[10:24] hi there gmb: have you red the email from Gary? [10:26] Hi frankban. Yes, I have. Want to pair on it? [10:26] gmb: sure [10:27] 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] ok gmb [10:53] frankban, I'm in https://plus.google.com/hangouts/extras/canonical.com/lxc-start-whut# [10:57] (export OTHER=gmb REMOTE=chinstrap.canonical.com; trap reset 0 1 2 3 15; stty raw -echo isig && ssh chinstrap.canonical.com "cat /tmp/$OTHER-term") [11:07] sudo mount -t overlayfs -oupperdir=$LXC_DIR/ephemeral-binding,lowerdir=$LXC_BIND none $LXC_DIR/rootfs$LXC_BIND [11:35] gmb: sed -i '$ d' filename [13:04] * benji waits until after the call to apply updates... just in case [13:08] benji frankban gmb call in 2 [13:08] ack [13:43] I can see it now: all of our machines die and we have to resort to e-cannibalism. [14:03] heh [14:15] 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) < https://launchpad.net/bugs/609986 > [14:16] frankban, I should be ready to pair again in a moment. Just need to reboot. [14:16] gary_poster: do you want a spectator/heckler? [14:16] benji, sure :-) [14:16] 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. [14:17] (i.e. I knew what I was sending, but not how it was ending up) [14:17] gmb: cool [14:18] 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 [14:19] benji, Yes, that thought occurred to me, but the hangout was pretty sluggish (probably because of my VM( [14:19] ) [14:19] benji, I'm in https://talkgadget.google.com/hangouts/extras/canonical.com/goldenhorde [14:19] 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] oh, cool [14:20] gary_poster: I'll be there in a moment. I need to run out to the car to get my headset. [14:20] cool [14:26] * gmb reboots [14:46] frankban, I'm finally back up and running, and I'm in https://plus.google.com/hangouts/extras/canonical.com/lxc-start-whut [14:47] gmb: ok [14:55] gmb: sudo lxc-create -t ubuntu -n lp -f /etc/lxc/local.conf -- -r lucid -a amd64 -b graham [15:07] frankban, lp:~gmb/+junk/lxc-start-ephemeral.scratch/ [15:18] frankban, sed -i '/graham/d' /tmp/fstab [15:20] frankban, sed -i '/'$USER'/d' $LXC_DIR/fstab [15:49] frankban, https://pastebin.canonical.com/62032/ [16:05] benji, We have questions about lxc-start-ephemeral that you might be able to help us with. Are you free? [16:05] gmb: sure [16:06] gmb: are you hanging out? [16:06] benji, Cool, we're in https://plus.google.com/hangouts/extras/canonical.com/lxc-start-whut [16:32] 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] benji, :-) no, s'ok [16:40] * gary_poster about to babysit [17:04] gary_poster: Are you free for a brief chat about lxc-start-ephemeral and all its joys? [17:04] ] [17:27] * benji is back. [17:45] gmb/frankban: did you guys get the second ephemeral issue figured out? [17:45] benji: yeah, it's a bug in lxc-wait. [17:45] heh [17:45] Not our fault, and it's a problem we can hack around, probably. [17:46] Basically lxc-wait can only be used for one lxc instance at a time. [17:46] Because it always uses the same Unix socket. [17:46] 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] benji: I've just emailed the list with a patch and details of the problem. [17:46] Everything else works. [17:46] perfect [17:47] 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] lxc-start-ephemeral:204 in lp:~yellow/+junk/lxc-start-ephemeral [17:48] Because it just errors it means that the ephemeral instance starts and then (potentially) gets torn down again almost immediately. [17:48] Which I imagine would make testr a sad little panda. [17:52] Anyway, a relatively productive day. [17:53] (Said in much the same way as one would talk about a relatively productive cough) [17:53] (It's good that stuff's coming out, but no-one likes having to deal with it very much) [17:59] gmb, I am now, but I imagine this is your perfect time to escape, yeah? [17:59] heh [18:00] cool sounds good gmb & frankban. Have a great weekend [18:01] you too gary_poster [18:02] gary_poster: who shall pick up their banner? [18:02] benji, you definitely; me maybe. [18:03] k [18:03] benji, I don't think the lxc-wait issue is a problem for our usage of lxc-start-ephemeral; do you agree? [18:04] benji, I will file a bug for this right now [18:04] gary_poster: I don't /think/ so but I get the feeling that you have some deeper thought that I'm not following. [18:06] 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. [18:07] 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] So neither instance is an issue for us AFAICT [18:08] gary_poster: right [18:10] 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] yes, benji. [18:10] sweet [18:42] benji, call? I would like to talk about the ephemeral changes [18:43] gary_poster: sure, one second [18:43] cool, I'm in the horde [19:33] benji, I'm in https://talkgadget.google.com/hangouts/extras/canonical.com/goldenhorde but stepping away. back in 5 [19:33] gary_poster: ok