[09:09] <USER__> plop
[09:10] <weweje> hello
[09:10] <weweje> what's the timezone of kim0 // cairo?
[09:45] <kim0> weweje: hey
[10:40] <weweje> kim0 plop
[10:41] <weweje> sry
[10:41] <kim0> :)
[10:41] <kim0> what's up
[10:41] <weweje> nice
[10:41] <weweje> sent you a mail about the clouds banners
[10:42] <kim0> weweje: Bruno ?
[10:42] <weweje> and came here to know more about what's needed
[10:42] <weweje> yes ;)
[10:42] <kim0> great :)
[10:42] <kim0> Thanks for your interest man
[10:43] <kim0> Well we'll mostly need a few front page banner images .. I haven't exactly defined exactly what it should be though
[10:44] <kim0> so .. I'll ping you back once we've settled on something to be done .. sounds good ?
[10:45] <weweje> okaaaay
[10:46] <kim0> weweje: thank you :)
[10:46] <weweje> KIT
[16:36] <smoser> SpamapS, just random bit of infomration i just thought about.  there were 2 reasons for the .tar.gz rather than .img.gz .  1, we have extra files that are useful, and small, so makes sense for archive.  2.) the images are 2G, largely of zeros.  there was very real savings in using tar's sparse file handling rather than compressing hundreds of megs of zeros.
[17:00] <jamespage> smoser: do you have a good reference for accessing the ec2 API from behind a firewall?  Need to open up the right outbound access and it ain't working right at the moment
[17:02] <SpamapS> smoser: *OH* the sparse file handling does make sense.
[17:05] <smoser> jamespage, you just have to be able to access the end point over http (or https, depending on url)
[17:05] <smoser> but if you need IP addresses, its annoying.
[17:06] <smoser> i think they add load balancers every once in a while, and redirect from the main end point, and then your firewall rules don't work.
[17:06] <smoser> from time to time our image builds start failing to reach a region and i have to ask for IS to open something else up.
[17:07] <smoser> that said, there is this: http://aws.amazon.com/articles/3912?_encoding=UTF8&jiveRedirect=1
[17:08] <smoser> incomplete bugs suck
[17:08] <smoser>  * user opens bug
[17:09] <smoser>  * triagers asks some (silly or legitimate) question
[17:09] <smoser>  * new user opens clear duplicate, but doesn't see the incomplete bug because its incomplete
[17:09] <smoser>  * no one realizes there is a duplicate
[17:21] <jamespage> smoser: thankyou!
[17:28] <SpamapS> smoser: so, the template I wrote up for lxc downloads the SHA256SUMS, and the .gpg, and verifies that, and then verifies the tarball against it..
[17:29] <SpamapS> smoser: I'm not sure why we don't just have a .gpg for every file in that dir tho
[17:50] <smoser> SpamapS, probably no good reason.  the checksums are generated by some tools that also run on cdimage and they have some whitelists of what to checksum
[17:52] <SpamapS> well it works
[18:07] <smoser> hallyn, bug 795427
[18:07] <smoser> did you try to reproduce using /dev/sdX rather than /dev/vdX ?
[18:11] <hallyn> smoser: yes i tried both
[18:12] <hallyn> smoser: did you reproduce?
[18:25] <Kyle__> Er.  Is there a virtual graphics device on the vms started?
[18:52] <smoser> hallyn, no. it just sems like it should b reproducible
[19:15] <SpamapS> hallyn: got a link to that wiki page you used to get dnsmasq working in your resolv.conf ?
[19:32] <hallyn> smoser: you'd think so
[19:33] <hallyn> SpamapS: i think it was the one for vm-tools, testingenvironment, one sec
[19:33] <hallyn> SpamapS: https://wiki.ubuntu.com/SecurityTeam/TestingEnvironment
[19:38] <SpamapS> hmmm... avahi
[21:06] <SpamapS> hallyn: so, if I don't use cgroup-bin , whats the magic way to mount cgroups?
[21:06] <hallyn> fstab :)
[21:06] <SpamapS> hallyn: meaning what?
[21:06] <hallyn> cgroup          /sys/fs/cgroup          cgroup  defaults        0       0
[21:06] <SpamapS> just that?
[21:06] <hallyn> put that in fstab and do mount -a
[21:07] <hallyn> yeah
[21:07] <hallyn> i wish i'd have succeeded in getting my blog on planet.ubuntu.com so more than 3 people would read it :)
[21:07] <SpamapS> haha
[21:07] <hallyn> not that i deny having named it precisely so that i coudl remain anonymous.  originally.
[21:07] <SpamapS> yous hould be an ubuntu member by now
[21:08] <hallyn> i can't write to that bzr tree.  tried very recently
[21:08] <SpamapS> You asked?
[21:08] <hallyn> no.  figured that would happen automatically bsaed on membership
[21:08] <hallyn> automation.  it's good for the soul
[21:09] <hallyn> just like, i need to have lxc suggest debootstrap.
[21:13] <SpamapS> cgroups on /sys/fs/cgroup type tmpfs (rw)
[21:13] <SpamapS> That was already there, w/o the fstab entry
[21:13] <SpamapS> did cgroup-bin do that?
[21:14] <SpamapS> umount: /sys/fs/cgroup: device is busy.
[21:15] <SpamapS> hallyn: I can't seem to get things working w/o cgroup-bin .. and with cgroup-bin, lxc-stop always "fails" (but works anyway)
[21:15] <hallyn> why aren't things working w/o cgroup-bin?
[21:16] <hallyn> the cgroup-bin is second on my list to address
[21:16] <SpamapS> lxc-start: open /sys/fs/cgroup/cpu/lucid-test2/devices.deny : No such file or directory
[21:16] <hallyn> have you rebooted?
[21:16] <SpamapS> since when?
[21:16] <hallyn> since removing cgroup-bin
[21:16] <SpamapS> ew, no
[21:16] <SpamapS> I have to?
[21:16] <hallyn> yeah, i'ma fraid.  cgroup composition is not flexible
[21:16] <SpamapS> *booo*
[21:16] <hallyn> kernel magic
[21:17] <hallyn> well, you can try just umounting all cgroup fs's,
[21:17] <hallyn> then doing mount a- to mount /cgroup
[21:17] <SpamapS> it won't let me
[21:17] <hallyn> what does it say?
[21:17] <SpamapS> device busy
[21:17] <hallyn> btw the fstab line i gave you was for /cgroup existing.  you probably want to make it /sys/fs/cgroup ?
[21:17] <hallyn> ok, yeah you'd have to reclassify all tasks i guess
[21:17] <SpamapS> cgroup /sys/fs/cgroup cgroup defaults 0 0
[21:18] <hallyn> thumb up
[21:18] <SpamapS> So there are things that already have used my cgrou-bin thing? Its a package that can't be safely removed w/o reboot?
[21:18] <hallyn> cgred reclassified tasks
[21:19] <hallyn> i thin it's be possible to do it, but it woudl take work
[21:19] <hallyn> i.e., manually rmdir child cgroups,
[21:19] <hallyn> reclasiffy tasks as needed,
[21:19] <hallyn> then youc an umount
[21:19] <hallyn> and *then*, im unsure as to whether you can remount with all cgroups composed
[21:20] <hallyn> used to be, no.  i know patches were  floated, but not sure if they went in
[21:20] <SpamapS> rm: cannot remove `/sys/fs/cgroup/cpu/cpu.rt_period_us': Operation not permitted
[21:20] <SpamapS> ok so cgroups is still crazy
[21:20] <SpamapS> will just reboot
[21:21] <hallyn> SpamapS: i didnt' say rm -rf
[21:21] <hallyn> you can nev er delete the files
[21:21] <hallyn> but you can rmdir empty cgroup dirs
[21:21] <hallyn> poorly documented issue :(
[21:28] <SpamapS> well either way, reboot solved it
[21:29] <SpamapS> hallyn: is there a command like lxc-ls but for running containers?
[21:31] <hallyn> lxc-ls shows running containers
[21:31] <hallyn> in the second line
[21:31] <hallyn> improving lxc-ls and lxc-ps is one of those things everyone wants done (but noone wants to do) :)
[21:31] <SpamapS> no it shows all containers
[21:31] <SpamapS> running or not
[21:31] <hallyn> in the second line?
[21:31] <SpamapS> there is no second line
[21:31] <hallyn> not in mine
[21:32] <hallyn> then none are running
[21:32] <SpamapS> I have 2 running
[21:32] <SpamapS> oh they were starting up still
[21:32] <SpamapS> so .. -1 then.. is kind of ... useless
[21:33] <SpamapS> the output is actually kind of hard ot use.. have to run through until I see one I've seen before and thats where running containers start :-P