=== StoneTable is now known as aisrael | ||
=== stokachu_ is now known as stokachu | ||
=== \b is now known as benonsoftware | ||
=== wgrant_ is now known as wgrant | ||
=== Ursinha_ is now known as Ursinha | ||
=== kadams54 is now known as kadams54-away | ||
=== mwhudson_ is now known as mwhudson | ||
=== mwhudson is now known as Guest80728 | ||
jam | fwereade: you would remember that we discussed writing something like that, but I don't think we ever actualyl did | 08:33 |
---|---|---|
fwereade | jam, ah right :) | 08:33 |
jam | fwereade: what are you doing around now ? :) | 08:33 |
fwereade | jam, taking the opportunity to code a little while (almost) nobody's around ;p | 08:34 |
jam | understood | 08:34 |
fwereade | jam, (although if you fancy reviewing either of my branches (or just conversing on the open issues in the first one) I'd be most grateful | 08:36 |
=== mwhudson is now known as mwhudson__ | ||
=== mwhudson__ is now known as mwhudson | ||
fwereade | jam, hey, I thought I needed something like this: http://paste.ubuntu.com/11749789/ ...but it turns out it's not quite as simple as that | 09:38 |
fwereade | jam, but that general structure looks very familiar | 09:39 |
fwereade | jam, do you think there's a place for a worker that just serializes ops onto a single goroutine and lets us focus on the op implementation? | 09:41 |
fwereade | jam, possibly I should be calling them commands not ops? | 09:41 |
=== kadams54 is now known as kadams54-away | ||
=== Beret- is now known as Beret | ||
=== Tribaal_ is now known as Tribaal | ||
=== kadams54 is now known as kadams54-away | ||
=== kadams54 is now known as kadams54-away | ||
=== _thumper_ is now known as thumper | ||
mup | Bug #1467331 opened: configstore lock should use flock where possible <juju-core:Triaged> <https://launchpad.net/bugs/1467331> | 22:29 |
mup | Bug #1467331 changed: configstore lock should use flock where possible <juju-core:Triaged> <https://launchpad.net/bugs/1467331> | 22:38 |
mup | Bug #1467331 opened: configstore lock should use flock where possible <juju-core:Triaged> <https://launchpad.net/bugs/1467331> | 22:41 |
thumper | wallyworld: https://github.com/juju/juju/pull/2613 | 23:20 |
thumper | wallyworld: review board seems to be not picking up new pulls | 23:20 |
wallyworld | thumper: will look after standup | 23:21 |
wallyworld | thumper: lgtm | 23:37 |
thumper | wallyworld: cheers | 23:37 |
thumper | wallyworld: the other thought I had was the 1 second timeout | 23:38 |
thumper | wallyworld: do you think they might be hitting that? | 23:38 |
wallyworld | thumper: which timeout? is that mentioned in the bug report? | 23:39 |
thumper | no... but it is the timeout that that fslock uses | 23:40 |
thumper | the lock is only held during the duration of a read or write of the file | 23:40 |
thumper | so I originally thought 1s should be more than enough | 23:40 |
thumper | however if there are many ops running concurrently, they are serialized through this file | 23:40 |
thumper | so perhas 2s? | 23:41 |
thumper | I'm not sure | 23:41 |
thumper | could just leave it as it is for now | 23:41 |
thumper | I suppose | 23:41 |
thumper | pretty sure we never hit this before | 23:41 |
* thumper asks the bot to merge | 23:42 | |
thumper | will forward port to master | 23:42 |
thumper | damn it is going to be cold at the gym | 23:42 |
* thumper psychs himself up | 23:42 | |
wallyworld | thumper: yeah leave it for now if we haven't seen it before. but if the system has lots to do. may be a factor. i had to introduce reties to lxc ops due to i/o ontention recently | 23:42 |
thumper | meh, just noticed bug 1464356 | 23:50 |
mup | Bug #1464356: TestCloudInit fails <blocker> <ci> <gccgo> <regression> <test-failure> <windows> <juju-core:Invalid> <juju-core 1.24:Triaged> <https://launchpad.net/bugs/1464356> | 23:50 |
* thumper off to the gym | 23:50 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!