=== matsubara-afk is now known as matsubara [00:26] hi all, I'm looking for help with an old Ubuntu server? Is this the right place? === platinum_ is now known as platinum === platinum is now known as platinum_ [00:27] Anyone online? [00:28] this is the right place, but it's a relatively quiet room, need some !patience [00:30] !ask [00:30] Please don't ask to ask a question, simply ask the question (all on ONE line and in the channel, so that others can read and follow it easily). If anyone knows the answer they will most likely reply. :-) See also !patience === matsubara is now known as matsubara-afk === AntORG_ is now known as AntORG === Myrtti is now known as Guest71494 === ahs3` is now known as ahs3 === guampa_ is now known as guampa [01:56] I'm getting ready to set up my Ubuntu server on an actual machine insteadd of a VM. I'm kinda curious, does anyone have any up-time monitoring software/service to recommend? [02:12] hi all. so I've got a syntax error in /etc/sudoers on 12.04, obv wasn't smart enough to be logged in as root when making changes [02:13] so now can't use sudo, can't use visudo, pkexec isn't installed... what other options do I have other than rebuilding the VM? [02:14] boot it into recovery mode [02:15] can I initiate a reboot without sudo? [02:15] or rather: how would one do that, I thought you couldn't [02:16] I can power off the machine but then it seems I can't get into a VNC console quick enough to hit escape to halt on the grub menu [02:17] mariooo: I'm not sure [02:17] mariooo: alternatively you could mount the vm disk and edit it manually [02:18] lifeless: awesome, thanks for the tips [02:25] lifeless: managed to get into recovery mode, but everything seems to be readonly...? have I missed something? [02:39] lifeless: nvm, bug in 12.04 recovery mode. enabling networking first remounts disk as rw. success! [02:43] mariooo: excellent [02:43] mariooo, for future reference, pressing ctrl alt del on a 'local' terminal reboots [02:43] like from the VM console [02:51] qman__: awesome, will keep that in mind. I wonder how you ctrl+alt+del over VNC with a mac keyboard hah [04:47] eth0 192.168.1.20 is pingable.. but eth0:1 192.168.1.21 or eth0:2 192.168.1.22 are not, from remote host.. what gives? [06:03] any ideas why this command, if executed after logging in, works as expected, but not if I pass it to ssh as the command to execute: [06:03] (nohup watch ls -l & disown -h) [06:03] in the first case, the command continues to live after logging out [06:04] in teh second case it dies [06:13] mgw: I don't understand what you're trying to do, but try using "ssh -T ... heh .. what use it to run watch in the background? :P [06:15] That confused me too :) [06:17] It's just a test [06:17] I'm trying to run ngrep [06:18] stgraber, you aroud still? [06:19] <_ruben> cleanest would probably to write a (perl/python/whatever) script that properly forks and runs the command(s) you want === stiv2k_ is now known as stiv2k === smb` is now known as smb === myrtti is now known as Myrtti [07:58] hey guys got my server working with iptables for masquerading and ufw for firewall, but for the life of me, how can I forward ports from ppp0 to a internal server [08:08] jtv: Hey, i think i missed your comment about python-django regarding translations.. Do you know what i need to change? [08:09] Daviey: its translations uploads (as produced by the build) should follow our naming conventions: [08:09] - One template per directory. [08:09] - Translation files named .po [08:09] - Each translation file in the exact same directory as its template. [08:10] The translations import queue gardener uses this to figure out which template and language a PO file upload is for. [08:11] Hey there mrevell === allenap` is now known as allenap [08:13] Hello jtv [08:20] jtv: Hmm, ok.. is there a way i can test changes work, without doing another upload? [08:23] Daviey: there's not too much to test, so I can just have a look at whatever you get, but you could try tar'ing up the .po & .pot files you get, uploading those to qastaging or staging, and then approving (or have someone like dpm or myself approve) the templates for import. The templates (i.o.w. the .pot files) should then be imported, and the .po files should be approved and then be imported themselves. [08:23] (You can't upload to a package that way, but the effect is much the same if you upload to a project release series on [qa]staging) [08:25] jtv: ok.. will try to do it, or get it done at least by EOD [08:25] Thanks! [08:25] One of those things I wish I had enough priority for: overhauling automatic approval. [08:26] jtv: ah, so it does work.. just a manual process to accept? [08:26] Just for the templates, yes. [08:26] but the existing translations are dropped on the floor? [08:27] And when we open a new series, whatever templates are current in the existing Ubuntu get copied over (and any applicable translations shared) so that the new series will populate itself very quickly. [08:27] Translations of the same strings in the template with the same name are dynamically shared between release series. [08:27] right [08:27] but 'upstream' translations are lost for new uploads? [08:27] So as soon as a message gets translated in Oneiric, it's translated in Precise, and vice versa. [08:28] There is also sharing between upstream and Ubuntu. By and large, Ubuntu automatically gets the upstream ones. [08:28] The other way is dependent on permissions etc. [08:28] right [08:28] thanks jtv [08:35] morning all === popey_ is now known as popey [09:48] Daviey: You do per-commit package builds of Essex for testing, right? [09:51] soren_: yes [09:51] Daviey: Which packaging branch are you using? [09:51] soren_: although, it's offline at this very minute.. the hardware had to be relocated to a different DC [09:52] http://bazaar.launchpad.net/~ubuntu-server-dev/nova/essex doesn not apply for me right now. [09:53] soren_: I suspect it needs rebasing.. the c-i is down, and it seemed like the best time to fixer it. [09:53] as in, early in the cycle. [09:54] Oh, ok, so that is indeed the right branch? [09:54] yep [09:54] Cool. [09:54] I'll look no further [09:54] soren_: any rebasing is most welcome :) [09:54] I'm on it. [09:54] \o/ [10:02] hi [10:03] i think this bug 1004316 is reported on the wrong project [10:03] Launchpad bug 1004316 in ubuntu-cloud-portal "Cloud server image OVF image fails to load in VirtualBox" [Undecided,New] https://launchpad.net/bugs/1004316 [10:12] it is indeed incorrect... smoser will fix0r that when he starts for the day i believe [10:12] thanks daker [10:13] yw ツ [10:20] Hi, I'm having trouble getting dnat to work on my 11.10 KVM server [10:21] Daviey: Merge prop pushed. [10:21] I'm tryint for DNAT port 25 to 192.168.122.10 using -A PREROUTING -p tcp -m tcp --dport 25 -j DNAT --to-destination 192.168.122.11 [10:22] dnat rule shows up: DNAT tcp -- anywhere anywhere tcp dpt:smtp to:192.168.122.11 [10:22] nocturn: You're misspelling the IP. [10:22] OK [10:22] have been trying this for a while! [10:22] "OK"? [10:23] I was expecting an "Oh, so I am" or something. Not "OK". [10:23] You say you're trying to DNAT to 192.168.122.10. [10:23] ...but then proceed to DNAT to 192.168.122.11. [10:24] soren_, OK meant that I just noticed my mistake. I'm trying with the correct rule [10:24] I had port 2222 forwarded to .11, and in my tries, the rules got mixed up -> my mistake... [10:25] Ok. [10:25] nocturne you using iptables for port forwarding> [10:27] so, corrected the rule [10:27] DNAT tcp -- anywhere alpha tcp dpt:smtp to:192.168.122.10 [10:27] added iptables -A INPUT -p tcp --dport 25 -j ACCEPT [10:27] You changed something else, too. [10:27] but still no luck [10:27] It says "alpha" where it used to say "anywhere" [10:27] yes added -d [10:28] the host has 3 public ip's [10:28] only alpha should accept traffic [10:28] nmap now shows the port open [10:28] Pastebin the output of "iptables-save" [10:28] You know that DNAT only changes the destination address, right? [10:29] The guest vm (I'm guessing these are VM's based on the subnet) will still see the real source IP. If you're multihomed, you may need to ensure that traffic goes back the same way it came from. [10:30] soren_ http://paste.ubuntu.com/1006196/ [10:30] I need to add an snat rule? [10:31] I have no way to know. [10:33] nocturn: Are you aware the ordering matters with iptables? [10:33] yes [10:33] -A FORWARD -o virbr1 -j REJECT --reject-with icmp-port-unreachable [10:33] -A FORWARD -d 192.168.122.0/24 -o virbr0 -m state --state RELATED,ESTABLISHED -j ACCEPT [10:35] this is incorrect? They are 2 different bridges. [10:36] Err.. [10:36] the mailserver is on virbr0 [10:36] Yeah, sorry. [10:36] I just got confused by all the repeated rules. [10:36] But there are definietely rules in there that'll never ever be used. [10:37] Very high up, you have: [10:37] -A FORWARD -o virbr1 -j REJECT --reject-with icmp-port-unreachable [10:37] -A FORWARD -i virbr1 -j REJECT --reject-with icmp-port-unreachable [10:37] ..but then further down, you have other rules that refer to -o virbr1 or -i virbr1, but by then, the packets have already been REJECTed. [10:38] Same for virbr0, although you don't have those REJECT rules quite as early. [10:38] I just commented out the rejects as a test [10:38] for both bridges [10:38] it made no difference [10:39] BTW, INPUT is irreleveant when you're doing DNAT. [10:39] When you're DNATing, you're forwarding traffic. [10:40] * soren_ needs lunch [12:16] how can I get a list of packages I have explicitly installed on a system? running "aptititide search '~i'" gives me the list of installed packages marking by "A" automatic packages, but it also includes packages which where installed together with the system. [12:22] is there any other place in ubuntu server 12.04 LTS version that controls the total number of connections into a server besides net.core.somaxconn?????????? [12:28] <_ruben> KingKatari: there's also a connection tracking limit as part of netfilter, but that only applies if there's any firewall rules present [12:29] any other [12:30] <_ruben> not that i'm aware of (which doesn't mean all that much really :)) [12:30] lol === matsubara-afk is now known as matsubara [12:54] man. [12:54] <_ruben> woman. [12:54] i have got to upgrade to quantal so i stop getting that annoying update manager popup. [13:10] <_ruben> hehe [13:22] soren_: I solved my issue. It was my isp blocking outgoing access to port 25! If you're ever in Antwerp/Belgium, I owe you a beer [13:26] smoser: you should be dog fooding anyways ) [13:26] zul, are you on quantal? [13:26] smoser: my server is [13:26] i eat dog food every morning for breakfast. [13:26] server. ppft. [13:27] "i have a virtual machine that i never use that is using quantal" [13:27] smoser: because on the internet no one knows you are a dog right? [13:27] hardly means anything. [13:27] primary work station, man. [13:27] :) [13:27] smoser: well i work on the sever so i dog food it :) [13:27] oh internet [13:27] smoser: ppphpppt...i dont think so [13:28] no one *knew* i was a dog, until zul spilled the beans. [13:28] smoser: i work for tmz [13:37] ruff [13:43] hallyn: ruff? makes me think of http://www.sadanduseless.com/2012/05/oh-you/ :) [13:45] highvoltage: nice collection :) [13:58] hallyn, On a very quick and not too sophisticated testing your libvirt-0.9.12-0ubuntu1 looks to be working (just installed on a 64bit quantal xen hv and booted one hvm via virt-manager (plus shutdown)). [13:58] smb: zul's, you mean :) cool, thanks! [13:58] hallyn/smb: cool [13:59] hallyn, whoever did it... :-P [13:59] That could help to ignore that strange ftbs on i386 that http://people.ubuntuwire.org/~stefanor/lp-ftbfs-report/historical/primary-quantal.html#main shows [14:21] hallyn: im going to upload libvirt today === dannf` is now known as dannf [14:36] zul: excellent. so dwarves MIR was approved? [14:36] hallyn: yep [14:37] yeah, that ftbfs (if it's the one i'm thinking of) was due to gnulib. hopefully it just doesn't happen in 0.9.12 [14:37] stgraber: around? [14:38] I'm doing the lxc patch to mount rootfs under /var/lib/lxc//root [14:38] But, I'll need to also do a mount --make-shared in src/lxc/conf.c for that dir [14:38] I'm looking for how to decide on whether to do that [14:38] (i can't do that when containers are using a shared dir) [14:40] I could just say "if strncmp(target, "/var/lib/lxc/", 13)" then make it rshared [14:40] kind of ugly, but should DTRT [14:40] Daviey: what version should we be calling the openstack SRUs [14:41] Daviety: for the snapshots? 2012.1~201205XX-0ubuntu2.1? [14:41] or, i could add a separate 'lxc.root.public = [true/false]" to the config, but that's kind of hacky too [14:43] hallyn: you could have something like lxc.mountfs = to override the default of /var/lib/lxc//rootfs [14:44] hallyn: then anything that uses a shared /var/lib/ will have to set it to something unique [14:44] stgraber: there already is a lxc variable for where to mount. So I'm just having the templates set that [14:44] Question is fo when that dir should be make MS_SHARED [14:44] but i think i've got it, [14:44] i'll just do it always so long as that dir != LXCROOTFSMOUNT [14:44] (where LXCROOTFSMOUNT for us is /usr/lib/lxc/root) [14:45] (lxc.rootfs.mount is the config variable for that, btw) [14:56] hallyn: is lxc-shutdown upstream? [14:57] looks like no [14:57] pretty sure i sent it. [14:59] zul: git$DATE-$(shortsum) [14:59] Daviey: ack [15:01] hallyn: ok. I replied to Christian regarding clean shutdown of container on lxc-devel, would be great if you could push the two lxc-shutdown patches. [15:02] ok, put on my list. [15:11] zul, here is fine. [15:11] smoser: i havent looked at the image but i assumed it was qemu-system-arm [15:13] he does say "highbank kernel" [15:13] which would imply arm kenrel and not qemu-arm [15:14] smoser: obviously we need more info :) [15:14] http://bazaar.launchpad.net/~dannf/+junk/make-highbank-ami/files [15:14] there is a qemu version for highbank [15:14] is reported the scripts used to make it. [15:14] dannf, ping [15:14] but neither packaged nor released yet afaik [15:14] zul: the deal is, libvirt automatically notices new systems [15:23] smoser: yo [15:24] ogra_: supposedly highbank support is all upstream, might even work in precise - but the upstream code isn't ideal for what i'm doing here (it boots kernel from host system, not from disk image) [15:24] dannf, so the ami you set up [15:24] i launch it, ssh in, and i'm in a qemu-system-arm vm, is that right? [15:24] smoser: that's correct [15:24] you modified the host to pass a single (ssh) port thorugh? or everything? [15:25] is there a way to get to the host? [15:25] smoser: all bridged through, no way to get to the host [15:25] oh, and btw, this is really cool, thank you for doing that. [15:25] thx [15:25] http://bazaar.launchpad.net/~dannf/+junk/make-highbank-ami/view/head:/mac-switcheroo.sh [15:26] dannf, ok. so zul was inteerested in reproducing this on a local system (obviously without mac_switcheroo) [15:28] zul: sure, let me know if you need any help w/ that [15:28] dannf, you should have hacked cloud-init in the "guest" [15:28] to get ports.ubuntu.com sources.list [15:29] although i guess we can actually pass that in as user-data. funny. [15:29] smoser: true [15:29] (userdata does work?) [15:29] * dannf should probably respin for that [15:29] i don't know if userdata works or no - i've a lot to learn about cloud-init [15:39] dannf, testing userdata [15:42] whats the best way to monitor dell raid cards in ubuntu these days? [15:43] i saw a post about a 3rd party repo to install the LSI megacli tool [15:43] not sure if theres a good dell method without installing the entire OMSA tools === nhandler_ is now known as nhandler [15:57] stgraber: (finally read the shutdown thread) boy some people want to make things complicated... [15:57] hallyn: as usual ;) [15:58] hallyn: I actually skipped most of Christian's e-mail when I saw that he didn't look at lxc-shutdown and just suggested he looks at it :) [15:58] there are probably irc logs of our discussions about whether to change lxc-stop [15:58] if i were smoser i coudl find them in a heartbeat [16:00] hallyn: my IRC logs say March 18th on this channel [16:00] i'm looking at mar 19 log right now :) [16:01] ok, so basically we didn't want to change lxc-stop behavior without upstream's consent. [16:01] (i'll reply) [16:05] stgraber: GAH! [16:06] if i make /var/lib/lxc MS_SHARED, then the kernel doesn't let me pivot_root under it :) [16:06] This is goin gto turn out more complicated than i'd thought [16:06] the kernel is being annoying... :) [16:06] And I hate to bring it up on lxc-devel right now, bc people are already in argue mode from other threads [16:07] this means that the only choice, if we want a shared root, is to chroot [16:07] uh, by shared i mean one we can maniuplate from the host [16:07] hm, no, let me try ONE more thing [16:10] dannf, for your record, user-data does work [16:10] and: [16:10] euca-run-instances --key brickies --instance-type=m1.large ami-aef328c7 --user-data-file=/tmp/my.ud [16:10] where '/tmp/my.ud' is http://paste.ubuntu.com/1006618/ [16:11] sets the mirror correctly and runs the 'runcmd' === Lcawte|Away is now known as Lcawte [16:11] smoser: cool! is there a ref doc for what i can do in user-data? [16:12] https://help.ubuntu.com/community/CloudInit [16:12] thx [16:12] hm.. [16:12] dannf, what does security.ubuntu.com do for armhf? [16:12] er... where do security updates come from? [16:13] i guess they come from ports too. [16:13] thats broken in the userdata above (ie, 'apt-get update' will fail due to still trying to use security.ubuntu.com) [16:13] good question [16:14] you could use a boothook to edit template file (just write it straight away with what you want it to have) [16:15] or even a runcmd if you weren't trying to install packages via cloud-config. [16:16] oh, snazzy :) [16:17] hallyn: I only just noticed http://git.kernel.org/?p=linux/kernel/git/torvalds/linux.git;a=commit;h=644473e9c60c1ff4f6351fed637a6e5551e3dce7 ! [16:17] anyone who wants can look at dannf's handywork if they're interested [16:17] be nice, it will die in 40 minutes [16:18] stgraber: man, if you get lxc-start to fail at the right time, you *really* pollute your mounts table [16:18] i think start.c should probably try to clean up... though that'snot a trivial matter [16:19] anyone interested can: [16:19] ah yes. we have part of user namespaces :) [16:19] ssh ubuntu@ec2-23-20-232-135.compute-1.amazonaws.com [16:19] with password 'abcdefg' [16:19] but that instance will die at 17:00 UTC (41 minutes from now) [16:19] smoser: permissiond enied (publickey) [16:19] hallyn: hehe, I've got a fix-mount-table script I use when I'm working on arkose, basically a while loop iterating through /proc/mounts and unmounting everything until all entries are gone :) [16:21] stgraber: gah. yes. we can't pivot_root into MS_SHARED dir. what do you think about introducing a lxc.use_chroot option? [16:21] maybe i should just bring it up on lxc-devel [16:21] hallyn, try again [16:22] hallyn: is chroot "safe" now? and if it's, why still use pivot_root? [16:22] stgraber: no. but we have apparmor for that :) [16:22] it seems that if you enable empty passwords, rather than enabling password auth in openssh's config file, it doesnt actually enable password auth. [16:22] (whoops) [16:22] though, i haven't tested that. i probably should [16:22] smoser: still hanging on motd :) [16:22] its not fast. [16:23] nice. highbank [16:23] you're runnign qemu-system-arm inside a ec2 m1.large [16:41] hallyn: its uploaded [16:49] zul: thanks [16:59] Anyone know of a ppa for jboss as 7.1? [16:59] stgraber: when i chroot instead of pivot_root()ing, mountall is not happy. tries to mount all the things you taught it not to mount [16:59] It looks like jboss as 4.2.3 is the only version included with ubuntu [17:00] hallyn: can you dump /proc/mounts before mountall runs? something must be quite wrong in there for it to try and mount them all... [17:01] stgraber: well here is mounts after mountall runs... http://paste.ubuntu.com/1006687/ [17:01] hallyn: I didn't put any container specific logic in there, I just taught it not to mount something that'd hide an existing mount (/dev being mounted when /dev/pts is already mounted for example) [17:01] right, and chroot doesn't change the /proc contents. so i guess that's a reason chroot does not suffice. IIUC === nxvl_ is now known as nxvl [17:02] yeah... we'll have to stick with pivot_root then [17:05] adam_g: im in the middle of collecting bug numbers for the openstack SRUs [17:05] zul: cool. they should all be in the git commits, no? [17:06] adam_g: well there is two that doesnt have a bug assoicated with them [17:06] for nova at least [17:06] SpamapS: which ones [17:06] er [17:06] ^ [17:06] SpamapS: around? [17:07] adam_g: one is for generating a changelog for the git commits when you run sdist, and the other is make tests past on osx (???) [17:14] adam_g: I am, wassup? [17:15] wassup is you're supposed to accept my SRU packages .... [17:15] :-) [17:15] ScottK++ [17:30] stgraber: which means that if we want to support shared mounts between host and container, we need to go back to the idea of a /shared [17:30] I'm going to let that sit for a bit and get back to it on wed or thu [17:34] CIFS mounting issue on Ubuntu server... I have a drobo NAS with a share called central. I can 'smbclient -L drobo-primary' and see the share called central. However when I run the command 'sudo mount -t cifs -o username=bustabust,password=mypass //10.10.1.150/central /mnt/drobo' error says Unable to find suitable address. Any ideas? I've googled the heck out of this thing [17:37] zul: drat, I should have warned you. The debdiff for 0.9.8-2ubuntu19 got lost in the 0.9.12 upload [17:38] doh...link? [17:38] ill upload ubuntu2 this weekend [17:41] zul: it was the patch attached to https://bugs.launchpad.net/ubuntu/+source/libvirt/+bug/1001625 [17:41] Launchpad bug 1001625 in libvirt "Guest clock stops after live migration on Ubuntu 12.04 hosts" [Medium,Confirmed] [17:41] (or just dget dget https://launchpad.net/ubuntu/quantal/+source/libvirt/0.9.8-2ubuntu19/+files/libvirt_0.9.8-2ubuntu19.dsc) [17:42] hallyn: thanks ill get that one in [17:42] zul: thanks. (my fault) [17:42] im guessing any SRUs as well [17:47] zul: ? [17:47] what do you mean, you're guessing sur's? [17:47] hallyn: is there an SRU for libvirt in precise or am i halucinating [17:48] this bug needs to be SRU'd for it yes. i haven't done that yet [17:50] Hello, I'm trying out MAAS with LXC. I've gotten to the point where Maas is fully installed in a container, but after adding a node, it says that it's "Commissioning", and stays that way. Any help? [17:55] zul: i need to run out but ill look through and confirm that bug list when im bcak. we should try to script something that generates that for us at point [17:55] adam_g: yeah im working on a script now :) [17:56] zul: im sure whatever scripts they have to parse that stuff out of gerrit proposals is available somewhere in the openstack-ci repo [17:56] back in a while === dendro-afk is now known as dendrobates === dendrobates is now known as dendro-afk [18:39] how does one report spam in a comment on launchpad? === dendro-afk is now known as dendrobates === dendrobates is now known as dendro-afk [19:14] smoser: do you have a cloud-init patch for switching the default apt server, or should i just do a sed hack? [19:15] ... or should i actually help out and send you a patch taht fixes the bug :) === dendro-afk is now known as dendrobates [19:21] dannf, there is no pathc in lcoud-init (you're welcome/encouraged to fix upstream to chose better) [19:21] the cloud-config is actually a supported path (for internal mirrors or othe rmirro) [19:23] but the security. was not thought about. [19:23] (the fact that security.ubuntu.com is not used for ports). === dendrobates is now known as dendro-afk === erichammond1 is now known as erichammond [19:52] hi, this ec2 image ami-5c892f35 in us-east-1 doesn't work [19:52] it's quantal [19:52] here is console output: [19:52] http://pastebin.ubuntu.com/1007018/ [19:53] I got that id from http://uec-images.ubuntu.com/quantal/current/ [20:01] hello, iam learning to work with linux . can someone pls provide me ssh access to some server ? [20:03] hello [20:03] is it possible to set a hostname without rebooting the server ? [20:04] hujifalak: search for "free ssh" - there are still several available [20:05] hujifalak: install your own virtual machine? [20:05] hujifalak: Amazon AWS/EC2 has a free tier that you can use to run and experiment with Linux: http://aws.amazon.com/ec2/#pricing === dendro-afk is now known as dendrobates [20:08] erichammond i wanted to try it but it requires credit card information even when they are not charging anything ? [20:10] hujifalak: correct. If you use resources that are not covered by the free tier, you will be charged for them. [20:29] stgraber, hi. I'm updating bug 994752's description per your request. In preparation for that, I was looking at the quantal packaging branch, and noticed you took out lxc-ip in favor of an in-line function within lxc-start-ephemeral. lxc-ip is actually quite nice to have for other reasons as well. Is there a way we could advocate for getting it added back? Or is getting it upstream our only chance? [20:29] Launchpad bug 994752 in lxc "lxc-start-ephemeral's use of dhcp lease table is fragile" [High,Fix committed] https://launchpad.net/bugs/994752 [20:30] gary_poster: no :) [20:30] stgraber, which one [20:30] the new lxc-attach will support attaching to only the network namespace if you want [20:31] and the API will have a call to get all the container's IPs that will likely be added to lxc-info [20:31] stgraber, in bash? [20:31] so there was no good reason to keep lxc-ip in 12.10 just to replace it in a few months (and backward compatibility being a pain, it was easier to get it out before anyone noticed :)) [20:31] heh [20:31] if lxc-info then C [20:32] * gary_poster wants more Python and less bash [20:32] right, we have patches on lxc-devel now that let you do "lxc-attach -s NET -n container -- ip -4 addr", we're also working on API design now and I'm trying to get a get_ips() call in there that lxc-info can use [20:32] for 12.10 I'll also be providing a python binding for the new API [20:33] so after these are done, there really shouldn't be any need for lxc-ip :) [20:33] we use Virtualbox on servers :) [20:34] how long can we stay at 10.04 LTS ? [20:34] stgraberm, sounds great (though a mild shame we went to the trouble for naught). The one thought I have for lxc-info's IP stuff is that it would be nice if it had a "just get the IP and you don't have to use sed/awk to parse the output" option. The sed/awk stuff always feels pretty fragile to me [20:35] gary_poster: sure, that's why I want to get the IP list in the C API so we can have it returned by lxc-info and from the python module [20:36] gary_poster: so you should be able to do: [20:36] import lxc [20:37] mycontainer=lxc.Container("mycontainer") [20:37] mycontainer.get_ips()[0] [20:37] stgraber, that sounds awesome :) [20:38] stgraber, that's great! [20:39] stgraber, one other only mildly related thought: in lxc-start-ephemeral when we start up 32 instances of these (and we do, with 24 planned for production) the timeout gets out to 2.5 minutes before we can actually connect. Is there a chance we could squeeze in a configurable timeout for bug 994752 also? [20:39] Launchpad bug 994752 in lxc "lxc-start-ephemeral's use of dhcp lease table is fragile" [High,Fix committed] https://launchpad.net/bugs/994752 [20:39] 2.5 minutes before we can connect to all of them, I should say [20:41] essentially, make TRIES something you can pass in [20:43] gary_poster: I have no problem adding a parameter for that in quantal, though that's technically a new feature so I'm not sure we can really SRU that back to precise [20:44] gary_poster: isn't there a magic command that lets you run a given command when your load average is below a given threshold? I think you could use that to fix your problem (assuming it takes that long because of all the containers starting at the same time) [20:45] stgraber, we need it to truly address that bug, is the issue. Alternatively, we can simply bump TRIES to something gigantic...hm, that sounds interesting, lemme search for it [20:46] niceload is close but no cigar... [20:48] and also it is not available in ubuntu afaict [20:52] gary_poster: parallel -l 1 -- "echo 1" "echo 2" [20:54] stgraber, nice, thank you. We'll play with it. [21:01] stgraber, there are still some problems with that. I'll mention them in the bug description and we can go from there. [21:01] thank you again [21:22] smoser: if i do ec2-run-instances -f userdata.sh -n 3 ami-whatever, is [0-2] or [1-3] passed to the userdata scripts by chance? [21:24] hallyn: You will want to query the ami-launch-index from the user-data script. [21:24] just echo $ami-launch-index? [21:24] hallyn: It'ts available from the instance at: http://instance-data/latest/meta-data/ami-launch-index [21:25] or using a tool that queries that for you. [21:25] starts at "0" [21:25] erichammond: awesome, thanks [21:28] adam_g, is there a document around quantum which can help me deploying it on precise [21:31] koolhead17: not that im aware of [21:32] adam_g, any plan to have it in future :P [21:33] koolhead17: probably at some point. i hope to take my first dive into quantum next week or so [21:33] that be great [21:36]