[06:26] davecheney, lesson learned - never eat [06:27] Bug #1445338 was opened: Win builds fail: cloudconfig/userdatacfg.go:65: undefined: unixConfigure

[06:33] Bug #1445338 changed: Win builds fail: cloudconfig/userdatacfg.go:65: undefined: unixConfigure

[06:39] Bug #1445338 was opened: Win builds fail: cloudconfig/userdatacfg.go:65: undefined: unixConfigure

=== kadams54 is now known as kadams54-away [07:51] Bug #1445369 was opened: Juju core freaks if /etc/os-release is not present [07:57] Bug #1445369 changed: Juju core freaks if /etc/os-release is not present [08:03] Bug #1445369 was opened: Juju core freaks if /etc/os-release is not present [09:45] Bug #1445146 changed: juju run fails after upgrade to 1.23-beta4.1 [10:12] jam: looks like irc dropped out - the maas guys are in another meeting, did you want to chat about the error stuff? [10:13] wallyworld_: I'm happy to chat if you'd like [10:13] I'm in the hangout [10:13] jam: ok, give me a sec and i'll change rooms [13:01] Bug #1445369 changed: Juju core freaks if /etc/os-release is not present [13:13] Bug #1444537 was opened: Log files from units deployed in lxc containers are shared on the physical node

[13:22] Bug #1444537 changed: Log files from units deployed in lxc containers are shared on the physical node

[13:31] menn0: I still need to ammend that branch so one sec on review [13:31] mgz: kk [13:31] Bug #1444537 was opened: Log files from units deployed in lxc containers are shared on the physical node

[13:35] menn0: it's okay, just wanted to run the unit test on vivid, they're good [13:35] mgz: menn0: would it be reasonable to add a logger.Debugf to that code you landed ? [13:36] jam: yup, totally. [13:36] (generally if you are suppressing an error it would be good to log it at least) [13:36] mgz: ship it [13:37] jam: you mean regarding upstart detection? [13:37] menn0: when you get the error that /sbin/initctl isn't there, just log at Debugf level [13:37] then if we get it for weird reasons [13:37] we can enable debug logging [13:37] he means a bit like we added when trying to debug, dump the actual error back from exec.Command [13:37] and see what the error we're getting is [13:37] jam: yep, that's a good idea [13:37] I shall add now [13:38] as long as it's just at debug [13:38] mgz: menn0: absolutely. But anytime you "add reporting to figure out what's going on", that's a good sign we may want a logger.Debugf for future use [13:38] for sure [13:40] logger.Debugf("exec %q failed: %v", initctlPath, err) [13:40] maybe? [13:40] anything else? [13:40] mgz: looks good to me [13:41] maybe %#v so we see the field names for the error [13:41] okay, done [13:41] mgz: can you test the output to confirm? [13:42] mgz: sorry that should be %+v [13:42] (hard code a different /sbin/init, or make a test fail) [13:42] talk about backseat coding... :) [13:42] menn0: well, %+v is just the field names for auto format, #v isthe go syntax which often includes field names [13:43] jam, mgz: whatever works [13:52] I pushed the log statement, will land shortly [15:16] Bug #1445338 changed: Win builds fail: cloudconfig/userdatacfg.go:65: undefined: unixConfigure

[15:45] going to need re-check review [15:45] for test junk I missed [15:49] okay, have pushed new test-fixing revision === FunnyLookinHat_ is now known as FunnyLookinHat [20:56] Bug #1445658 was opened: juju fills logs with attempts to do work on implicitly removed containers