[00:01] zroysch: yeah I think [00:01] SpamapS: ping, whenever you have 5 mins, upstart question [00:02] anyone setup nat on an ec2 instance? [00:02] lynxman: fire away, but I really do only have 5 minutes [00:03] SpamapS: cool, will take 3 :) [00:03] SpamapS: trying to setup an upstart job for a process that was using an init.d wrapper, and I need to user the wrapper (it's python calling to internal functions) [00:05] lynxman: you can have an upstart job with no actual running process to track.. just use pre-start to call the 'thing start' and pre-stop to call the 'thing stop' [00:05] SpamapS: like this http://pastebin.ubuntu.com/667787/ [00:05] SpamapS: smoser suggested the expect fork, right now the upstart job hangs indefinitely as it is, I guess it's due to that [00:06] For that, I'd just drop the expect fork, and move the script to pre-start script [00:06] The upstart job will work like a place holder for boot/shutdown... [00:06] but otherwise will get out of the way. [00:06] SpamapS: cool, was just a bit concerned of using pre-start instead of start [00:06] SpamapS: thanks :) [00:08] lynxman: its a little obtuse, but works wonders for getting things into upstart without fully migrating them. :) [00:08] btw, your start on is *WAY* too verbose [00:08] just start on runlevel [2345] [00:08] SpamapS: hehe yeah we were discussing it with smoser last Friday so I wanted to be safe [00:08] both of those events are guaranteed to happen before runlevel 2 [00:08] SpamapS: cool, will reduce that as well [00:09] plus this way when you go from runlevel 1 -> 2 .. it will actually get started back up. [00:09] so what does 'status' show for a job that has no job ? [00:09] SpamapS: neat [00:10] smoser: I guess "nothing to see here, move along" [00:10] and why would you not want upstart to correctly track a pid ? [00:10] how would it know if the thing died ? [00:10] smoser: its start/running but no pid [00:10] i'm generally curious [00:10] generally and genuinely [00:10] smoser: because upstart cannot correctly track most pids [00:10] They've come up with a good plan for that finally. [00:11] SpamapS: when are we switching to systemd? [00:11] * lynxman hides [00:11] instead of 'expect fork' it will be 'expect exit' .. which will watch for the parent's fork, but then wait for it to exit. [00:11] lynxman: http://pastebin.com/kWFtcb30 [00:12] smoser: tracking pids is over rated, some services actually do it all by themselves. ;) [00:12] anyway [00:12] I'm late.. good luck! [00:15] zroysch: out of ideas then, you definitely have something weird going on there... [00:15] well i do have the data backed up [00:15] i guess i could just start over from scratch [00:16] zroysch: if that's an available option I would go for it, and make both fd this time ;) [00:16] zroysch: software raid gets confused way too easily [00:28] lynxman: i've never run into this problem before. [00:30] zroysch: me neither :) [00:37] http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=500309 [00:37] hmm [00:37] Debian bug 500309 in mdadm "mdadm thinks disk is not large enough to be added, but it is (v1 superblock)" [Grave,Fixed] [00:37] old bug resurfacing? [00:37] i have mdadm v3.1.4 [00:38] does ubuntu server come with lesser repositories? [00:38] cant seem to find git-core [00:39] nm, after an update i got it [00:49] an update will not get you a new package [00:52] zulax: it was renamed to "git" about twelve months ago [00:52] ok [01:06] Having a issue with nginx 1.0.5, php-fcgi with php-fpm over a socket is using its cache a little to aggressive, it seems that ALL requests are being send to the same file, which is the one cache file that was made with the first request, my configs http://pastebin.com/RND2E9fY http://pastebin.com/vxbHZf3W http://pastebin.com/xKaKAAgH [01:06] Is your PHP app sending the appropriate cache-related header fields? [01:07] my first test was apc.php [01:07] but if i do apc.php first [01:07] then do test.php [01:07] it will show what apc.php sent to nginx [01:07] Dunno, then [01:08] You could try #nginx and/or ##php, too [01:08] http://survivorzero.com:82/apc.php then http://survivorzero.com:82/test.php [01:08] test.php doesnt exist [01:08] and its showing the contents from apc.php [01:09] http://paste.debian.net/126447/ [01:11] is the only cache file [01:11] http://pastebin.com/aLkPwh8p [01:11] no matter what you call it will serve that file [01:12] whats fun is that d41d8cd98f00b204e9800998ecf8427e = md5("") [01:12] thats the hash its stored in the cache [01:12] so... i dont think the cache has a clue to what the request URI is [01:15] New bug: #827681 in cobbler (universe) "Cobbler does not generate pxelinux.cfg files for ARM" [Medium,In progress] https://launchpad.net/bugs/827681 [01:18] ARM supports PXE? [01:18] I thought PXE was an x86ism [01:18] why wouldnt it? [01:19] As opposed to what I *have* seen on ARM/uboot, which is to hard-code the TFTP server address and the filename [02:12] New bug: #827755 in tftp-hpa (main) "package tftpd-hpa 5.0-21ubuntu2 failed to install/upgrade: subprocess installed post-installation script returned error exit status 2" [Undecided,New] https://launchpad.net/bugs/827755 [03:45] hallyn: ping, I think there's a serious problem with your cgroups/lxc [03:45] ^implementation [03:45] NCommander: Error: "implementation" is not a valid command. [03:46] * NCommander facepalms [03:46] NCommander: whats aup? [03:46] lifeless: so lxc uses/requires cgroups to make sure containers are separate from each other, right? [03:47] yes [03:47] key component of the implementation [03:47] lifeless: right, but then why does LXC work if cgorup namespaces are unavailable? (they were removed with 39(?)) [03:48] http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=a77aea92010acf54ad785047234418d5d68772e2 ? [03:48] The replacement is clone_process, but that's not enabled by default and there's no way to get cgroups to use them with cgconfigparser [03:49] lifeless: as it stands, I think cgroups aren't being properly used, but LXC doesn't emit a warning or error (aside from a misleading message in lxc-checkconfig) [03:49] hmmm [03:49] I think this is out of my expertise ;) [03:50] (this is true on i386/amd64 as well) [03:50] lxc is just a thin shim around cgroups [03:50] so I think ns_cgroup isn't needed [03:50] twb: right, but namespaces areno longer available [03:50] and never really was, it was a convenience [03:50] But a default lxc configuration will not virtualize much at all -- e.g. no namespace (containerization) for the filesystem, network stack, etc. [03:50] lifeless: no, they were replaced with clone_instance [03:51] whichis a mounting option when you create the cgroup (there is no way ot specify the necessary flag via cgconigparser) [03:51] I see your point [03:51] like I say, we've reached the edge of know-how. [03:51] twb: lxc-checkconfig says namespaces are 'required' although if clone instances are enabled, it says 'enabled' with a new line [03:52] NCommander: can you file a bug @ https://bugs.launchpad.net/ubuntu/+source/lxc ? [03:52] NCommander: serge is pretty active there [03:52] lxc-stop appears to be broken like this [03:52] lxc-stop is probably broken for unrelated reasons [03:53] twb: lxc-stop works properly if clone_children is set on the cgroup instances [03:53] Of course my experience is limited to .32 so I may be talking bullshit [03:53] NCommander: thats extremely interesting [03:57] I got lxc-stop to work after a few tries. No idea why [04:01] NCommander: so it 'works' first time for me, always has (though it whines but actually does its thing) [04:01] NCommander: but the lack fo power-off means that processes don't go down quite as they might expect [04:06] twb: lifeless: https://bugs.launchpad.net/ubuntu/+source/lxc/+bug/827798 [04:06] Launchpad bug 827798 in lxc "LXC works without warning regardless if cgroup namespaces are properly available" [High,New] [04:16] New bug: #827798 in lxc (main) "LXC works without warning regardless if cgroup namespaces are properly available" [High,New] https://launchpad.net/bugs/827798 [04:18] the bot is lagged :-/ [05:49] slow server 10.04 in hyper-v vm, can anyone help? [05:57] nvm, thank you google [06:03] rickjaruiz: what was the issue? [06:03] Rather, what was the cause and the fix [06:04] disable the frame buffer module: edit /etc/modprobe.d/blacklist-framebuffer.conf and add the following line: blacklist vga16fb [06:05] every line to be drawn for a page refresh [06:05] God damn I hate that module [06:05] And I hate that it's on by default [06:05] All it gives you is 80x30 instead of 80x25, it's not even useful when it DOES work :-/ [06:06] anyway to turn off before installation? [06:06] rickjaruiz: during lucid install, no. Post-install and I *think* in newer d-i installs, yes [06:06] or during installation [06:06] There's a bug in the lucid installer that prevents all documented techniques from working [06:06] yeah it makes install take twice as long [06:06] Oh, also of course you can do the install via serial instead of VGA [06:06] 11.04 installed in 10 minutes and 10.04 installed in 20 [06:07] im running it as a virtual machine in hyper-v [06:07] If Microsoft's virtualization technology can't emulate a COM port, it ought to be scrapped. [06:08] how can i get a remote gui file access? [06:08] I don't know or care; GUIs are not a server issue. [06:08] or whats a good way to transfer files between serveres [06:08] i new to linu [06:08] linux [06:09] SFTP. [06:09] how do u install in server? [06:09] Or scp, which is pretty much the same thing. [06:09] Both are components of SSH [06:09] so just udo apt-get install openssh-server? [06:09] *sudo [06:09] rickjaruiz: if you have not already done so, you should read the ubuntu server guide (mentioned in /topic) [06:09] rickjaruiz: yes. [06:10] i mgoing through it and asking questions at the same time [06:10] helps me remember [06:12] OK [06:12] but yeah so much to learn from that guide [06:12] i have to take it in sections === smb` is now known as smb === otubo is now known as otubo[AFK] [06:37] Is there any known problems / bugs affecting 11.04-Server and Samba? I've been tryng to setup a simple share for hour, and its just not happening, even with SWAT [06:41] Evening.. Is there a way to have initrd get all ip's from dhcp servers before doing it's iscsistart ? [06:43] for now , the moment it booots , its gets an ip , then immediately tries to make iscsi connections. ending up being unable to connect. [07:23] Daviey: about my rootwrap, I think I'll have to defer to Essex, I'm a bit busy atm [07:24] Daviey: on the plus side, this will allow to generally discuss security improvements at ODS [07:24] Daviey: and nobody strictly asked for it in MIRs afaict [07:24] Daviey: thoughts ? [07:26] ttx: Well nova MIR hasn't been ack'd yet [07:26] Daviey: sure but that wasn't part of it -- anyway, I commit to doing it for Essex, so would be in LTS [07:26] But yes, apparmour and sudo limitation seemed more favourable to me, but i understood i was overuled on that. [07:27] ttx: How much work is involved forking what we had before? [07:27] Daviey: I'm almost done with the wrapper -- the devil being in the "almost" [07:27] Daviey: also I feel a bit uncomfortable pushing it now [07:28] ttx: Having the proof-of-concept published somewhere would allow it to be mentioned in docs. [07:28] Daviey: first shot is at https://code.launchpad.net/~ttx/nova/privsep2 [07:29] but I can't really spend more time on it this week(s), and would like to have it properly testcased [07:29] so to do it right, I need more time -- and that time would allow us to discuss how to do it [07:29] ttx: Where will this live eventually, packaging branch? contrib/ in upstream or a core part of upstream? [07:30] morning [07:30] hey RoyK [07:30] Daviey: if we follow what I started, it will live in bin/nova-rootwrap in nova + contrib filter files in Nova. [07:31] again, some discussion around it at ODS sounds like a good idea [07:31] I need to work on the registration site now [07:31] ttx: Why did you start it from scratch? [07:31] Daviey: as opposed to ? [07:31] forking what we used to use === Guest69853 is now known as Ard1t [07:32] (1) because there is no way I can write security code in C right and (2) I'd like it to live in Nova [07:32] ok, thanks [07:32] so that someone adding a sudo command can fix the file [07:32] ttx: Thanks for letting me know. [07:32] Daviey: that was the result of a discussion with mdeslaur, fwiw [07:33] anyway, I prefer not to rush security code :) [07:33] i'll catch up with mdeslaur regarding this, hopefully today. [07:34] Daviey: I think it's better to discuss how it's best done, and implement it with plenty of safeguards in essex [07:34] ttx: yah [07:34] rather than adding one more branch merge proposal to our FF queue [07:34] (deadline for feature merges being Monday) [07:34] If I had one more week I would probably have done it [07:35] aw frack [07:35] I was a bit optimistic on how much free time I'd get :) [07:49] hey guys what is needed to do a remote network installation [08:06] eagles0513875: a good preseed :) [08:06] ? meaning what [08:06] eagles0513875: entirely hands off, or over SSH> [08:07] ? [08:07] over ssh [08:07] https://help.ubuntu.com/community/Installation/OverSSH [08:07] ^^ not the best way IMO. [08:07] Daviey: Error: "^" is not a valid command. [08:08] https://help.ubuntu.com/community/Installation/NetworkConsole <-- much better [08:08] humm ok [08:09] i just dont want to migrate my keyboard nad monitor over to my room where my server is at hence why im askin [08:10] Daviey: so i boot the iso or the bootable usb on the machien im currently on right then follow the steps from there? [08:11] eagles0513875: If you are doing this for the first time, and mainly doing it to save time.. you are better off moving the monitor and keyboard. [08:11] ok [08:11] :-/ [08:12] Daviey: why, because of the 1:1 correspondence with the udeb name? [08:13] Oh, nm, those articles are different. Thought you were discussing the name of the article [08:14] hehe [08:14] The overssh version is a pita [08:14] last question guys i need to back up my current email server which is running dovecot postfix and spamassassin [08:14] Better off just put the mini.iso kernel and ramdisk and preseed in the /boot of the unix system you want to blow away [08:15] thing is i need to change disks i bought 2x1tb hdds for raid [08:15] twb: Yeah, suggesting against debootstrap - in favour of using the ssh support within d-i [08:15] in regards to my question above relating to email [08:15] Then you reboot, pick d-i from the grub menu, and it'll load into RAM. At that point, you have everything in RAM and can blow away the HDD you booted from [08:15] i just need to backup the postfix conf file the dovecot config and spamassassin config maildir correct and i should be fine [08:15] debootstrap has a whole bunch of hairy edge cases like creating /etc/hosts, and it's a PITA to remember all of them [08:16] And if you're using grub, that's a bloody huge nightmare to install via a chroot, unlike extlinux [08:18] eagles0513875: as a general rule, I would back up all of /etc, tar -cfp etc.tar /etc, then save the .tar file where you can get to it. [08:18] greppy: why though? [08:18] guess it woudlnt hurt [08:18] wouldnt [08:19] because there have been times when I have remembered that I needed something AFTER I wiped the disk :) [08:19] this disk isnt getting whiped [08:19] its going into storage [08:19] atm my server only has one 750 hdd [08:19] and having groups/passwd/shadow, etc backed up is handy [08:19] im putting in or hoping to put in 2x1tb hdd raid 1 [08:20] agreed [08:20] :) thanks for your input greppy :) [08:20] *nod* still easier to have a tar ball of all of /etc instead of having to shove the drive into another system to get data off of it. [08:20] im gonna take my server offline for a sec to check something [08:38] guys is it easy to setup raid if i add a 2nd hdd at a later point in time [08:39] what option after ifconfig shows me all avaible interfaces not just the up ones [08:54] <_ruben> philipballew: don't use ifconfig, it's been depreciated for ages, use the 'ip' command instead .. 'ip link' for all interfaces, 'ip address' for all ip addresses on thos einterfaces, 'ip route' for all routes, etc [08:54] <_ruben> 'ip help' and 'man ip' for more info :) [08:54] _ruben, ive never even herd of that command. its better? [08:55] _ruben: havent heard of that command either [08:57] <_ruben> 'ip' offers way more options than 'ifconfig' and is less braindead .. ifconfig for instance can't see multiple ip addresses per interface, only ip aliases, which basically are a nasty workaround to the actual problem [09:00] ifconfig can see multiple interfaces per interface, no problem [09:03] ip's easier to read, too [09:03] <_ruben> multiple interfaces per interface ? :) [09:04] http://paste.debian.net/126482/ [09:04] <_ruben> nice example of flaws: http://paste.ubuntu.com/668056/ [09:05] Yep [09:05] this conversation is still goin? [09:05] nice! [09:06] ifconfig's only good feature is that it's available on SUS systems you might have to deal with [09:06] Like say SCO or Solaris [09:07] Although ISTR they don't have a route(8) that does what you expect, for some reason that functionality is in netstat(8) [09:33] <_ruben> twb: hence lots of people using netstat -r on linxu as well :) [09:33] I still use netstat -rn for things like the routing table [09:34] <_ruben> yuck [09:34] <_ruben> `ip r` is so much shorter to type and cleaner output-wise :) [09:35] netstat is clean enough and works on all the platforms I use as a standard format [09:35] <_ruben> and netstat -rn is semi-braindead too, doesnt list source ip address for instance :) [09:35] <_ruben> then again, to each his own :) [10:45] New bug: #827922 in tomcat6 (main) "On Tomcat6 shutdown web apps do not get shutdown" [Undecided,New] https://launchpad.net/bugs/827922 [10:55] hmm apparently one cannot edit their lp-comment? [10:59] <_ruben> sounds more appropriate to be asked in #launchpad i guess :) [11:03] reisi: no, editing comments isn't possible right now :/ [11:03] well if there was a way i guess someone would had shouted, no need to go bully the launchpad crowd :) [11:04] reisi: https://bugs.launchpad.net/launchpad/+bug/80895 [11:04] Launchpad bug 80895 in launchpad "comments on bugs/answers/merge proposals/etc cannot be edited" [Low,Triaged] [11:42] hey guys , how would i add a file to a .war file , i tried tar , but it can use war files [11:58] tdr112: war files are zip files, just rename, unzip, add file, rezip, rename [12:00] thanks lynxman [12:25] I'm running ubuntu server 11.04 on vmware player. I added a second interface in /etc/network/interface and now I can't resolve outside network addresses. Any insight into this? [12:25] Added the second interface by adding the following lines: #Host-Only / auto eth1 / iface eth1 inet dhcp [12:50] Hey, just a quick question: is there something like the ubuntu-desktop^ (task) for the server seed? ubuntu-server^ does not work. Just may be handy sometimes as at least on the desktop normal upgrade sometimes "forgets" to install vital things... [12:51] ah, it seems like /etc/resolv.conf is getting re-written [13:02] smb: Well, there is a task simply called server, as well as a set of more specific server tasks (openssh, postgres, etc) [13:02] tasksel --list-tasks [13:03] Even if they might not give you the exact same thing as the server install seed. [13:03] (Don't see anything about the server kernel for one thing.) [13:04] andol, Ah, cool. So it seems there already is one thing missing here... Well, I usually know when I am missing the kernel... ;) But other things not so much === otubo[AFK] is now known as otubo [13:11] ... hm, not that those would force back useful things like tasksel or command-not-found which were probably removed due to broken dependencies. But ok, I can bring those back... [13:31] Daviey: did you get my comment about -enroll? [13:31] and good morning [13:33] hggdh: i did not :( [13:33] hggdh: where did you send it? [13:34] Daviey: lp:~hggdh2/+junk/cobbler-enroll2 [13:37] hggdh: will look shortly [13:37] How is it looking> [13:39] Daviey: seems good (but I still have no clue on how to do the actual config via a postinstall [13:40] which is to say, apart from the postinstall (and a man page for it, lintian is not very happy about lacking it), it is ready [13:42] hggdh: db_get :) [13:42] hggdh: you rock [13:44] Daviey: send me what you have, and I will try it [13:44] hggdh: not done yet [13:45] Daviey: k. BTW, I changed the templates slightly, and changed the NIC dialog to a selection [13:46] hggdh: ooo, nice [13:46] hggdh: so it lists all the NIC's? [13:46] hggdh: how much code did that take? [13:46] Daviey: 3 lines in the template :-) [13:47] I put the logic for the nic's as much as possible into C, to try and keep it small. [13:47] hggdh: wow. Is that multi-select? [13:47] yes [13:48] Daviey: one thing we might ant to look at is if the user may select more than one interface [13:48] s/ ant/ want/ [13:48] hggdh: yeah, i'm not going to be a perfectionist over that TBH [13:49] heh [13:49] the main target audience is using it to preseed; or selecting "all" [13:49] man, perfection is an unattainable dream [13:49] 'it works' is almost perfect [13:50] Hah, landing it is a dream :) [13:50] lol [13:50] brb [14:11] New bug: #828036 in mysql-5.1 (main) "MySQL daemon keeps dying and restarting when using ssl connections" [Undecided,New] https://launchpad.net/bugs/828036 [14:20] zul, so today (not sure whether it is because its Wednesday or because I pulled int the virt-host task which added libvirt-bin) bonding comes up ok, though the pbond0 that used to be created is not. PVM seems ok with net. HVM started with xm does fail/hang early trying to do large order allocations of some kind. Started with xl comes up to the point where the IDE emulation wait and resets itself endlessly... [14:20] zul, Guess I check that again tomorrow... :-P [14:23] Daviey: ping [14:23] RoAkSoAx: pong [14:24] Daviey: ok so I got ARM booting, not installing thouhg :) [14:25] Daviey: but it is in process [14:26] New bug: #828047 in postfix (main) "postfix init script copies smtp_tls_CApath /etc/ssl/certs to /var/spool/postfix/etc/ssl/certs/etc/ssl/certs" [Undecided,New] https://launchpad.net/bugs/828047 [14:33] hello! I have a 11.04, kvm based VM server with a handful of VMs running. Some should only be accessible via a private network, but not externally addressable, others should get external IPs routed through the VM server. I have the private net working fine (though I did not set this up), and I have seen other systems with a bridge interface setup to allow VMs the ability to hop on the external network/subnet and get DNS pointing at them direct [14:33] I'm not sure I fully understand how the bridge and VM server ought to be setup to allow for that [14:33] * patdk-wk uses bridges for all of that, private and non-private [14:33] guidance / topics to research / etc are all appreciated [14:34] patdk-wk: could you give me an example? is this all configured in /etc/network/interfaces? [14:34] or is there VM configuration as well (aside from a VM having brX as its source?) [14:34] in my case no, I did it manually [14:34] how do you mean? [14:35] vm's would never have a brX [14:35] not as a source if? [14:35] manually = not using a script [14:35] patdk-wk: would you consider /etc/network/interfaces a script? [14:36] crazygir, /etc/network/interfaces is a config for a script, so yes, it's using a script [14:36] does that configuration (your manual method) persist across reboots? [14:36] do you use puppet for system management at all? [14:36] cause I don't really see that working if you are leveraging puppet [14:37] that said, I would be curious how you did it [14:37] I'm sure I can figure out how that translates [14:37] who said I used puppet? [14:37] I'm just going into uncharted territory at this point and unsure how the pieces I understand connect [14:37] I don't think puppet even existed in 8.04 [14:37] it seems to be documented pretty good in the manual: https://help.ubuntu.com/community/KVM/Networking [14:37] I didn't say you did, I asked if you did [14:39] so using bridges is the way to do this? [14:39] using bridges is an easy way to do it [14:39] Daviey: would you be able to generate the list of bugs we have in http://status.qa.ubuntu.com/reports/ubuntu-server/release-bugs.html other than using tags? [14:39] dunno if it's the best for your case or not, but generally it's simple and easy, and repeatable [14:39] * CrazyGir nods [14:39] if you just think of each bridge as a seperate switch [14:39] and you plug the vm's into those switchs (bridges) [14:40] fantastic, I'll dig a little deeper here and see how well this will fit [14:40] yea, that makes sense [14:40] but how to create that in linux does not always make as much sense to me [14:40] I would be able to work my way through most all of this setup on OpenBSD with the manpages only [14:41] that doesn't happen in linux for me for some reason :) [14:41] * CrazyGir shrug [14:41] either way, I'll see where this gets me, thanks patdk-wk [14:41] the moment i add a second NIC to a system that boots over iscsi , initrd screws me over. are there docs somewhere on how to force initrd to either pause till both my nics have an ip before continuing with iscsi , or force it to only start "iscsistart" when a particular nic has an ip. [14:42] i have tried scsi-if=ethX [14:42] but nada [14:43] if i stick in the cable in the second nic it's broken. [14:43] heh [14:43] i have tried 3 different systems btw [14:43] second nic was added after the first one? and is normally called eth1? [14:43] it's the same [14:44] if i install with eth0 or eth 1 as a primary [14:44] or if i do an install with only 1 and then add it later [14:45] even if ubuntu server is instyaled and configured with only 1 nic , the moment i add the second and it has a cable , it fucks up in initrd [14:45] it gets an ip for the other nic [14:45] and tried to contact my target [14:45] through the other interface [14:45] ouch [14:46] which is on another subnet and has no iscsi targets [14:46] guess the iscsi script grabs first to get dhcp working [14:46] dunno, would have to check the iscsi initrd scripts [14:47] it should wait till all nics are up .. [14:47] you could have it run last? [14:48] what do you mean Crazy ? [14:48] Yes , it all works peachy on all systems untill i stick in another card [14:48] gxt4, not really, it should know what one is used for iscsi maybe [14:48] brands? [14:48] Daviey: is libcgroup not in the list of packages server team watches? I don't see new bugs announced here by the bot... [14:48] smb: cool i hope to get to it next week at a conference this week [14:48] well yes [14:49] @patdk-wk initrd should keep a record of which nic the target is on [14:49] zul, ack [14:50] @patdk-wk aparently it doesn't .. [14:50] Gxt4: can you do a static network configuration [14:50] hallyn: no.. not as yet, you think it should be? [14:50] @crazy once it's past initrd it's all peachy [14:51] @ i can configure ip's via dhcp or fixed then [14:51] dunno, I'm not a ubuntu guru [14:51] @ the problems are before os is loaded [14:52] Daviey: well it affects lxc and libvirt. so maybe. Of course, so do most things in foundations :) [14:52] hallyn: what does? [14:53] libcgroup [14:53] zul: ^ [14:53] ah ok [14:53] hallyn: I wonder if this is best on your personal list? [14:53] but maybe it's just my problem and no one else here woudl care to see them [14:54] hallyn: The rest of us will no doubt have NFI how to triage those issues. [14:55] Daviey: when it breaks it can break your whole openstack/uec/ensemble/whatever stack... it definately seems server related to me. [14:55] hallyn: argh, ok [14:56] (as it's doing right now - ask smoser :) [14:56] Daviey: but ok, if noone else is going to bother looking anyway then forget it [14:56] i have a few more reboot tests to do, biab [14:57] hallyn: added [14:58] Daviey: tbh my sense is that libcgroup has fundamental flaws, and for our deafults we would be better off with a simple upstart job that mounts the cgroups [14:58] maybe i should discuss on -devel... [14:58] hallyn, did you not author that ? [14:58] smoser: no! [15:00] you're talking about cgroup-bin, right? [15:01] yes [15:01] and libcgroup1 in general [15:01]