/srv/irclogs.ubuntu.com/2013/11/11/#juju.txt

freeflyingis it possible to destroy a subordinate hacluster unit in dying state01:09
=== axw_ is now known as axw
=== axw__ is now known as axw
freeflyingthumper, if we wanna to destroy a service properly, and its has relation with others, what shall we do exactly03:24
freeflyingremove relation firstly and then destroy service?03:25
thumperfreeflying: I think with the fix that landed on friday, just destroying the service should work, but for now, best to be safe and remove the relations first03:34
thumperfreeflying: the fix won't be released until 1.17 i think03:34
freeflyingthumper, so remove-relation works in 1.16.0?03:35
thumperfreeflying: I hope so... seems pretty fundamental03:36
freeflyingthumper, :)03:39
=== axw_ is now known as axw
=== thumper is now known as thumper-afk
=== freeflying is now known as freeflying_away
=== freeflying_away is now known as freeflying
sbbrtnwhat is an easy to test charms?  I have a charm written but now need to debug it.  Is there a tool to do this?  Or do you just have to deploy it and see what happens?05:21
thumper-afksbbrtn: you can use the local provider and deploy there, it uses lxc locally06:46
sbbrtnI'm not able to ssh into the local provider06:47
freeflyingsbbrtn, do you have ssh key pair06:49
sbbrtnyeah.  I created an ssh key like the tutorial said06:49
sbbrtnmaybe I am running the wrong command...06:50
sbbrtni run "juju ssh 1" and it gives me an error.06:51
thumper-afksbbrtn: that should work06:51
thumper-afksorry06:51
thumper-afkno it won't06:51
thumper-afkuse 'juju ssh unit-name/n'06:51
thumper-afkit has to do with limitations I've not yet fixed06:51
thumper-afkalternatively06:51
thumper-afkall the log files are local06:51
thumper-afkby default in ~/.juju/local/log06:52
=== thumper-afk is now known as thumper
* thumper goes to do a few dishes06:52
sbbrtnahh.  thanks!06:52
sbbrtnthat works06:52
freeflyingI destroy-service, is it normal that the machine hasn't been released from maas?07:07
thumperfreeflying: yes, the machine isn't destroyed automatically (yet)07:19
thumperfreeflying: we are looking to change the default behaviour back to what pyjuju did07:19
thumperfreeflying: right now, when the last unit is removed from a machine, the machine is still around07:20
thumperfreeflying: we want to change this (soonish)07:20
freeflyingthumper, I just did it :)07:21
thumperfreeflying: just did what?07:21
freeflyingthumper, destroy services and then destroy machines07:21
freeflyingthen get the machine released from maas07:21
* thumper goes to roll his calves07:22
freeflying:)07:24
=== CyberJacob|Away is now known as CyberJacob
AskUbuntuCeph did not deploy to machine 0 why? | http://askubuntu.com/q/37532108:01
=== CyberJacob is now known as CyberJacob|Away
DanChapmanHey when i run 'juju bootstrap -e azure'. My memory completely fills (16GB + 8 GB swap) to the point my box crashes. I need to file a bug for this, is there any specific logs you would need included with the bug report?09:27
axwDanChapman: I can't think of anything off the top of my head. I'd say just file it, and when someone looks at it they may ask for more info.09:47
DanChapmanaxw, thanks I will do that then :-)09:48
axwDanChapman: do you have any non-default parameters in your environments.yaml? apart from the credentials09:48
axwperhaps list them09:48
axwDanChapman: also, the version of juju09:49
DanChapmanaxw all I changed were the credential: subscription-id, storage-account, cert path and location. I've included the juju version I'll add some info about the .yaml file09:50
axwcool, thanks!09:51
freeflyingafter destroy a service, some relation stays in dying, how can I get them removed?09:54
axwfreeflying: can you please pastebin a "juju status" for me?09:56
=== freeflying is now known as freeflying_away
bloodearnestheya all, 2 simple questions:11:33
bloodearnest1) is there a standard way to get a file path friendly version JUJU_REMOTE_UNIT? I see a lot charms implementing their own sanitizing functions11:34
bloodearnest(am using charm-helpers, but can't see one in there)11:35
bloodearnest2) is there a standard solution to subordinate charms possibly fighting with it's main charm over apt locks if their install hook execution overlaps? Does it indeed overlap? Is hook execution serialized per-unit?11:36
stubbloodearnest: no, everyone has implemented their own. It is probably worth adding to charm-helpers if you can think of a good name for it.11:55
stubbloodearnest: 2) is interesting - could be worth adding locking to the charm-helpers helper?11:56
stubbloodearnest: I don't know what the current serialisation guarantees are with hooks11:56
bloodearneststub: I'm seeing code in gunicorn subordinate charm to retry in a loop, which I've gone with. Just thought it was error prone if subordiante charm authors have to remember to do it11:59
stubbloodearnest: yeah, that sounds like a hack to me.12:00
bloodearnestas for names: urlsafe_juju_unit?12:01
stubI guess. Or sanitized_juju_unit if it is generic enough12:02
stubI mainly needed it for filesystem safe names12:02
stuboh - and sql safe for database names, usernames etc.12:03
bloodearnestgood point12:05
stub(think the last version I coded allowed only a-zA-Z0-9_ ?)12:07
stub(And now I think of it, it probably explodes if the service name starts with a number)12:07
=== freeflying_away is now known as freeflying
AskUbuntuHow to make my nodes online in maas? | http://askubuntu.com/q/37541213:09
jamespagebloodearnest, you should not hit 2) - within a container hook execution is serialized13:34
jamespageif you have that's a bug13:34
bloodearnestjamespage: ack, good to know. Not hit it, just working on a charm that has code to handle it, wasn't sure it was needed13:35
bloodearnestjamespage: I'll remove it13:35
jamespagebloodearnest, it used to be a problem13:35
jamespagebut then someone saw sense in that juju was much better positioned to resolve that sort of thing13:35
bloodearnestyep13:35
marcoceppijamespage: adam_g would you guys be opposed to me assigning reviews for openstack charms to ~openstack-charmers and having ~charmers abstain? In order to get them out of the queue?16:38
marcoceppiOr do you guys use the queue?16:39
marcoceppiI did one like that a few nights ago, then thought I should ask you all first16:39
marcoceppisbbrtn: local provider caveats have been documented https://juju.ubuntu.com/docs/config-local.html#caveats16:41
jamespagemarcoceppi, should we move the official branches over to that team as well?16:41
marcoceppijamespage: you're more than welcome to, it makes no difference to us at this point, though it might break a lot of existing deployer files. I know CTS explicitly uses lp:~charmers branches for their deployers16:43
jamespageok16:44
marcoceppiI just didn't want to break your review workflow16:44
marcoceppiand I didn't want new charmers to start reviewing stuff since it might be in flux with the openstack charms16:44
ahasenackhi, could I inderest someone into taking a look at a one-liner MP against the apache2 charm? https://code.launchpad.net/~ahasenack/charms/precise/apache2/apache2-no-failing-juju-log/+merge/19440317:23
contextMP ?17:30
contextmust be a bzr/lp term17:31
teknicoMP means Merge Proposal17:59
AskUbuntuJuju with GUI Installation | http://askubuntu.com/q/37557018:35
=== CyberJacob|Away is now known as CyberJacob
AskUbuntuUnable to install juju-gui locally? | http://askubuntu.com/q/37563521:25
rick_h_bac: lol at dual post answering22:14
rick_h_and <3 jcastro for editing22:15
=== freeflying is now known as freeflying_away
=== freeflying_away is now known as freeflying
=== CyberJacob is now known as CyberJacob|Away

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!