| === cronus__ is now known as cronus | ||
| === SJrX is now known as SJr | ||
| gdeeble | Curiosity, I know that having software raid causes GRUB issues, that you have to repoint GRUB to the proper location, but does having the bios on an older desktop cause the same issue? Like I have 2 250gb drives that I want to run as a RAID1 for my server OS partition so I have that redundant, but want to check before starting into that project seeing everything is working and this would | 03:04 |
|---|---|---|
| gdeeble | be an upgrade. | 03:04 |
| patdk-lap | heh? | 03:05 |
| patdk-lap | since when does raid cause grub issues? | 03:05 |
| gdeeble | I was reading that. Maybe I misunderstood. | 03:05 |
| TheLordOfTime | where did ya read that? | 03:05 |
| patdk-lap | if you setup raid, grub should install itself on all raid disks to boot | 03:06 |
| patdk-lap | so that if one is missing, it can boot from the other | 03:06 |
| SpamapS | yeah | 03:06 |
| patdk-lap | in order for that to work though, will depend on how your bios failsover from disk to disk to boot | 03:06 |
| gdeeble | I was looking at the zentyal setup guide, when I was googling. Apparently, I need to go back and re-read it. LOL. | 03:07 |
| gdeeble | I was going to run as RAID1 for boot/os, which should be a mirror image, meaning it should boot even if degraded due to 1 drive failure. Is this not right? | 03:07 |
| SpamapS | gdeeble: that is right | 03:08 |
| SpamapS | gdeeble: the point is, grub will just install itself onto both drives | 03:08 |
| SpamapS | so there's no issue with repointing | 03:08 |
| SpamapS | gdeeble: you're installing onto this software raid and building it during the installer, right? | 03:08 |
| gdeeble | What I want to do, is allow the bios to raid the 2 drives then install the OS. | 03:09 |
| patdk-lap | you want to use bios raid? | 03:10 |
| gdeeble | This is the first time I'm working with any raids with linux, as I know on my desktop running windows, I just set it up in the bios and mounted the drive and it showed as 1. | 03:10 |
| patdk-lap | then your stuck using dmraid | 03:10 |
| patdk-lap | I would recommend not using a raid option in the bios, and use mdadm | 03:10 |
| gdeeble | Wouldn't software raid use more processing power from the proc, than using bios? | 03:11 |
| gdeeble | or are they the same | 03:11 |
| patdk-lap | both are software raid | 03:13 |
| patdk-lap | the bios is not hardware raid | 03:13 |
| gdeeble | Interesting, didn't know that. | 03:13 |
| gdeeble | I always assumed(yes bad) that if handled by the bios, it would show as 1 drive to any other software. | 03:13 |
| patdk-lap | only if by bios you mean, add in raid card, and not the motherboard bios | 03:14 |
| patdk-lap | you have 3 different raids currently | 03:14 |
| patdk-lap | hardware raid, fakeraid, and softraid | 03:14 |
| patdk-lap | fakeraid is what bios raid is | 03:14 |
| gdeeble | fakeraid = bios | 03:14 |
| gdeeble | Ok. | 03:14 |
| patdk-lap | it's software raid + bios boot helper stuff | 03:14 |
| === james_ is now known as Guest90685 | ||
| gdeeble | It's not any better than just a standard software raid then, correct? | 03:15 |
| patdk-lap | nope | 03:15 |
| patdk-lap | well, it can be | 03:15 |
| patdk-lap | but only in the case of, a disk failure | 03:15 |
| patdk-lap | and then, it's still iffy if it's any better | 03:15 |
| patdk-lap | the worst thing that can happen for pure softraid is if a disk fails, but doesn't completely fail | 03:16 |
| patdk-lap | so the bios still attempts to boot it | 03:16 |
| patdk-lap | but I believe that is also going cause fakeraid the same issues | 03:16 |
| p7ank5te7 | Sorry, dropped connection. | 03:18 |
| p7ank5te7 | So either way it could turn out bad without a true raid card | 03:19 |
| === NomadJim_ is now known as NomadJim | ||
| paulorodrigues | hello | 04:50 |
| paulorodrigues | anyone can help me? | 04:50 |
| paulorodrigues | ls | 04:50 |
| jotterbot1234 | hey mate | 04:51 |
| jotterbot1234 | what's up? | 04:51 |
| paulorodrigues | I am having problems with my ubuntu server | 04:51 |
| paulorodrigues | after an reboot it doesnt boot anymore | 04:51 |
| paulorodrigues | I can only boot using a boot cd | 04:51 |
| paulorodrigues | I triend to install grub, but it dont fix it | 04:52 |
| jotterbot1234 | I would boot using a liveCD like you describe and repair your grub partition | 04:52 |
| jotterbot1234 | hmm | 04:52 |
| paulorodrigues | yes | 04:52 |
| paulorodrigues | how can i repair without losing data? | 04:52 |
| jotterbot1234 | Maybe someone else on here can contribute? Have you tried googling "grub restore"? DId you receive any error codes? any feedback at boot? | 04:53 |
| paulorodrigues | no, it simple ask for a boot device | 04:53 |
| paulorodrigues | its like i dont have operation system | 04:53 |
| paulorodrigues | but I can boot with a live CD and the server works well | 04:54 |
| jotterbot1234 | Your hard drive might be dead? Do you have a spare you can try and copy it over to low level. Using dd? and then try and boot from that? | 04:55 |
| paulorodrigues | but i can access my hard drive | 04:55 |
| paulorodrigues | with the boot cd | 04:55 |
| paulorodrigues | i have 2 disk in raid with the operation system | 04:57 |
| Laogeodritt | Has anyone used darkstat before? Know if it's possible to get stats on a per-port basis for all hosts? (If not, any suggestions for packages that can do that?) I'd like to track the traffic and data transfer for different services running on my server | 05:02 |
| === smb` is now known as smb | ||
| JHofmann | hi | 09:49 |
| JHofmann | apt-get error mkinitramfs... anyone experience? | 09:50 |
| zetheroo | when configuring 2 identical drives as RAID1 do I have to set one of them to be bootable - or is it enough to make the ext partition on the RAID1 bootable ? | 09:52 |
| JHofmann | server is installed on vmware vsphere 5 server.... "df -H" -> /boot is 11% used | 09:54 |
| JHofmann | here is screenshot : http://img109.imageshack.us/img109/6595/errored.jpg | 09:57 |
| JHofmann | what can be the reason of this error? anyone have an idea? | 09:58 |
| === Myrtti_ is now known as Myrtti | ||
| === baba is now known as Guest13870 | ||
| giaco | hello | 10:47 |
| giaco | I need help: struggling with tomcat6 on Quantal. I'm having difficulties in running tomcat6 over java6 instead of java7 | 10:48 |
| giaco | should I forward the question to a tomcat channel? | 10:49 |
| jamespage | giaco, are you using the tomcat6 packages from ubuntu? or an upstream tarball? | 10:55 |
| giaco | jamespage: tomcat6 package from ubuntu | 10:55 |
| jamespage | giaco, right | 10:55 |
| jamespage | giaco, what issue are you seeing? | 10:55 |
| giaco | but nevermind, I've just solved ( from JAVA_HOME=/usr/lib/jvm/java-6-openjdk to JAVA_HOME=/usr/lib/jvm/java-6-openjdk-amd64 ) | 10:55 |
| jamespage | giaco, great! | 10:56 |
| === aard__ is now known as aard_ | ||
| jamespage | the packaging init script should check in know locations for java | 10:56 |
| koolhead17 | jamespage: i found 1 missing line in /etc/swift/container-server.conf | 11:19 |
| koolhead17 | which comes from our cloud repo for folsom | 11:20 |
| koolhead17 | i had to simply run cat >> /etc/swift/container-server.conf << EOF [container-sync] EOF | 11:20 |
| koolhead17 | for the fix | 11:20 |
| koolhead17 | also i don`t understand why swift package by defualt does not allow logging :( | 11:22 |
| alex88 | hi guys, what are the advantages to use lvm? | 11:44 |
| jamespage | yolanda, any chance you could ack https://code.launchpad.net/~james-page/horizon/memcache-fixup/+merge/153109 please | 11:44 |
| yolanda | ok | 11:44 |
| yolanda | jamespage, you are fixing cache and email settings also? | 11:49 |
| jamespage | yolanda, no - I just use slightly different quilt settings to adam who did the last refersh | 11:50 |
| yolanda | the diff was confusing me | 11:50 |
| jamespage | yolanda, my ~/.quiltrc - http://paste.ubuntu.com/5610447/ | 11:58 |
| jamespage | yolanda, fyi oslo-config went 1.1.0 today - most projects are bumping pip-requires to >= 1.1.0 | 12:09 |
| jamespage | I've updated raring and backported to the PPA as well - should land ~1hr or so | 12:09 |
| jamespage | adam_g, ^^ fyi | 12:09 |
| yolanda | jamespage, do they pushed that changes upstream in some of the projects? | 12:21 |
| jamespage | yolanda, yes - glance and ceilometer have already bumped | 12:21 |
| yolanda | i can fix them | 12:22 |
| yolanda | i'm just going crazy with the ceilometer meterings so a change of topic is good | 12:22 |
| === mikal_ is now known as mikal | ||
| jamespage | yolanda, you really don't need todo anything | 12:47 |
| yolanda | i saw it, i was confused because i think we had patched it | 12:47 |
| jcastro | Daviey: do you know when we're supposed to hear about our talk submissions for ODS? | 12:53 |
| ivoks | jamespage: you can reproduce this one? https://bugs.launchpad.net/ubuntu/+source/maas/+bug/1069570 | 12:55 |
| uvirtbot | Launchpad bug 1069570 in isc-dhcp "1 MAC Address, two IPs - DNS is "out of sync" with DHCP leases databases, I think..." [Undecided,Confirmed] | 12:55 |
| jamespage | ivoks, yes | 12:55 |
| ivoks | within kvm? | 12:55 |
| jamespage | ivoks, yes - ipxe and the os present differently to the isc dhcp server so they get different IP's | 12:55 |
| ivoks | i wonder... | 12:55 |
| ivoks | dhcp server opens both ipv4 and ipv6 udp ports | 12:56 |
| ivoks | ipv6 dhcp doesn't use mac address for identification | 12:56 |
| ivoks | could it be that maas dhcpd offers ipv4 address over ipv6 udp (which is possible) | 12:57 |
| ivoks | and then, once os boots up, it requests ipv4, since it's defined like that within /e/n/interfaces | 12:57 |
| ivoks | if you can easily reproduce this, could you do me a favor and disable ipv6 on maas server and see if that happens again? | 12:57 |
| Daviey | jcastro: i do not | 12:58 |
| jamespage | ivoks, I don't think that is the case | 12:58 |
| jamespage | ivoks, but I can try | 12:58 |
| ivoks | jamespage: if it's something you could easily try, that would be awesome, since i'm low on resources right now to try it myself | 12:59 |
| jcastro | Daviey: any idea who I could ping? | 12:59 |
| Daviey | jcastro: ttx might know. | 13:00 |
| Daviey | ivoks / jamespage: ipxe is ipv4 only, no? | 13:00 |
| jamespage | ivoks, fwiw we have a patched version is isc-dhcp-server we use with vmaas | 13:00 |
| jamespage | which ignores the uid that ipxe sends | 13:00 |
| Daviey | well, makes the identifier purely MAC - rather than MAC + client-id | 13:01 |
| ivoks | jamespage: right, but ipxe should send uuid only when ipv6 pxe booting | 13:01 |
| ttx | jcastro: depends on the exact track, they started to tell people yesterday, I think | 13:02 |
| Daviey | ivoks: Are you sure our ipxe supports ipv6? | 13:02 |
| jamespage | ivoks, ah - right | 13:02 |
| ivoks | Daviey: yes | 13:02 |
| ivoks | well, pxe | 13:02 |
| ivoks | not sure about ipxe :) | 13:02 |
| Daviey | We did have this doubling up issue due to ipv4/ipv6 with UEC Avahi based registration. Ended up having to pach out ipv6 support in the registration code. | 13:03 |
| jcastro | ttx: that's a reasonably good enough answer! So I guess I should find out by EOW. | 13:04 |
| ivoks | if we can easily test it, it would at least tell us that ipv6 isn't the problem | 13:05 |
| ttx | jcastro: let me see what I can find -- what was your talk name ? | 13:05 |
| jcastro | ttx: the juju charm workshop | 13:05 |
| jcastro | ttx: we had another one too, something like "lessons learned from dogfooding openstack", but I can't remember the exact title. | 13:06 |
| koolhead17 | jcastro: salute | 13:07 |
| ttx | jcastro: see pm | 13:08 |
| === Pici` is now known as Pici | ||
| Daviey | smoser: did you SRU that isc-dhcp checksum patch that seemed to impact xen? If not, i think we should bite the bullet and carry this client-uid one aswell. | 13:19 |
| smoser | i did not sru the isc-dhcp, no. i did say i'd follow up on that. | 13:21 |
| smoser | and its not xen only. | 13:21 |
| smoser | it was kvm that was reported. | 13:21 |
| smoser | (virtuio driver) | 13:21 |
| Daviey | smoser: er, i thought qemu carried support already? | 13:22 |
| Daviey | smoser: http://git.qemu.org/?p=qemu.git;a=commit;h=1d41b0c .. so surely it's only on xen we are seeing this? | 13:23 |
| smoser | it was reported on kvm. | 13:23 |
| smb | Daviey, if that is about udp problem, libvirt installs a mangle rule for iptables | 13:24 |
| smoser | i think it happens only if you have the virtio kernel module loaded | 13:24 |
| smoser | which maybe (possibly) does in kernel what that qemu patch would do in user-space | 13:24 |
| smoser | ? | 13:24 |
| Daviey | smb: it does NOW, but didn't in earlier versions? | 13:24 |
| smb | smoser, Daviey The problem is with all pv network drivers | 13:24 |
| Daviey | smb: Some of this is cross-distro issues, with RHEL.. With ubuntu as a guest.. So i'm not sure what their libvirt was doing. | 13:25 |
| smb | Daviey, Not sure when it started, but its there in P at least | 13:25 |
| smoser | Daviey, i seem to recall that it only occurs if you have the virtio_net driver loaded in the host | 13:28 |
| smoser | or somethign liek that. | 13:28 |
| smb | All depends whether "sudo iptables -t mangle -S" shows | 13:28 |
| smb | -A POSTROUTING -o virbr0 -p udp -m udp --dport 68 -j CHECKSUM --checksum-fill | 13:28 |
| smb | So people that installed libvirt do not see it | 13:29 |
| Daviey | Ah! | 13:29 |
| Daviey | If you use the openstack xen support outside of libivrt, you would see this. | 13:30 |
| === wedgwood_away is now known as wedgwood | ||
| Daviey | (openstack supports direct xen and libivrt xen) | 13:30 |
| smb | Right | 13:31 |
| Daviey | smoser: virtio-net can be used by xen, no? | 13:31 |
| smb | Or people using kvm directly | 13:31 |
| smb | Daviey, no | 13:31 |
| smb | Daviey, They have their own pv driver | 13:31 |
| smb | netback or was it netfront...? one of the two :-P | 13:32 |
| koolhead17 | soren: hello | 13:32 |
| Daviey | smb: Oh, i thought you could now use qemu's virtio.. ok. | 13:32 |
| jamespage | adam_g, yolanda: switched the trunk builds for quantum to the the milestone-proposed branch as master is now open for havana | 13:34 |
| jamespage | Daviey, ttx: ^^ fyi | 13:34 |
| jamespage | ttx: I'll keep one eye on openstack-dev ML but can you ping me as projects switch over from master for grizzly rc1 | 13:35 |
| Daviey | jamespage: thanks | 13:38 |
| ttx | jamespage: I'll try to remember to ping you. You can also watch the appearance of a milestone-proposed branch | 13:51 |
| === matsubara_ is now known as matsubara | ||
| === highvolt1ge is now known as highvoltage | ||
| jamespage | ttx, ta - will do | 14:13 |
| soren | koolhead17: Ohai. | 14:15 |
| jamespage | Daviey, can I object to a later version dependency upgrade in cinder? https://bugs.launchpad.net/cinder/+bug/1150720 | 14:28 |
| uvirtbot | Launchpad bug 1150720 in cinder "There is now a dependency on paramiko v1.9.0" [Undecided,In progress] | 14:28 |
| Daviey | jamespage: why aren't we seeing this error? | 14:30 |
| Daviey | jamespage: Oh, it's only visible with storwize_svc ? | 14:31 |
| jamespage | Daviey, yep | 14:31 |
| jamespage | Daviey, its a pita because cinder uses stevedore for extension loading; and if pip-requires is not 100% correct then #ERROR | 14:32 |
| jamespage | Daviey, I can patch it out if need be and pick a fix for paramiko... | 14:32 |
| Daviey | jamespage: well, it looks like a 1 line fix | 14:32 |
| Daviey | jamespage: https://github.com/paramiko/paramiko/commit/f8239015ec427a2b5e62afa8370885894483a356 | 14:32 |
| jamespage | yes | 14:32 |
| Daviey | jamespage: Seems annoying to bump this late stage, doesn't it | 14:34 |
| Daviey | dosaboy: Hey, Your nova upload to precise... i think that the outcome of our discussion a few days ago was to reject. I have now rejected it, but can be salvaged easily if needed. | 14:39 |
| jamespage | Daviey, yes - I've raised a task on that bug for Ubuntu explaining the situation. | 14:40 |
| jamespage | Daviey, phew - paramiko has a test suite | 14:40 |
| jamespage | Daviey, guh - we would need to SRU that fix | 14:43 |
| jamespage | it does not qualify for the cloud-archive | 14:44 |
| Daviey | jamespage: yep, find out what exactly it does.. and see if you feel it meets the SRU criteria | 14:44 |
| dosaboy | Daviey: sorry got the right channel now ;) | 14:44 |
| Daviey | dosaboy: hola! | 14:44 |
| Daviey | dosaboy: Hey, Your nova upload to precise... i think that the outcome of our discussion a few days ago was to reject. I have now rejected it, but can be salvaged easily if needed. | 14:44 |
| dosaboy | howdi | 14:44 |
| dosaboy | ok so that patch has been applied through a different patch set? | 14:45 |
| dosaboy | or are we rejecting it outright? | 14:45 |
| dosaboy | seem to remember another patch set incluing it (possible from adam_g?) | 14:45 |
| Daviey | bah | 14:48 |
| streulma | hey, is there a ppa for updated php 5.3 and updated mysql-server ? | 14:48 |
| Daviey | streulma: not a supported one AFAIK. So if you find one, check the policy on security updates | 14:50 |
| streulma | Daviey: in Debian we have Dotdeb, I just run Debian on the server right now, for updates... | 14:51 |
| streulma | Daviey: but, love Ubuntu :) | 14:51 |
| Pici | PPAs are provided by the community. We can't ensure that you will be getting security updates in a timely manner. | 14:51 |
| Daviey | streulma: I'm not entirely sure of dotdeb's security policy. | 14:53 |
| streulma | Pici: what to have to do to get security updates for mysql and php on Ubuntu Server then ? | 14:54 |
| Pici | streulma: Nothing. If you are running a supported release then you will be getting security updates. | 14:55 |
| streulma | Pici: ok, then choose a VPS with Ubuntu Server 12.04 LTS then :) | 14:56 |
| jamespage | yolanda, w00t - I have a running ceilometer | 15:11 |
| jamespage | yolanda, spotting a couple of bugs but nothing major | 15:11 |
| yolanda | nice! | 15:11 |
| jamespage | (in your charms that is) | 15:11 |
| yolanda | is that working without stevedore problems? | 15:11 |
| jamespage | yolanda, it is now that I've got the up-to-date oslo-config package | 15:11 |
| Daviey | dosaboy: I *think* the situation was.. you upload needed re-basing as it was trumped by a security upload.. And adam_g confirmed that as it was now in stable/essex, a refreshed snapshot would include it. | 15:12 |
| Daviey | dosaboy: The issue now, is that nobody has created a new set of snapshots. | 15:12 |
| Daviey | (they need doing quickly really) | 15:12 |
| yolanda | jamespage, did you notice the "connection lost" problem, after doing the relation between nova and ceilometer? i always have to reboot nova machine to make that work | 15:12 |
| jamespage | yolanda, I'm not running in canonistack so don;t get that issue | 15:13 |
| yolanda | and the duplicate nodes? | 15:13 |
| jamespage | yolanda, I have seen that tho - but I think we already discussed this | 15:13 |
| yolanda | jamespage, but you told that with latest juju versions it should work, i use juju from lp source, and i have that problem | 15:14 |
| dosaboy | Daviey: gotcha, let me know if I can help with anything | 15:14 |
| jamespage | yolanda, I don't see duplicate nodes | 15:14 |
| yolanda | jamespage, how about if you reboot nova, or stop/start it? | 15:20 |
| fakhir | Hello. I have a server with a network interface and a virtual network interface. how can I chose to always use the vitrual network interface for outgoing connections? | 15:24 |
| === matsubara is now known as matsubara-lunch | ||
| SpamapS | fakhir: that is a routing issue | 15:51 |
| SpamapS | fakhir: your default route determines where most outgoing connections go | 15:52 |
| fakhir | SpamapS, ok how do I set the default route? | 15:52 |
| SpamapS | fakhir: in /etc/network/interfaces | 15:54 |
| SpamapS | fakhir: man 5 interfaces | 15:54 |
| hallyn | stgraber: did you pull alex's patch for lxc-archlinux? if not lemme push it real quick before you cut | 16:27 |
| hallyn | oh i see you did, nm :) | 16:27 |
| hallyn | are you pulling the lxc_map_ids patch right now? | 16:28 |
| hallyn | (that one should be in rc, it's a regression) | 16:28 |
| stgraber | hallyn: yeah, I think I pulled it too | 16:29 |
| stgraber | hallyn: hmm, I don't see it applied actually, let me fix that. I certainly meant to have it | 16:29 |
| stgraber | hallyn: pushed | 16:30 |
| hallyn | thx | 16:31 |
| rbasak | jamespage: around? I've got mongo past the alignment fault, but mongod hangs in atomic_intrinsics_gcc.h. I presume the ARM part of that needs fixing. Did you say that you'd done something with this already? | 16:33 |
| jamespage | rbasak, right! | 16:33 |
| jamespage | so that was the second part of the problem | 16:34 |
| jamespage | lemme push my patch for that bit | 16:34 |
| jamespage | rbasak, http://paste.ubuntu.com/5611149/ | 16:34 |
| jamespage | to start with I refactored stuff to use the gcc builtins for __sync_* | 16:35 |
| rbasak | jamespage: ah - I didn't realise that wasn't being applied already. | 16:35 |
| jamespage | but I'm not comfortable that I have it right | 16:35 |
| rbasak | Or is it? I'm confused. Are you saying that this is the fix, or that it doesn't work? | 16:35 |
| jamespage | rbasak, well the original patch from fedora-arm is - but it does not work | 16:36 |
| jamespage | that is my revised patch | 16:36 |
| rbasak | Ah, OK | 16:36 |
| djkorn | good day guys. | 16:39 |
| djkorn | need info for virtualbox-server | 16:40 |
| Fieldy | djkorn: can you be more specific? | 16:52 |
| === matsubara-lunch is now known as matsubara | ||
| koolhead17 | hi all | 17:41 |
| koolhead17 | seems like patch for this https://bugs.launchpad.net/swift/+bug/1024477 | 17:42 |
| uvirtbot | Launchpad bug 1024477 in swift "SAIO: logs filling up with "exception dumping recon cache"" [Undecided,Fix released] | 17:42 |
| koolhead17 | has still not landed in Ubuntu | 17:42 |
| koolhead17 | i am using 1.7.4-0ubuntu2~cloud0 & can still see the bug | 17:43 |
| === highvolt1ge is now known as highvoltage | ||
| Pest | i need help | 18:42 |
| Pest | i need that webcam saves screenshots something like /home/pictures/screen.jpg but not rewrite each time. like screen 1, screen 2 etc | 18:44 |
| Pest | advices, gentlemen? | 18:44 |
| sarnold | Pest: maybe you could convince logrotated to rotate your photos as if they were logs | 18:45 |
| stgraber | hallyn: http://paste.ubuntu.com/5611525/ does that look reasonable to you? | 18:46 |
| stgraber | hallyn: also, what exactly is going on with the cupset cgroup? why can't we put pid 1 in there? | 18:46 |
| Pest | i mean, should i put in sfwebcam.conf line like /home/pictures/screen %N .jpg ? | 18:46 |
| stgraber | hallyn: with that change, I now get the following result for /proc/1/cgroup in a "b" container running in a "a" container. http://paste.ubuntu.com/5611528/ | 18:47 |
| hallyn | stgraber: hold on... | 18:47 |
| hallyn | what are you trying to do? | 18:48 |
| hallyn | the hook as is was working for me, | 18:48 |
| stgraber | hallyn: I'm trying to fix the hook failing to setup the cpuset cgroup here, making the rest of it fail (as it's set -e) | 18:48 |
| hallyn | stgraber: i think you need a reboot | 18:48 |
| hallyn | then the first time lxc creates /sys/fs/cgroup/xxx/cpuset it will set clone_children to 1 | 18:49 |
| hallyn | and all should just work | 18:49 |
| hallyn | it did for me last night | 18:49 |
| stgraber | hmm, ok. Let me try that then | 18:49 |
| hallyn | stgraber: well, or just set clone_children to 1 by hand, | 18:49 |
| hallyn | and make sure to fill the mems and cpus fields in the lxc directory. after that all should go well | 18:50 |
| * hallyn has caused trouble yet again | 18:50 | |
| stgraber | hallyn: what do you mean by "fill the mems and cpus fields in the lxc directory"? | 18:51 |
| Pest | oh, "tnx" guys... so helpful | 18:52 |
| hallyn | stgraber: make sure that /sys/fs/cgroup/cpuset/lxc/cpuset.{cpus,mems} both are not empty | 18:52 |
| RoyK | Pest: cheese should do | 18:53 |
| Pest | RoyK: ? | 18:53 |
| RoyK | Pest: cheese is an application to use webcams or similar | 18:54 |
| stgraber | hallyn: ok, a reboot sorted it. The only cgroup that's wrong is the systemd one but I'm not sure I care ;) http://paste.ubuntu.com/5611746/ | 20:01 |
| hallyn | cool | 20:12 |
| kieppie | hi folks | 20:13 |
| kieppie | I have a headless 12.04 server doing my KVM/libvirt stuff, but I'd like to drasticly scale down the power-consuption on it - just shy of hibernation/standby when not doing any heavy lifting. how can I aggressively tune the power-consumption for on-demand use? | 20:14 |
| lisandrop05 | hello, I'm join files splitted using zip -s | 20:21 |
| lisandrop05 | but there is no way | 20:21 |
| lisandrop05 | which command I need to use to join the files again? | 20:21 |
| lostme | Hey, I work on Openvpn on Ubu Server, I tried to get a certificate signed by my own with "sudo openssl ca -config ca.cnf -in xx.req -out newcerts/xx.crt -notext", but I get only "usr/ssl/serial: no such file or directory" and "error while loading serial number". What's wrong? | 20:44 |
| sarnold | lostme: usr/ssl/serial ? | 20:47 |
| sarnold | lostme: can you copy-and-paste to a pastebin somewhere the entire command and output? | 20:48 |
| lostme | serial is a file, and I cannot copy and paste, sry | 20:50 |
| ScottK | lostme: Use the pastebinit package to put it in a pastebin. | 20:51 |
| stgraber | lxc-stop: failed to stop 'test-lucid': Operation not permitted | 20:54 |
| stgraber | FAIL | 20:54 |
| stgraber | runTest (__main__.SimpleLxcTests) ... lxc-stop: failed to stop 'test-lucid': Operation not permitted | 20:54 |
| stgraber | hallyn: ^ | 20:54 |
| hallyn | sigh. where is this? | 20:54 |
| * hallyn goes to replace all lxc testsuites with "if true: pass" | 20:55 | |
| stgraber | hallyn: running integration-test on current staging | 20:55 |
| hallyn | what is integration-test? | 20:56 |
| * hallyn still waiting for edubuntu to install :) | 20:57 | |
| hallyn | maybe i'll take the opportunity (if it boots :) to introduce the kids to it | 20:57 |
| stgraber | hallyn: bzr+ssh://bazaar.launchpad.net/~serge-hallyn/+junk/lxc-test/ | 20:58 |
| stgraber | root@castiana:~# lxc-stop -n test-lucid | 20:58 |
| stgraber | lxc-stop: failed to stop 'test-lucid': Operation not permitted | 20:58 |
| hallyn | <grimace> | 20:58 |
| stgraber | I can't remember ever seeing that kind of thing :) | 20:58 |
| hallyn | stgraber: so what happens if you lxc-stop it yourself? | 20:59 |
| stgraber | well, I can't ;) that's what I just tried | 20:59 |
| hallyn | stgraber: the lxc.stopsignal seems a likely culprit | 20:59 |
| stgraber | hallyn: yeah... I have that problem with any container on my machine, so it's a global problem, very likely introduced by the stopsignal stuff | 21:00 |
| hallyn | can you strace it and see what is getting the eperm? | 21:00 |
| * hallyn waiting on his test instance to finish installing ubuntu-dev-tools | 21:01 | |
| stgraber | hallyn: kill(32498, -1) = -1 EINVAL (Invalid argument) | 21:01 |
| stgraber | hallyn: so for some reason it uses -1 as the default value for the signal, instead of 9 (I guess our default is SIGKILL right?) | 21:02 |
| hallyn | stgraber: but the patch to lxc_stop_callback() defaults to using SIGKILL which was used before as explicitly | 21:04 |
| stgraber | hallyn: yeah, that doesn't make sense ... the -1 value should only be possible if "handler->conf->stopsignal" was set, which it isn't in my case... | 21:08 |
| stgraber | hallyn: if stopsignal isn't set, it should just be using the standard SIGKILL value as was the case in the past... | 21:08 |
| hallyn | right | 21:09 |
| stgraber | hallyn: shouldn't stopsignal be initialized to 0? | 21:11 |
| stgraber | (in conf.h) | 21:11 |
| stgraber | hmm, no, was missing context looking at the diff | 21:12 |
| hallyn | right, memset | 21:14 |
| === keithzg_ is now known as keithzg | ||
| hallyn | why does rootfs get an extra / prepended? | 21:20 |
| hallyn | stgraber: you've verified that without that patch it works? | 21:21 |
| hallyn | great, apparently i mistyped my interfaces file | 21:22 |
| stgraber | hallyn: I haven't tried reverting yet, but I can't think of anything else which would change the second parameter of kill() to -1 | 21:23 |
| stgraber | I'm building a version with some extra debug statements now | 21:23 |
| hallyn | oh wait, heh, i guess my instances are hanging from the cgroup-lite bug :) | 21:34 |
| stgraber | hallyn: so handler->conf->stopsignal is definitely -1 in my case, now to figure out why | 21:39 |
| stgraber | hallyn: and I also confirmed that config_stopsignal isn't called, so that part works too | 21:40 |
| === james_ is now known as Guest11140 | ||
| hallyn | buffer overrun? | 21:43 |
| hallyn | stgraber: on my end, i can't get a raring instance with eth0 as bridge_port to br0 to boot today. | 21:44 |
| stgraber | hallyn: so I confirmed that at the end of lxc_conf_init stopsignal is indeed 0, so if nothing touches it afterwards and it's -1 at the time of lxc_stop it looks like something is overwriting it by accident | 21:52 |
| hallyn | right, i'm looking over git wc -p for clues | 21:52 |
| stgraber | added a ton more debuging here, hopefully that should show me roughly where the value changes to -1 | 21:57 |
| mkander | I have a strange problem here. After I installed plex media server on the server it is requesting and receiving multicast stream from the IPTV from the provider. Even though Plex is turned off. Any idea how I can debug? I dont want the server to get flooded. | 21:59 |
| mkander | cant find out what is generating the request | 22:00 |
| sarnold | mkander: as I understand IGMP, the router will periodically poll the network to see if multicast groups are still needed, and unsubscribe from ones that aren't needed any longer | 22:02 |
| sarnold | mkander: see e.g., http://en.wikipedia.org/wiki/Internet_Group_Management_Protocol#IGMPv3_membership_query | 22:02 |
| sarnold | mkander: you might be able to just wait a little while.. | 22:02 |
| mkander | sarnold: Thanks for the tips, read the article | 22:06 |
| mkander | sarnold: I have waited for one week but something is still requesting with IGMP | 22:06 |
| sarnold | mkander: hahaha | 22:06 |
| sarnold | mkander: oh jeeeze :) that's ... wow. | 22:07 |
| sarnold | mkander: I thought perhaps it'd been ten minutes or something. | 22:07 |
| mkander | it was 5 :P | 22:07 |
| mkander | but still it is requesting the stream somehow | 22:08 |
| mkander | verified this by telneting into the router and running igmpproxy | 22:08 |
| mkander | after some minutes it gets a request from the server ip | 22:09 |
| stgraber | hallyn: so it doesn't appear to be anything during initialization... I'm now digging into the mainloop/command handler | 22:13 |
| hallyn | stgraber: i gotta run, sorry. i *woudl* say just push without that patch, but that's like a ticking timeobomb if we're overrunning memory | 22:16 |
| hallyn | i'd guess my monitor changes for cgroup are to blame | 22:16 |
| hallyn | will look later tonight wehn ig et back | 22:16 |
| * hallyn out | 22:16 | |
| stgraber | I'll dig some more. I'd rather not push rc1 with a potential memory corruption of the conf struct ;) | 22:16 |
| stgraber | hallyn: yay, finally found the first place where the value appears to be -1. It's in lxc_command_mainloop_add. Looking at what happens before that one is called | 22:18 |
| hubuntu | silly question: when using virtual manager I would like to share clipboard between my ubuntu desktop and my ubuntu server in the VM | 22:40 |
| hubuntu | is that possible? there were some tool for such things for vmware last time I used that stuff in 2007 | 22:40 |
| sarnold | hubuntu: I think the virt-manager just fires up a VNC client. I'm not sure if any vnc clients make it easy to share clipboard or the selection buffer, but perhaps one exists? (I don't know that the standard one _doesn't_ do that..) | 22:42 |
| hubuntu | using spice | 22:45 |
| hubuntu | i think it is the spice-vdagent package | 22:45 |
| hubuntu | but thanks a lot sarnold | 22:45 |
| hubuntu | :) | 22:45 |
| sarnold | hubuntu: woo :) good luck :) | 22:46 |
| hubuntu | aptitude in precise is always a bit weird and wants to delete a lot of stuff all the time... I end up using apt-get and I get what i want | 22:47 |
| hubuntu | is that normal? | 22:47 |
| ScottK | Using apt-get is the sane way to go. | 22:49 |
| === arosales1 is now known as arosales | ||
| === wedgwood is now known as wedgwood_away | ||
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!
