frankban | hi there gmb: have you red the email from Gary? | 10:24 |
---|---|---|
gmb | Hi frankban. Yes, I have. Want to pair on it? | 10:26 |
frankban | gmb: sure | 10:26 |
gmb | frankban, Cool. I just need to finish doing the update/upgrade/reboot cycle and then I'll be ready. I'll ping you shortly. | 10:27 |
frankban | ok gmb | 10:27 |
gmb | frankban, I'm in https://plus.google.com/hangouts/extras/canonical.com/lxc-start-whut# | 10:53 |
gmb | (export OTHER=gmb REMOTE=chinstrap.canonical.com; trap reset 0 1 2 3 15; stty raw -echo isig && ssh chinstrap.canonical.com "cat /tmp/$OTHER-term") | 10:57 |
frankban | sudo mount -t overlayfs -oupperdir=$LXC_DIR/ephemeral-binding,lowerdir=$LXC_BIND none $LXC_DIR/rootfs$LXC_BIND | 11:07 |
frankban | gmb: sed -i '$ d' filename | 11:35 |
* benji waits until after the call to apply updates... just in case | 13:04 | |
gary_poster | benji frankban gmb call in 2 | 13:08 |
gmb | ack | 13:08 |
benji | I can see it now: all of our machines die and we have to resort to e-cannibalism. | 13:43 |
gary_poster | heh | 14:03 |
gary_poster | benji, I'm going to futz with bug 609986 while we are waiting on the ephemeral script changes. | 14:15 |
_mup_ | Bug #609986: layer setup failures don't output a failure message (they spew straight to console) <lp-foundations> <paralleltest> <Launchpad itself:Triaged> < https://launchpad.net/bugs/609986 > | 14:15 |
gmb | frankban, I should be ready to pair again in a moment. Just need to reboot. | 14:16 |
benji | gary_poster: do you want a spectator/heckler? | 14:16 |
gary_poster | benji, sure :-) | 14:16 |
gmb | benji, Inicdentally, we used your term-sharing-via-chinstrap thing this morning; worked very well, though it was disconcerting to not be able to see what frankban was seeing. | 14:16 |
gmb | (i.e. I knew what I was sending, but not how it was ending up) | 14:17 |
benji | gmb: cool | 14:17 |
benji | gmb: something Gary has done when we were debugging the hack is to use the hangout screen sharing to then share the window back with me so I can see how it's ending up on his side | 14:18 |
gmb | benji, Yes, that thought occurred to me, but the hangout was pretty sluggish (probably because of my VM( | 14:19 |
gmb | ) | 14:19 |
gary_poster | benji, I'm in https://talkgadget.google.com/hangouts/extras/canonical.com/goldenhorde | 14:19 |
benji | gmb (and everyone): earlier this morning I made a slight change to the terminal sharing page that should make it a little more reliable | 14:19 |
gary_poster | oh, cool | 14:19 |
benji | gary_poster: I'll be there in a moment. I need to run out to the car to get my headset. | 14:20 |
gary_poster | cool | 14:20 |
* gmb reboots | 14:26 | |
gmb | frankban, I'm finally back up and running, and I'm in https://plus.google.com/hangouts/extras/canonical.com/lxc-start-whut | 14:46 |
frankban | gmb: ok | 14:47 |
frankban | gmb: sudo lxc-create -t ubuntu -n lp -f /etc/lxc/local.conf -- -r lucid -a amd64 -b graham | 14:55 |
gmb | frankban, lp:~gmb/+junk/lxc-start-ephemeral.scratch/ | 15:07 |
gmb | frankban, sed -i '/graham/d' /tmp/fstab | 15:18 |
gmb | frankban, sed -i '/'$USER'/d' $LXC_DIR/fstab | 15:20 |
gmb | frankban, https://pastebin.canonical.com/62032/ | 15:49 |
gmb | benji, We have questions about lxc-start-ephemeral that you might be able to help us with. Are you free? | 16:05 |
benji | gmb: sure | 16:05 |
benji | gmb: are you hanging out? | 16:06 |
gmb | benji, Cool, we're in https://plus.google.com/hangouts/extras/canonical.com/lxc-start-whut | 16:06 |
benji | gary_poster: I'm done with them and am starving so I think I'll do lunch now unless you're in desperate need of heckling. | 16:32 |
gary_poster | benji, :-) no, s'ok | 16:32 |
* gary_poster about to babysit | 16:40 | |
gmb | gary_poster: Are you free for a brief chat about lxc-start-ephemeral and all its joys? | 17:04 |
gmb | ] | 17:04 |
* benji is back. | 17:27 | |
benji | gmb/frankban: did you guys get the second ephemeral issue figured out? | 17:45 |
gmb | benji: yeah, it's a bug in lxc-wait. | 17:45 |
benji | heh | 17:45 |
gmb | Not our fault, and it's a problem we can hack around, probably. | 17:45 |
gmb | Basically lxc-wait can only be used for one lxc instance at a time. | 17:46 |
gmb | Because it always uses the same Unix socket. | 17:46 |
benji | gmb: I guess you're about to EOD, in which case Gary and I can take over if you can transfer enough state to us. | 17:46 |
gmb | benji: I've just emailed the list with a patch and details of the problem. | 17:46 |
gmb | Everything else works. | 17:46 |
benji | perfect | 17:46 |
gmb | benji: There's only one place where lxc-wait is really an issue, and that's where it's used to block until the ephemeral instance stops. | 17:47 |
gmb | lxc-start-ephemeral:204 in lp:~yellow/+junk/lxc-start-ephemeral | 17:47 |
gmb | Because it just errors it means that the ephemeral instance starts and then (potentially) gets torn down again almost immediately. | 17:48 |
gmb | Which I imagine would make testr a sad little panda. | 17:48 |
gmb | Anyway, a relatively productive day. | 17:52 |
gmb | (Said in much the same way as one would talk about a relatively productive cough) | 17:53 |
gmb | (It's good that stuff's coming out, but no-one likes having to deal with it very much) | 17:53 |
gary_poster | gmb, I am now, but I imagine this is your perfect time to escape, yeah? | 17:59 |
gary_poster | heh | 17:59 |
gary_poster | cool sounds good gmb & frankban. Have a great weekend | 18:00 |
frankban | you too gary_poster | 18:01 |
benji | gary_poster: who shall pick up their banner? | 18:02 |
gary_poster | benji, you definitely; me maybe. | 18:02 |
benji | k | 18:03 |
gary_poster | benji, I don't think the lxc-wait issue is a problem for our usage of lxc-start-ephemeral; do you agree? | 18:03 |
gary_poster | benji, I will file a bug for this right now | 18:04 |
benji | gary_poster: I don't /think/ so but I get the feeling that you have some deeper thought that I'm not following. | 18:04 |
gary_poster | benji, the script only uses lxc-wait twice. The second one is not an issue because it only comes into play if you do not pass a command to lxc-start-ephemeral. We always do for the case of the parallel tests. | 18:06 |
gary_poster | The first time we use lxc-wait, we are waiting on RUNNING, and as long as we proceed if that fails (as I am pretty sure we do) we will be ok because we just try to ssh in, and don't do anything until it is ready. | 18:07 |
gary_poster | So neither instance is an issue for us AFAICT | 18:07 |
benji | gary_poster: right | 18:08 |
benji | gary_poster: if the g&f team fixed the overlayfs issue, then we should be ready to try running the tests again; right? | 18:10 |
gary_poster | yes, benji. | 18:10 |
benji | sweet | 18:10 |
gary_poster | benji, call? I would like to talk about the ephemeral changes | 18:42 |
benji | gary_poster: sure, one second | 18:43 |
gary_poster | cool, I'm in the horde | 18:43 |
gary_poster | benji, I'm in https://talkgadget.google.com/hangouts/extras/canonical.com/goldenhorde but stepping away. back in 5 | 19:33 |
benji | gary_poster: ok | 19:33 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!