhost. Maybe that is from the state server machine 0? Can you ssh to that machine and see if mongo is running? [17:16] hackedbellini: maybe provide some info on the version and such of ubuntu, juju, etc this environment is on? === roadmr_afk is now known as roadmr [17:37] rick_h_: well, I think we fixed it. We restarted juju-agent-juju-local and everything worked again [17:37] really strange though. In our logs, the service restarted itself at 10am (UTC-3). Nobody did nothing for it to happen [17:37] hackedbellini: well yay and hmmmm, wonder what that was [17:39] rick_h_: hahaha exactly my reaction to this, "yay and hrmmmm". If I find something, I'll report here or open a bug ;) [17:40] hackedbellini: thanks, and glad it's working and not a bug in the GUI :) === ryan is now known as Guest80317 [18:10] Guest80317: o/ Hey Ryan [18:11] lol well i guess im guest [18:11] jcastro: Guest80317 is the ILOC guy I was telling you about that may or may not be able to assist us in this endeavour [18:11] hi! [18:12] hey [18:13] i need to get my nick registered but ill get to that === Guest80317 is now known as ILOC === ILOC is now known as ILOC_guy [18:15] that will have to work for a min [18:17] hah, nice handle ILOC_guy [18:17] i figured that would work [18:18] So, we have some new stuff coming down the pipeline, and one of which is a HyperV virtualization solution for people interested in Juju. How do you feel about pioneering VM image creation with us on HyperV? [18:18] Well i think i could help ya out [18:18] i will of course need some detail [18:18] jcastro: ^ drive my man. drive away. [18:20] ILOC_guy, yeah that would be awesome! [18:21] basically, we want hyperv users to have an out of the box box (heh) so they can turn around and deploy onto hyper-v if they want [18:21] that would be nice, since VMM isnt exactly the best software that i have used [18:22] i dont know if you have used it but its about as much fun as a sharp stick in the eye [18:23] that snap in based management console for the hyp-v stuff? [18:23] i'd rather herd carts.... [18:23] *cats [18:24] Yea its part of M$'s system center [18:24] well, as long as it's repeatable ... [18:25] it can be but its not fast or flexable .... as im dealing with currently [18:26] ILOC_guy: hows the overarching experience with building machine images from powershell? [18:27] That's where we're going to see this go I would imagine. Building a base image with powershell and letting cloud-init handle all the heavy lifting inside the base box. [18:28] PS is better.... [18:29] ILOC_guy: we should have a group sync at some point in the near future and talk about a path to success and get some feedback on how you would tackle this problem [18:29] jcastro: thoughts? [18:30] yeah sounds good to me! [18:30] ILOC_guy: pick a date thats not the 30'th through june 2nd. [18:31] and a timeslot. I'll make a calendar invite and get everyone on board and we can go from there. give you a chance to talk to your boss if you want to do this on office hours or not. [18:32] marcoceppi: hi [18:33] Hello everybody [18:34] Well my boss is in the M$ camp so i dont know if i can use HER work time to do it. [18:35] ILOC_guy: no worries, if it's a hobby its a hobby :) Just a thought. [18:36] well im not saying that i wont push the subject but i need to have ammo to take to this fight [18:41] ILOC_guy: i'm crafting an email to the group of us that's responsible for our VM story. This way you can get answers out of band without tracking us all down on here. Thanks for agree'ing to help - we certainly appreciate the time to talk to us today. [18:43] of course any time, its nice to be distracted from the every day roll [18:44] how to restart unit agent? === sebas538_ is now known as sebas5384 [18:49] ILOC_guy: you've got mail. [18:49] o/ sebas5384 [18:50] hey lazyPower o/ [18:50] well look at that i do... lol [18:52] I have a region controller and I am trying to bootstrap node0 to precise, is that not possible? [18:58] marcoceppi: need your help. in config-changed hook, i need to restart my server after configuration changed but do not want to run restart code for the first time like, install->config-changed-start [18:59] marcoceppi: how to block restart code when config-changed runs after install hook? === CyberJacob|Away is now known as CyberJacob [19:07] themonk: whats the problem with restarting? [19:08] themonk: if you have, you can always make an intelligent function that knows about that rule [19:09] sebas5384: problem is how can i restart before start hook because it is not started yet after install->config-changed [19:10] themonk: you can divide in others scripts [19:10] sebas5384: how do i know that confir hook is running for the first time? [19:11] you can touch a file :) [19:11] touch .installed [19:12] if you don't started but you are doing: stop, start [19:12] will show something like a warning, but not an error [19:13] sebas5384: :) i thought about touch like solution. but is there any think juju provide? [19:14] sebas5384: *anything [19:15] themonk: well if juju can give you that i didn't know [19:15] sebas5384: thanks :) [19:16] themonk: o/ [19:33] themonk: juju doesn't have a concept of if a service is started or not. That's more the domain of whatever you're using to handle your process [19:33] eg: upstart [19:57] HELLO [19:57] WHERE IS JOJO? [19:57] JUJU === vladk is now known as vladk|offline === CyberJacob is now known as CyberJacob|Away [21:34] Is juju for Openstack reliable as Packstack ? [21:34] (the juju charm) [21:39] lazyPower and jcastro alpha draws: https://s3.amazonaws.com/uploads.hipchat.com/45897/307411/beJXOwF8csdcR9C/info_taller_juju_maas-01.jpg [21:39] thats an awesome infographic [21:39] i wish i read spanish [21:39] nothing finished yet, we are going to do more series, etc... [21:39] lazyPower: portugues !!! [21:39] hehe [21:39] that ^ === sebas538_ is now known as sebas5384 [21:40] cool :) [21:40] :D [21:41] jose maybe you understand hehe [21:42] sebas5384: spanish? :P [21:42] haha [21:42] hahaha [21:42] jose: dont pull a lazypower mistake [21:42] its protugues with three exclamation points. [21:43] hahaha [21:47] portuguese usually is with three exclamation points [21:48] sarnold: ever the ambassador of the obvious :P [21:48] * sarnold shines his Capt Obvious badge [21:50] https://www.youtube.com/watch?v=t85OkP9bQEw#t=407 [21:50] hehe [21:58] sebas5384, dude that is so cool! [21:58] thanks jcastro!! :) [22:01] jcastro: those are not the finals [22:01] so any feedback will be great! [22:01] :D === swebb is now known as zz_swebb === yo61_ is now known as yo61 === jcsackett_ is now known as jcsackett === dpritchett_ is now known as dpritchett === _mup__ is now known as _mup_