[00:48] smoser: fyi tested the same test case on a quantal image, didn't affect that c-i version [00:49] right. [00:50] maas is the worst case for us here. [00:50] we can block just about everything else out by not releasing an updated image. [00:50] utlemming, ^ [00:50] please do not release a cloud-image until bug 1100491 has worked its way through. [00:50] Launchpad bug 1100491 in cloud-init "[SRU] cloud-init 0.6.3-0ubuntu1.3 failing to process juju-generated userdata" [Critical,In progress] https://launchpad.net/bugs/1100491 [00:56] roaksoax: comin atya https://code.launchpad.net/~gandelman-a/charms/precise/hacluster/avoid-osras/+merge/143626 === cpg|away is now known as cpg [01:29] hi all.. [01:30] I'm trying to install nvidia cuda on an ubuntu server 12.10 installation [01:30] for some reason i am unable to blacklist the nouveau driver [01:30] it seems to always be loaded [01:31] azbyin, i think you have to update-initramfs for the blacklist to make it into the initrd === Err404NotFound is now known as Error404NotFound === th0mz_ is now known as th0mz [08:29] Hello... is there a known howto page available for drbd+ocfs2+pacemaker installation in 12.04 ? [08:53] hi everyone [09:05] hi, i want to know is motherboard compatible with ubuntu server 12.04? Motherboard s1155 Intel H77 ATX DH77KC Intel BLKDH77KC [09:06] cpu and ssd does not matter, correct? [09:49] is Intel H77 ATX DH77KC Intel BLKDH77KC motherboard compatible with ubuntu server 12.04? any driver problems or other negative response about it? [09:51] mocha: check the hardware database [09:53] http://www.ubuntu.com/certification/catalog/ [09:54] mocha: and https://friendly.ubuntu.com/ [09:59] lifeless: thank, and if i dont find anything... then what? is only way "plug and pray" or is something else i can do before buy hardware? [09:59] you could ask the vendor [13:03] jamespage: ping can you have a look at the nova ftbfs for me im totally stumped (the testsuite failures) [13:04] zul: in the lab? [13:05] jamespage: yeah if you back out the fix-ubuntu-tests.patch you can see what im talking about [13:06] zul, just disable that patch right? [13:06] jamespage: yeah i rather not have to skip tests i would like to see why they are failing (was thinking about this last night) [13:07] zul, +1 [13:08] zul, on another matter who's uploading updated dependencies for the grizzly cloud archive? [13:08] http://reqorts.qa.ubuntu.com/reports/ubuntu-server/cloud-archive/grizzly_versions.html [13:09] looks sparse to me [13:09] just been debugging the build failure for alembic in the grizzly-trunk-testing PPA - needs new sqlalchemy [13:09] lovely === smw_ is now known as smw [13:12] jamespage: ill take a look at the cinder failure as well today [13:13] zul, actually leave that with me - I can fix the current build failure for precise easily [13:13] zul, are you referring to the skipped tests? [13:13] jamespage: ok yeah [13:14] also i have 2 mp for novaclient and quantumclient i think === cpg is now known as cpg|away [13:34] zul, just review - both look OK but can you fixup the trailing , on the list of BD on one of them [13:40] jamespage: sure === Err404NotFound is now known as Error404NotFound [14:14] jamespage: does openvswitch (1.9.0~git20130115.ca71f5b-0ubuntu1) works for you? I mean the dkms kernel module, here the module can't be loaded with kernel 3.8.0-0-generic in raring [14:15] freeflyi1g, it was working for me yesterday [14:15] I can check again... [14:16] freeflyi1g, does it build OK? [14:16] i.e. the dkms modules [14:16] jamespage: it was built successfully [14:16] ERROR: could not insert 'brcompat': Unknown symbol in module, or unknown parameter (see dmesg) [14:16] freeflyi1g, ah! I know what that is [14:17] freeflyi1g, that probably means you have the stock openvswitch module which ships with the kernel loaded [14:17] jamespage: lemme check [14:17] OR the bridge module is already loaded [14:19] jamespage: yep, works by unloading stock openvswitch module [14:19] jamespage: thanks [14:22] freeflyi1g, np - I had a head scratching moment about that yesterday [14:22] freeflyi1g, you know the brcompat module is deprecated now and will be removed from openvswitch (probably this year) === slank_away is now known as slank [14:53] jamespage: ping https://code.launchpad.net/~zulcss/openstack-ubuntu-testing/glance-grr-mad/+merge/143702 [14:55] zul, have you pinged upstream about the distruption that is going to cause === Ursinha_ is now known as Ursinha [14:55] jamespage: sent the info ttx but he hasnt responded yet [15:00] zul: hmm? [15:00] oh. pastebin. looking [15:00] yeah [15:02] zul: ok, continuing discussion in #openstack-infra. no idea where that comes from [15:03] I have a little box running LTS that sits and does its thing without my thinking about it. While this is great it does cause one problem for me, I forget to check for security updates. This is especially problematic because the box faces the internet. What automatic or semiautomatic solutions are there for such a situation? [15:06] voxadam, https://help.ubuntu.com/community/AutomaticSecurityUpdates === gondoi_ is now known as gondoi === duckydan_ is now known as duckydan === schmidtm_ is now known as schmidtm [16:10] yolanda: can you retarget the branch for lp:~openstack-ubuntu-testing/ceilometer/grizzly please, otherwise looks good [16:16] zul, done it === Ramjett_ is now known as Ramjett [16:46] smoser, getting this error from cloud-init on precise - http://paste.ubuntu.com/1541976/ [16:46] any ideas? [16:47] https://bugs.launchpad.net/ubuntu/+source/cloud-init/+bug/1100491 [16:47] Launchpad bug 1100491 in cloud-init "[SRU] apt_sources broken in 0.6.3-0ubuntu1.3 [regression]" [Critical,Fix committed] [16:50] smoser, ta [16:51] jamespage, just begging in -devel now to get that into -updates sooner. [16:51] smoser, yes please - its breaking all of the automated deployment testing on precise [16:51] juju never makes it onto the server.... [17:10] does anyone know if the debian experimental 2.4 packages are usable on Ubuntu (12.04)? [17:10] does anyone know if the debian experimental Apache 2.4 packages are usable on Ubuntu (12.04)? [17:18] how do I reinstall grub from a chroot? [17:18] I booted my another linux up and mounted the drive, I just need to chroot and install grub again === yofel_ is now known as yofel [17:33] why is the default niceness for new processes 20 on my ubuntu server? isn't it usually 0? [17:36] hm.. nice is usualy 5 [17:36] seems it's 10 on precise [17:37] RoyK, yeah any ideas? limits.conf is empty too [17:37] happyface: that's odd; do you have any /etc/security/limits.conf settings that would fiddle with your nice values? Or somewhere else in login path? [17:37] happyface: which ubuntu version? [17:37] happyface: 0 is normal priority, btw [17:37] happyface: are you positive it's nice value and not priority that's 20? all my nice 0 tasks are running with priority 19 on my 12.04 LTS laptop... [17:38] nice 0 == '' [17:38] sarnold, ah I think you're right [17:38] priority is defined by the kernel dynamically [17:39] happyface: ps -A -o pid,nice,pri [17:39] or just run top :P [17:39] yup, silly mistake :p [17:39] ah, so they are both in top :) hehe [17:39] I only got onto this because my other developer was freaking out [17:40] thanks :P [17:47] how can I tell if it is a LUN or physical disk? [17:47] lspci shows [17:47] 07:00.0 RAID bus controller: Adaptec AAC-RAID (rev 09) [17:47] mdstat show [17:47] md0 : active raid1 sdb1[1] sda1[0] [17:48] how can I tell if sda is either JBOD disk or a LUN provisioned by the raid controller? [17:49] axisys: use a tool for the RAID controller - it's usually exposed as a SCSI device, and linux normally can't see what's underneath [17:49] most RAID controllers have tools for linux showing the underlying structure [17:52] I wonder if this is a pass through one.. would be easier to replace with bigger size disk then [17:53] axisys: what controller? [17:53] 12:47:39 < axisys> 07:00.0 RAID bus controller: Adaptec AAC-RAID (rev 09) [17:54] axisys: http://hwraid.le-vert.net/wiki/Adaptec [17:54] RoyK: thanks [17:54] iirc that raid controller is a 'true' raid controller, so if configured with RAID, it should show up as a single device in linux [17:55] but then, you can always configure it in JBOD mode, dunno [17:55] I usually only create single disk raid0 luns [17:55] easier to manage at OS that way [17:56] yeah, I prefer software RAID over hw raid too [17:56] configuring each disk as raid0 is normally the same as JBOD [17:57] RoyK: yep.. I would prefer rip the controller out and talk to the disk directly if I could.. lol [17:57] axisys: perhaps better if you have SSDs, since most (or all?) RAID controllers don't support TRIM/UNMAP [17:57] hi all [17:58] they have 300G SSDs relatively cheap? [17:58] RoyK: ^ [18:00] I already have the drivers ofcourse.. cuz I am using the disks.. but trying to find out if that url has a link for the arcconf binary.. using ubuntu 10.04 64bit [18:01] axisys: what sort of machine is this? [18:01] aac-raid on motherboard? [18:01] http://hwraid.le-vert.net/wiki/DebianPackages got it! [18:01] [ 19.006845] Adaptec aacraid driver 1.1-5[2461]-ms [18:01] [ 19.006891] aacraid 0000:07:00.0: PCI INT A -> GSI 17 (level, low) -> IRQ 17 [18:01] [ 19.006897] aacraid 0000:07:00.0: setting latency timer to 64 [18:02] [ 19.460000] IRQ 17/aacraid: IRQF_DISABLED is not guaranteed on shared IRQs [18:02] !pastebin [18:02] For posting multi-line texts into the channel, please use http://paste.ubuntu.com | To post !screenshots use http://imagebin.org/?page=add | !pastebinit to paste directly from command line | Make sure you give us the URL for your paste - see also the channel topic. [18:02] [ 22.599531] scsi2 : aacraid [18:02] RoyK: my bad [18:07] http://pastebin.com/c0figCNK [18:08] looks like JBOD to me [18:09] please explain (learning) [18:09] I guess you'll have to reconfigure new devices in the controller's BIOS if replacing drives, though, but that may be different with other BIOSes than the ones I've tried [18:09] I am thrown off by logical device with raid level simple_volume [18:09] JBOD == just a bunch of disks [18:10] aka no RAID config on the RAID controller [18:10] i thought each disk is configurd as raid0 lun [18:10] no? [18:10] yes, which means no redunancy on the controller [18:11] RoyK: correct.. [18:11] i do the raid1 using the software raid [18:11] yes, so did also mdstat show [18:12] so if I want to replace with larger disk.. I will have to take it offline in md first and then offline in the raid controller while system is live.. [18:12] correct? [18:13] in your raid-1, yes [18:13] do you plan something raid-[56] for the other drives? === negronjl` is now known as negronjl [18:13] no .. i only use raid1 and raid10 in OS [18:14] disks are cheap [18:15] just thinking if you were using raid-5 or raid-6 (which may be slower, but in the case if raid-6, safer), it would be easier to use a disk as a spare, so you could replace that and mdadm --replace to that before removing a device [18:17] gotcha.. [18:18] raid-10 will allow for a single drive failure per mirror, raid-6 will allow for any two drives failing without data loss [18:19] so unless you're placing a high-traffic database or something like that needing high iops, better use raid-6 (IMO) [18:24] so if I want to replace it bigger size disk I need to take the system down? [18:24] that depends on the controller firmware [18:24] I was hoping arcconf can offline it and allow to put a new disks and online it back [18:25] last I worked with that controller, it needed the new device to be defined in its BIOS === wedgie_ is now known as wedgie [18:26] iirc it's designed to handle hotplug, but only if it's controlling its own RAID [18:26] http://pastebin.com/MgBGn2sF [18:26] firmware version [18:26] no idea [18:26] better try it [18:26] i am looking for how to offline a disk [18:26] axisys: if you have another drive that will fit in there, just try it [18:27] axisys: I guess this system isn't in production yet? [18:27] RoyK: it is in production [18:27] DELETE | deletes one or more logical devices [18:27] that might do it [18:27] once i remove it from md raid1 [18:28] just make sure to run a scrub before you remove a disk [18:28] something like echo check > /sys/block/md0/md/sync_action [18:29] I thought this should be enough [18:29] mdadm /dev/md0 --fail /dev/sdb1 --remove /dev/sdb1 [18:30] that works well [18:30] but then, if there are errors on the other side of the mirror, you're in trouble [18:30] so better run a check first [18:32] what is the application for this server? [18:35] jamespage: zul http://people.canonical.com/~agandelman/g2_deps/ are these okay to upload to grizzly-staging CA pocket? they'll get rid of all/most the red in staging @ http://status.qa.ubuntu.com/reports/ubuntu-server/cloud-archive/grizzly_versions.html [18:36] damn it i havent looked at it [18:36] gimme a couple of minutes [18:37] shouldnt ceph be precise-grizzly? [18:37] adam_g: ^^^ [18:37] axisys: ? [18:38] adam_g: alembic looks ok [18:39] adam_g: why is leveldb needed? [18:39] zul: actually, yeah, exclude the ceph stuff [18:40] adam_g: k [18:45] adam_g: they look good although you will need a newer testrepository (ubuntu6 runs the testsuite during the build) [18:45] and we already talked about ceph [18:46] RoyK: oops.. sorry went to pick up something for lunch.. [18:46] adam_g: i think we might need the openvswitch snapshot from raring in there as well [18:46] RoyK: it is for splunk [18:46] so I guess no big needs for high iops :P [18:47] RoyK: yep [18:49] then my suggestion is, get a couple of 1-2TB drives and create a raid-5 on those, copy the data over to them (or perhaps you have another server somewhere), and start from scratch with two drives in raid-5, add more drives as needed, extend to raid-6 if you're paranoid [18:49] looks like only disks (raid0 luns) under raid10 will be replaced with bigger disk [18:49] RoyK: ^ [18:49] you can do all that dynamically [18:49] (mdadm raid, I mean) [18:49] smoser: hey, are you aware of any bugs in cloud-init's handling of the legacy 'user:' cloud-config option? [18:49] anyone had success with user --groups staff in kickstarting ubuntu Precise? whenever I add --groups staff to the user line in my kickstart file the install stops at the user creation bit, as if the whole line is made invalid by this. [18:49] smoser: having some trouble with it in 12.10 images [18:50] axisys: I'm quite sure that's not right - I think replacing drives with bigger ones in a raid-[56] should work well [18:50] SpamapS, it would not surprise me if there were a bug, but i thought we'd got it working right. [18:50] SpamapS, example ? [18:51] raid10 gives best performance [18:51] axisys: yes, it does, but do you need that performance? [18:51] I guess I will trun off splunk and then run check and offline and fail on each disk.. at a time [18:51] axisys: if you don't, better spend time on safety [18:52] axisys: install sysstat and enable it and you'll see over time what the i/o load is [18:52] probably very low [18:52] smoser: trying to use heat w/ stock 12.10 images [18:52] smoser: http://paste.ubuntu.com/1542277/ [18:53] smoser: ec2-user is created, but never gets ssh keys [18:53] we have all splunk servers on raid10.. wont have enough time to redo all [18:54] smoser: root does actually end up getting the specific ssh key though [18:54] adam_g, yes - leveldb, ceph (in that order - otherwise it won't pickup the new version) [18:54] thats the bobtail release [18:54] and python-hp3parclient as well [18:54] for cinder [18:57] jamespage: okay, adding hp3parclient. what are we doing with ceph? 0.56.1 into precise-grizzly? [18:58] smoser: also in root's authorized_keys file, the user is not 'ec2-user', it is 'NONE' [18:59] axisys: well, raid-10 should do as well, but I don't know it too well in terms of replacing drives [18:59] smoser: hrm, you know what? This was with 0.7.0-ubuntu2 ... let me try again w/ the latest cloud image. [19:00] zul: [19:00] http://people.canonical.com/~agandelman/g2_deps/testrepository/ [19:01] +1 === io is now known as IdleOne === matsubara is now known as matsubara-afk [19:07] zul: this guy, too http://people.canonical.com/~agandelman/g2_deps/python-hp3parclient/ [19:07] +1 [19:10] axisys: just make sure you scrub the pool before replacing anything [19:11] SpamapS, it is a bug. [19:11] axisys: try at off hours first, after a scrub, to remove a drive and reinsert another, perhaps the raid tool can do a rescan and recreate a raid-0 on that drive, after that you should be able to mdadm --add that drive [19:15] adam_g, yes please re ceph [19:15] but please make sure the leveldb package is updated first; there are some improvements worth having :-) [19:16] 0.56.1 is the next lts release [19:16] jamespage: ok [19:20] echo check > /sys/block/md10/md/sync_action should do for md10 or something else? [19:20] RoyK: ^ [19:20] smoser: good to know. Is there a good workaround? [19:21] you can just declare it in the new form [19:21] smoser: heat wants to work with 10.04 [19:21] that would be ignored then. [19:21] (the new form would be ignored on 10.04). [19:21] so just declare both. [19:21] i think [19:21] axisys: it works for all raid levels [19:21] smoser: I'll try that [19:21] but please opne a bug. [19:26] smoser: http://paste.ubuntu.com/1542348/ [19:26] smoser: so, like that? [19:27] no [19:28] ditc line 6, SpamapS [19:28] smoser: the example sucks then :) [19:28] that refers to the OS's "default" user [19:28] ie, you'd get 'ubuntu' created. [19:28] ah [19:28] the example is right, it just didnt' fit what you wanted [19:29] smoser: perhaps.. comments? :) [19:29] * SpamapS admits, he did not read the comments [19:29] or the docs [19:29] so.. yeah.. perhaps SpamapS sucks [19:29] i think you might want to add 'default: true' to your user's definition === alienth_ is now known as alienth [19:36] SpamapS, please open a bug. [19:37] smoser: \o/ [19:37] smoser: new style works [19:39] actualy no [19:39] I mis-read the ci-info [19:40] still only the root user gets the key [19:45] smoser: ah, default: true so that it gets the ssh key.. [19:46] well, adding 'default' turns it into the "default" user. [19:46] which the ssh-import uses [19:46] can you show me what you have now ? [19:47] smoser: http://paste.ubuntu.com/1542389/ [19:48] smoser: trying with default: true [19:51] SpamapS, just fyi, apparently (through documentation i just read) you can snpecify ssh authorized keys for each user. [19:51] and even per-user ssh-import-ids [19:51] but it still "should work" what you want [19:51] smoser: in this case I just want the instance keypair for the user. :) [19:52] anyone else having issues with ec2 today? [19:52] euca-describe-instances is blocking [19:52] smoser: still no ssh keys for ec2-user with 'default: true' [19:59] smoser: I am thinking this might be something else in the stack causing issues [19:59] smoser: booting a quantal instance with just that userdata produces a key just fine [20:00] smoser: as in, just 'user: ec2-user' [20:01] * SpamapS decides to consider it over lunch [20:03] well, SpamapS please open abug if its not working right. [20:03] but i thought i had reproduced === frojnd_ is now known as frojnd [20:25] I have a file that's been modified after installation. when dpkg prompts for how you want to handle it [20:25] how can I force apt to accept the maintainer's version? I tried -y & --force-yes === blkperl_ is now known as blkperl [20:46] jamespage: thoughts on retargetting the nova live migration charm work under review as merges into the respective branches of ~openstack-charmers, to condense the # of in-flight changesets? === Ursinha is now known as Ursinha-afk [20:58] adam_g, sounds like a good idea - I've been having trouble keeping track [21:01] jamespage: yah, same here. ill see about resolving conflicts and proposing some merges to the ~openstack-charmers branches for nova-compute, nova-c-c, keystone, glance. also gonna stick a rabbitmq-server charm with hacluster support and propose changes to the amqp hooks of the other charms [21:07] +! [21:07] +1even [21:13] ciao [21:13] !list [21:13] capauciell: No warez here! This is not a file sharing channel (or network); read the channel topic. If you're looking for information about me, type « /msg ubottu !bot ». If you're looking for a channel, see « /msg ubottu !alis ». === Ursinha-afk is now known as Ursinha [22:03] say (looking at an untriaged bug) is tomcat7 explicitly not meant to run with oracle java? [23:02] is it possible to restrict only certain SSH logins to have key-auth only? [23:02] thereby prevent password auth for a user? [23:03] most admins prefer disabling password auth for everyone [23:04] true, however i can't do that globally because of one user. [23:04] who is already on ther eceiving end of hell for using insecure passcodes [23:08] roaksoax: ping [23:09] TLoT: the only thing I can think of -- and it's a bit ugly, but I'm saying it in the hopes that someone else will have a good idea :) -- is to run a second sshd, with a different set of PAM modules (pam_succeed_if?), on a different port for your special user. [23:09] eww that does sound ugly :p [23:09] i'd rather just kick the one user who needs password-only auth to /dev/null :P === ariel__ is now known as Guest88403 [23:12] anybody here? [23:12] TLoT: oh! pam_succeed_if could probably do it just for the one user [23:13] TLoT: .. without a second daemon. [23:14] sarnold, still ugly [23:14] Guest88403, sup [23:15] TLoT: sure, but much less ugly. [23:16] can you show me how to put my ip on the map [23:24] adam_g: pong [23:26] how do i make a directory writable by my user [23:27] roaksoax: are you doing work on DRBD /w the hacluster stuff? [23:27] adam_g yes [23:28] roaksoax: where at? and hows it going? thinking maybe need that for this rabbit stuff [23:28] adam_g: most of the stuff is here: https://code.launchpad.net/~andreserl/charms/quantal/mysql/hacluster-support [23:28] all of the stuff for now [23:29] so it is up to mysql to set its own DRBD stuff and then passes resources,constraints, etc to hacluster to be configured [23:36] roaksoax: hmm. wondering if theres anyway to make the drbd stuff more consumable from other charms. [23:39] adam_g: we can have some of it in an external lib or something. I wanted to make it in the hacluster charm at first but in reality it didn't fit there === cpg|away is now known as cpg