uvirtbot | New bug: #1057320 in glance (main) "Glance api and swiftclient not compatiable versions" [Undecided,Fix released] https://launchpad.net/bugs/1057320 | 00:01 |
---|---|---|
pmatulis | anyone here get sssd working with sudo configured in ldap? | 00:02 |
=== n0ts_off is now known as n0ts | ||
uvirtbot | New bug: #1061961 in horizon (main) "Upgrading openstack-dashboard does not upgrade dependencies." [Undecided,In progress] https://launchpad.net/bugs/1061961 | 00:21 |
ipl31 | Hello, so I had a qlogic 10 Gig card that worked oneiric but now seems to not be happy because of a firwmare version in precise. Did something change in the linux-firmware package? | 00:27 |
sarnold | ipl31: were there any enlightening notes in the /usr/share/doc/<foo>/changelog* files? | 00:31 |
ipl31 | sarnold: unfortunately no | 00:33 |
sarnold | :( | 00:33 |
sarnold | ipl31: if you steal the firmware from the old version, does it work? | 00:34 |
ipl31 | good question I should try that | 00:34 |
ipl31 | although box is remote and I have not network access only remote KVM, but I will see if there is way I can get it on there | 00:35 |
sarnold | oof | 00:35 |
ipl31 | yeah :) | 00:35 |
sarnold | would it be easier to get a local card? | 00:35 |
ipl31 | yeah it might be, its 10G card so I can't buy one locally would need to have it shipped | 00:35 |
ipl31 | but I might end up doing that | 00:36 |
sarnold | .. it doesn't sound cheap :) but still, that also sounds like a system that's best to leave alone as much as you can. :) | 00:36 |
ipl31 | actually system is not in service so I can do whatever I want | 00:36 |
sarnold | woo : | 00:36 |
sarnold | :) | 00:36 |
ipl31 | testing the upgrade to precise | 00:36 |
=== matsubara is now known as matsubara-afk | ||
uvirtbot | New bug: #1061977 in cloud-init "Machine fails to commission when console=ttyS0 is present on kernel opts" [Undecided,New] https://launchpad.net/bugs/1061977 | 01:21 |
stgraber | SpamapS: I'm testing on an ivy bridge laptop (i7) with a lot of RAM and a very fast SSD... | 01:49 |
=== jtv1 is now known as jtv | ||
SpamapS | stgraber: ok, so not that dissimilar from my mac book air (i5, 4GB RAM, average SSD) | 02:06 |
Guest2295 | Hello all!!! | 02:23 |
Guest2295 | I am coming I think to the end of a very drawn out problem with reference to my primary system running ubuntu. | 02:23 |
uvirtbot | New bug: #1061996 in samba (main) "Can Not Create Samba Share" [Undecided,New] https://launchpad.net/bugs/1061996 | 02:31 |
Tohuw | I need to install Magick-config, but it seems the only packages that provide it from the Ubuntu repos require X11. Can someone enlighten me as to why this might be, and if there's a suitable alternative if my only goal is to provide Magick-config for gem install rmagick? | 02:49 |
sarnold | Tohuw: drat, I wondered if graphicsmagick-libmagick-dev-compat might be easier .. but no luck, it also eventually depends upon libx11-dev | 02:52 |
sarnold | Tohuw: I believe I've read some mechanism to fake dpkg into thinking that some specific packages are already installed; I _hope_ that graphicsmagick or imagemagick would load libraries only as they need them.. | 02:53 |
Tohuw | sarnold: I could force the install of the package, but I'd rather not... based on my reading of Magick-config, it seems odd to require the X11 stack for it... | 02:57 |
sarnold | Tohuw: it'd be a package-level dependency, the individual tool may not need it at all. | 02:57 |
=== Psi-Jack_ is now known as Psi-Jack | ||
Tohuw | sarnold: I found a workaround via installing libgraphicsmagick1-dev, which installs an acceptably minimum number of X11 dev packages. | 03:05 |
sarnold | Tohuw: oh! hooray for just the -dev packages. that'll save a bit. | 03:05 |
Tohuw | Yes, considerably. x11-common, libx11-dev, x11proto-xext-dev. I can live with that. | 03:06 |
Tohuw | For the record, the correct package ended up being 'graphicsmagick-libmagick-dev-compat' to resolve rmagick not installing on Ubuntu via gem install. This is the most minimal provider I could find. | 03:17 |
Tohuw | I didn't catch that libgraphicsmagick1-dev provides GraphicsMagick-config, which is not quite the same. I also needed libmagickwand-dev. | 03:23 |
=== n0ts is now known as n0ts_off | ||
=== n0ts_off is now known as n0ts | ||
=== cyphermox_ is now known as cyphermox | ||
=== gary_poster is now known as gary_poster|away | ||
uvirtbot | New bug: #1006149 in libvirt (main) "PowerPC needs access to /proc/device-tree/ in apparmor perms" [Medium,Expired] https://launchpad.net/bugs/1006149 | 04:27 |
=== n0ts is now known as n0ts_off | ||
=== n0ts_off is now known as n0ts | ||
=== n0ts is now known as n0ts_off | ||
=== dendrobates is now known as dendro-afk | ||
basil60 | I'd like some advice on possible malware on ubuntu11.04 server please? | 05:55 |
basil60 | I've received a number of complaints from my ISP that unsolicited emails have originated from IP address. I've scanned for rootkits , malware a,d almost every other conceivable problem on my PC. As soon as I turned my server back on, the complaints returned. I'm sour it's some system change that I made that has permitted this problem... I'd just like some advice on shutting it down. | 06:06 |
=== kInOzAwA1 is now known as kInOzAwA | ||
basil60 | Thanks guys...I might try again later | 06:18 |
=== lifeless_ is now known as lifeless | ||
=== smb` is now known as smb | ||
=== n0ts_off is now known as n0ts | ||
=== dendro-afk is now known as dendrobates | ||
jellybean | Good morning all | 08:04 |
=== n0ts is now known as n0ts_off | ||
jellybean | Is there a recommended procedure for installing Ubuntu Server to a compact flash drive? | 08:04 |
=== dendrobates is now known as dendro-afk | ||
=== dendro-afk is now known as dendrobates | ||
=== n0ts_off is now known as n0ts | ||
DarkStar1 | hi all.. Is there anyway I can increase the size of a directory? specifically one under the /var tree | 09:52 |
=== n0ts is now known as n0ts_off | ||
chmac | DarkStar1: Directories don't have sizes exactly, unless you put more stuff into the directory, in which case it'll get "bigger" | 09:59 |
chmac | DarkStar1: Or are you running out of space? | 09:59 |
=== mcclurmc is now known as mcclurmc_away | ||
DarkStar1 | chmac: I ran out of space. I was importing old mail archives into a the directory path /var/vmail | 10:00 |
chmac | DarkStar1: Do you have space elsewhere on the disk? | 10:00 |
DarkStar1 | chmac: yeah. let me pastethe output of df -hl | 10:01 |
DarkStar1 | chmac: http://fpaste.org/05Cn/ so I'm guessing I need to increase the amount of space allocated to /dev | 10:03 |
chmac | DarkStar1: OVH server right? | 10:03 |
DarkStar1 | yeah | 10:03 |
DarkStar1 | chmac: houw'd you know :) | 10:03 |
chmac | They have this weirdness with rootfs and /dev/root, very strange. /dev/root doesn't actually exist. | 10:04 |
DarkStar1 | hhmm. | 10:04 |
DarkStar1 | mayhaps I can remount /var on dev2? | 10:05 |
DarkStar1 | on /dev | 10:05 |
DarkStar1 | ? | 10:05 |
chmac | DarkStar1: There is no /dev | 10:11 |
chmac | It's a virtual filesystem, your only real file system is /dev/md2 | 10:11 |
chmac | You could try symlinking your /var/vmail directory to /home/vmail or something like that. | 10:12 |
chmac | ovh are pretty stupid in the way they lay out your disk, I always repartition them. | 10:12 |
sri | hi | 10:14 |
=== sri is now known as Guest18447 | ||
Guest18447 | hi | 10:14 |
DarkStar1 | chmac: hhm……... | 10:15 |
DarkStar1 | good idea. Will have to trythat later. thanks | 10:15 |
chmac | DarkStar1: Have you been using the machine for a while? | 10:16 |
chmac | DarkStar1: You could also repartition it, that's one advantage of software raid, you can repartition on the fly. | 10:16 |
chmac | DarkStar1: I wouldn't recommend it with a production system, but ovh provide 100G of ftp backup, so you can push a backup to be doubly safe, take one disk out of the raid array, repartition, bring it back into the array, resync, wait, repeat. | 10:16 |
DarkStar1 | chmac: for over a week now. I've been setting it up for a client and there a few sites hosted on it so | 10:17 |
DarkStar1 | far | 10:17 |
chmac | DarkStar1: In that case you definitely want to repartition, all your /var/www/ and /var/logs/ will be on that 10G root partition, likewise /var/lib/mysql | 10:17 |
DarkStar1 | Oh.. I'll have to book sometime off to do that then | 10:17 |
chmac | You can do the whole thing with the machine still online, because you have 2 disks. | 10:17 |
DarkStar1 | shit!! | 10:18 |
DarkStar1 | all that time setting up >:/ | 10:19 |
feisar | hi my question's about virsh and vmbuilder ubuntu but #ubuntu-virt is a little quiet: if I created a vm using vmbuilder kvm ubuntu but i did it from the wrong directory, can I just mv my .qcow2 then edit the VM using virsh # edit VM_NAME? | 10:19 |
DarkStar1 | chmac: That just poo pooe'd my tgi friday feeling now | 10:19 |
chmac | DarkStar1: You will need to take down the machine in order to move data around onto new partitions, but it's not so hard | 10:20 |
DarkStar1 | chmac: what a stupid way to partition a machine | 10:21 |
DarkStar1 | Pissed off at them now | 10:21 |
chmac | DarkStar1: You pay peanuts, you get monkeys! | 10:23 |
chmac | DarkStar1: In my opinion, their SSD boxes are a better deal than their traditional disks, 120G is usually way more than we ever use on any of our web machines, and the SSDs perform much, much better. | 10:24 |
DarkStar1 | chmac: but let's be honest that kind of beetlejuice partition is spiteful | 10:24 |
chmac | DarkStar1: It's just stupid, but so is much of what I see at OVH! :-) | 10:24 |
DarkStar1 | SSds aren't worth the price atm and I need space. Lots of it. the mail archve I was trying to migrate is 100Gb zipped up | 10:26 |
chmac | DarkStar1: Then you want a spinning disk! | 10:26 |
RoyK | [slightly offtopic] hm... I just upgraded a server install to Quantal and installed ubuntu-desktop on it, and lightdm doesnt't start automatically | 10:30 |
AlexO | Hey, I opened the 3306 port (for mysql) but when I'm trying to connect with telenet on 3306, I get a connect refused, the strangest this is, that it's was working yesterday, but today It's not working anymore... | 10:33 |
AlexO | any ideas? | 10:33 |
AlexO | -this+thing* | 10:33 |
Lachezar | AlexO: netstat -antup | grep 3306 shows? | 10:34 |
AlexO | "tcp 0 0 0.0.0.0:3306 0.0.0.0:* LISTEN 15568/mysqld" | 10:35 |
Lachezar | AlexO: That seems OK. Are you sure you're telnet-ing into the right address? | 10:36 |
AlexO | Lachezar: yep, I just checked it again | 10:37 |
RoyK | AlexO: using ufw? | 10:38 |
AlexO | RoyK: iptables | 10:38 |
RoyK | AlexO: turn on logging in iptables, then, or just use ufw | 10:41 |
* RoyK uses ufw for 95% of his work | 10:41 | |
DarkStar1 | I only know a little iptables and 0 about ufw | 10:41 |
AlexO | RoyK: what do you mean but turn on logigng in iptables? You mean open the 3306 port ? | 10:42 |
AlexO | by* | 10:42 |
AlexO | I'm runing out of batery, need to find a plug brb | 10:44 |
AlexO | if you have any idea let me know my screen will still be there | 10:44 |
DarkStar1 | AlexO: he means you should enable iptables logging so that you can see infomation about the firewall | 10:45 |
=== n0ts_off is now known as n0ts | ||
=== doko_ is now known as doko | ||
=== n0ts is now known as n0ts_off | ||
=== cpg is now known as cpg|away | ||
=== n0ts_off is now known as n0ts | ||
=== gary_poster|away is now known as gary_poster | ||
=== mcclurmc_away is now known as mcclurmc | ||
=== cpg|away is now known as cpg | ||
jamespage | rbasak, is the subarch support SRU required to support diff arm archs? | 12:32 |
rbasak | jamespage: yes, but also any arch at all | 12:32 |
rbasak | jamespage: any ARM arch at all I mean | 12:32 |
jamespage | rbasak, ack | 12:33 |
jamespage | rbasak, looking at the branch now | 12:35 |
rbasak | thanks! | 12:35 |
jamespage | rbasak, 02- patch is missing from the branch :-( | 12:36 |
rbasak | jamespage: well that's embarrassing :-/ | 12:37 |
* rbasak digs it out | 12:37 | |
jamespage | rbasak, lol | 12:37 |
rbasak | jamespage: wondering how best to recover from this | 12:38 |
rbasak | jamespage: another upload with a bumped ubuntu revision? | 12:39 |
jamespage | rbasak, just push a new version to the branch | 12:39 |
rbasak | jamespage: roaksoax has already uploaded - it's in the unapproved queue | 12:39 |
jamespage | rbasak, ah! I wish he'd commented. | 12:39 |
jamespage | I think we can get that rejected | 12:39 |
rbasak | jamespage: ok, so do I need to bump the version, or does it count as not published? | 12:40 |
rbasak | I suppose nobody will have been able to install from it if it never actually entered precise-proposed? | 12:40 |
jamespage | rbasak, no - its not been published anywhere yet! | 12:40 |
rbasak | OK | 12:40 |
jamespage | rbasak, oh - was the upload for quantal or precise? | 12:40 |
rbasak | jamespage: both. roaksoax took the change upstream for quantal I think, and uploaded precise as-is | 12:41 |
jamespage | rbasak, sorry - LP was confusing me | 12:41 |
jamespage | rbasak, actually it looks OK in the queue | 12:42 |
jamespage | http://launchpadlibrarian.net/118134341/maas-enlist_0.4-0ubuntu1.1_0.4-0ubuntu1.2.diff.gz | 12:42 |
rbasak | jamespage: I'm confused. How? Or does that mean that roaksoax noticed and fixed it up? Also the pocket was wrong, which I know he fixed | 12:42 |
jamespage | rbasak, I suspect roaksoax fixed you up :-) | 12:43 |
rbasak | thanks roaksoax :) | 12:43 |
* rbasak commits fixes to the branch anyway | 12:44 | |
rbasak | At least it should match now | 12:44 |
rbasak | jamespage: sorry for the mess! | 12:44 |
jamespage | rbasak, no problem | 12:44 |
jamespage | rbasak, story of my day today | 12:44 |
jamespage | looking at stuff thats already been done! | 12:44 |
=== cpg is now known as cpg|away | ||
uvirtbot | New bug: #1057946 in mod-auth-mysql (main) "A patch to support Phpass hash (Used by Wordpress, phpBB3, etc)" [Undecided,New] https://launchpad.net/bugs/1057946 | 13:19 |
uvirtbot | New bug: #1062277 in nova (main) "092_add_instance_system_metadata migration fails when upgrading" [Undecided,New] https://launchpad.net/bugs/1062277 | 13:26 |
=== n0ts is now known as n0ts_off | ||
jamespage | zul: ^^ nova upgrade bug | 13:31 |
zul | jamespage: not cool | 13:33 |
=== n0ts_off is now known as n0ts | ||
selvodka | hello | 13:50 |
Ulfr | I just switched my frontend webservers from CentOS to Ubuntu because of an issue with PCRE, but now I'm trying to troubleshoot a problem my servers have after about a day of uptime where sessions will stack up in HAProxy and I haven't the foggiest why. Can anyone help me get started with troubleshooting? | 13:54 |
Ulfr | I use HAProxy to load balance to squid reverse proxies running apache2 on ubuntu 10.04 if that helps | 13:57 |
Mupfi | hi there :-) | 14:01 |
Mupfi | anybody here who is familiar with kerberos? | 14:03 |
feisar | hi what's the fix for the 'waiting for network configuration' on every boot with 12.04? | 14:03 |
Mupfi | i have problems with the krb5-kdc which woulden't start | 14:04 |
Mupfi | krb5kdc: cannot initialize realm ALKAR.INTERN - see log file for details | 14:04 |
uvirtbot | New bug: #1062314 in nova (main) "do_refresh_security_group_rules in nova.virt.firewall is very slow" [Undecided,New] https://launchpad.net/bugs/1062314 | 14:05 |
Mupfi | and the logfile tell me "krb5kdc: No such file or directory - while initializing database for realm ALKAR.INTERN" | 14:05 |
=== matsubara-afk is now known as matsubara | ||
=== niemeyer_ is now known as niemeyer | ||
=== kInOzAwA1 is now known as kInOzAwA | ||
=== n0ts is now known as n0ts_off | ||
=== n0ts_off is now known as n0ts | ||
=== dendrobates is now known as dendro-afk | ||
=== benji___ is now known as benji | ||
uvirtbot | New bug: #1062336 in nova (main) "nova-compute expects libvirtd group" [Undecided,New] https://launchpad.net/bugs/1062336 | 14:41 |
uvirtbot | New bug: #1062334 in nova (main) "nova-* not work - Scientific Linux 6.2" [Undecided,New] https://launchpad.net/bugs/1062334 | 14:43 |
hallyn | smb: hi, bug 914788, it's marked fix released and nominated for oneiric. but you seem to be talking about q new quantal patch? | 14:48 |
uvirtbot | Launchpad bug 914788 in libvirt "libvirt expexts qemu-dm in wrong path for xen" [Undecided,Fix released] https://launchpad.net/bugs/914788 | 14:48 |
smb | hallyn, Yes, since George made the patch for q but probably should have done a new report | 14:50 |
jamespage | zul, I enjoyed that bug ^^ | 14:51 |
smb | Things broke again after Debian removed the alternates setting which created xen-default link | 14:51 |
zul | i did as well :) | 14:51 |
hallyn | smb: can you mark it Triaged again? | 14:51 |
hallyn | I'll push the fix this afternoon then. tuesday is freeze :) | 14:51 |
smb | hallyn, not sure I can do that, but in theory we probably should make the main task triaged again and add a precise task marked fixed? | 14:52 |
smb | hallyn, Yeah, seem you have to accept the nomination at least. | 14:53 |
hallyn | smb: sigh, ok, let me set them how I THINK you mean them | 14:54 |
smb | hallyn, I could do the rest, I just cannot accept nominations for anything I cannot upload | 14:55 |
hallyn | smb: can you reload and tell me if that's ok? | 14:55 |
hallyn | smb: and then you're saying i can just take that patch, pop it onto quantal package, and expect itto work? | 14:55 |
smb | hallyn, I looks like I thought (probably need to decide to won't fix for O) and no not that simple | 14:56 |
smb | hallyn, Not sure it helps you but I liked a branch | 14:57 |
smb | hallyn, A bzr branch. Basically I had to pop back to the old patch and have pushed the modified ones on top | 14:57 |
hallyn | oh, *linked* a branch :) | 14:57 |
hallyn | i thought you were facebooking | 14:57 |
smb | errr yeah... :-P no | 14:57 |
smb | Just missing keys | 14:57 |
hallyn | smb: so does the branch work? | 14:58 |
smb | hallyn, I did the test-build from that | 14:58 |
hallyn | smb: ok, thanks. lemme look at it | 14:58 |
SpamapS | smb: so, any progress? | 14:58 |
SpamapS | smb: have not looked through my bugmail yet today | 14:58 |
smb | SpamapS, Tsk, well it works for me which is not progress I guess. And I asked for a more detailed list of steps | 14:59 |
smb | SpamapS, Not sure whether there is any special network setup involved or so | 14:59 |
SpamapS | smb: let me try my reproduction steps one more time.. | 15:01 |
uvirtbot | New bug: #914788 in libvirt (main) "libvirt expexts qemu-dm in wrong path for xen" [Undecided,Fix released] https://launchpad.net/bugs/914788 | 15:01 |
SpamapS | smb: hm, doing the basic steps doesn't cause it.. I'll try again with juju, maybe it is doing something special | 15:06 |
smb | SpamapS, Maybe I prepare another kernel in parallel that logs a lot of stack traces (for every dev_hold and dev_put). Its ugly but if you are the only one to be lucky I would let you run it and post the results. :) | 15:07 |
SpamapS | smb: for sure :) | 15:09 |
SpamapS | smb: indeed, something juju is doing in its destroy-environment is causing the issue reliably.. will boil it down to a test case | 15:11 |
=== n0ts is now known as n0ts_off | ||
=== n0ts_off is now known as n0ts | ||
hallyn | smb: i dunno, patch looks good to me (though i can't bzr import it bc of the way you moved the patch around in the queue) I'll run the qa regression tests against it, then push the package from source | 15:33 |
smb | hallyn, Grr and I was just using quilt and bzr locally. :-P But ok, just note patches, because there were two | 15:35 |
hallyn | smb: yeah, bzr just doesn't deal well with any sort of funky quilt usage. but I'm just doing 'bzr bd -S' out of your bzr tree, worked fine, so it'll have both your patches | 15:36 |
smb | hallyn, Ok, cool. Yeah, I probably should just have all patches unapplied and then replaced the file before applying them again | 15:37 |
smb | SpamapS, If you find a nice test case, please let me (err well the bug report) know. Otherwise the smb2 version is now up. | 15:43 |
SpamapS | smb: ok, working on it now | 15:47 |
uvirtbot | New bug: #1061964 in cloud-init "Config drive ensure local-hostname" [Medium,In progress] https://launchpad.net/bugs/1061964 | 15:50 |
=== basz is now known as basz_afk | ||
=== n0ts is now known as n0ts_off | ||
DarkStar1 | qq the cron task: 3 0 * * 1/3/5 will run this task 0003 every mon-wed-friday right? | 16:12 |
=== dendro-afk is now known as dendrobates | ||
=== ariel is now known as Guest54412 | ||
DarkStar1 | No one here? | 16:19 |
kendosan | hello guys, when i put my bash command in startup, when i reboot everything runs , but the bashscript that runs gives me error ffmpeg not found, i run a nohup on startup | 16:19 |
kendosan | so im not sure what is wrong | 16:19 |
kendosan | when i cd in directory and do nohup manualy everything is cool | 16:19 |
Guest54412 | hmm does anyone here have lamp installed | 16:19 |
patdk-lap | kendosan, that doesn't tell you anything? | 16:19 |
kendosan | hmm thats the thing im not sure what im doing :( | 16:20 |
kendosan | for example on startup commandline nohup sh /var/www/bash/nohup.sh >/dev/null 2>&1 & | 16:21 |
sarnold | kendosan: try without throwing away the error messages.. | 16:22 |
kendosan | er i will try | 16:22 |
patdk-lap | kendosan, oviously your ENVIROMENT is different | 16:22 |
patdk-lap | different path settings and other things | 16:22 |
kendosan | patdk, what do you mean ? i run ubuntu server 12.04 64bit | 16:23 |
TJ- | kendosan: Which 'nohup' is start-up using? The shell's version, or "/usr/bin/nohup" ? They have differences, especially if the start-up script isn't using a log-in shell | 16:23 |
kendosan | just a second please :) | 16:23 |
sarnold | TJ-: oh? I hadn't heard this :) thanks for the heads up. | 16:23 |
kendosan | i put the nohup commandline here /etc/rc.local | 16:23 |
patdk-lap | so no login, so you can't depend on anything being right | 16:24 |
patdk-lap | and have to make sure path and all are specified | 16:24 |
kendosan | it executes but, the thing is it does not run, when i type it in manual way :O | 16:24 |
kendosan | maybe im just confused with all this :D | 16:24 |
TJ- | kendosan: Usually that's because the shell environment at start-up is minimal compared to a log-in shell. Best debugging is to insert, in your start-up shell, a debug line of the form "env > /tmp/start-up.log" so you can look at the env it runs with | 16:25 |
kendosan | thanks i will try to do that | 16:26 |
TJ- | kendosan: s/your start-up shell/your start-up shell script/' | 16:26 |
hallyn | smb: my build of your libvirt tree oddly fails on an augeas lens test | 16:26 |
smb | hallyn, there is lenses in libvirt??? | 16:27 |
hallyn | in the tests, yeah | 16:27 |
hallyn | biam | 16:27 |
smb | Weird I have created a source package from my tree and ran it though sbuild... my only problem there is building sometimes ending up in /build which then bails because of an overlayfs bug | 16:29 |
smb | The second attempt with /build diverted to non-overlayed /home/... did succeed | 16:29 |
DarkStar1 | the cron task: 3 0 * * 1-5/2 will run this task every mon-wed-friday right? | 16:30 |
DarkStar1 | I'm new to this | 16:30 |
DarkStar1 | the aim is that it will run it at 0003h on those days. | 16:31 |
SpamapS | smb: installing your kernel next | 16:32 |
SpamapS | smb: the test case is still the same... just start and stop a container | 16:33 |
DarkStar1 | no one know? | 16:33 |
sarnold | DarkStar1: that's how I read it as well | 16:33 |
sarnold | DarkStar1: though I'd probably use 1,3,5 | 16:33 |
sarnold | no longer, no shorter, but more explicit | 16:33 |
SpamapS | smb: note that the container I start/stop also does an apt-get upgrade.. I wonder if there's something weird in there | 16:33 |
DarkStar1 | sarnold: ok | 16:33 |
DarkStar1 | with the commas right? I'm new to task automation. | 16:34 |
sarnold | DarkStar1: yes, with the commas | 16:35 |
smb | SpamapS, Hm, at least a bit more usage of the net devices. And the container is just created with "lxc-create -t ubuntu -n name -- -r quantal" ? | 16:35 |
=== n0ts_off is now known as n0ts | ||
SpamapS | smb: -t ubuntu-cloud | 16:41 |
SpamapS | smb: and -r precise | 16:41 |
SpamapS | smb: actually there's also some userdata fed in which does an apt-get upgrade and installs a few things... | 16:42 |
smb | SpamapS, I guess I will better wait for your detailed description... ;-P | 16:43 |
SpamapS | smb: you can use my juju branch which does it reliably.. | 16:44 |
SpamapS | smb: bzr branch lp:~clint-fewbar/juju/local-cloud-img | 16:44 |
SpamapS | smb: from that dir, PYTHONPATH=$PWD PATH=$PWD/bin:$PATH bin/juju bootstrap | 16:45 |
=== tyhicks` is now known as tyhicks | ||
SpamapS | smb: and PYTHONPATH=$PWD PATH=$PWD/bin:$PATH bin/juju deploy wordpress | 16:45 |
SpamapS | smb: then the same but 'juju destroy-environment' | 16:46 |
smb | SpamapS, Probably you should write that down then *in* the bug report. My memory is very limited (especially when it is actually the weekend) | 16:46 |
SpamapS | smb: I'm hoping to have a lower level reproducer | 16:47 |
smb | SpamapS, Me too :) | 16:48 |
SpamapS | smb: so far, unable to reproduce with your stacktrace happy kernel | 16:51 |
SpamapS | smb: n/m, reproduced | 16:52 |
=== n0ts is now known as n0ts_off | ||
SpamapS | smb: want syslog? | 16:52 |
smb | SpamapS, Yes please in the bug report. | 16:52 |
smb | SpamapS, Probably just a matter of timing as it now is a bit slower with the lots of output | 16:53 |
hallyn | smb: even plain libvirt is faling to build. hopefully i just messed up my localhost, lemme try on a clean instance | 17:03 |
smb | hallyn, Yeah, hopefully. Well, in some way. But I could not explain how that test did not bother me otherwise... | 17:05 |
frojnd | Hi there. I don't know why but when I try to change /etc/motd as root it won't update it. And when I reopen file with vim it's still last message | 17:27 |
SpamapS | hallyn: did you come up with a plan for bisecting btw? | 17:27 |
hallyn | SpamapS: bisecting what? | 17:28 |
SpamapS | hallyn: I can reproduce 100% of the time on this hardware with the fairly lengthy steps I posted in the bug.. but have not yet boiled it down to anything less than a long juju deploy/destroy | 17:28 |
sarnold | frojnd: see update-motd(5) | 17:28 |
SpamapS | hallyn: netns bug | 17:29 |
hallyn | SpamapS: so you can't reproduce with the test program attached to the bug? | 17:29 |
SpamapS | hallyn: no, this is a different problem | 17:29 |
SpamapS | hallyn: the simple fix of flushing the route cache addressed that | 17:29 |
hallyn | are you sure there's no zombie tasks pinning the netns? | 17:30 |
SpamapS | hallyn: to reproduce I have to boot up an ubuntu-cloud based container and then stop it | 17:30 |
frojnd | This happens if I'm editing with nano too.. So I must do something wrong. Why after editing /etc/motd as root this new login message won't exist? | 17:30 |
SpamapS | hallyn: no, how would I check? | 17:30 |
hallyn | anyway no, i've been leaving that to smb to not reproduce effort | 17:30 |
hallyn | ps -ef should show the defunct tasks... | 17:30 |
SpamapS | nothing there | 17:30 |
hallyn | SpamapS: what is the bug#? | 17:30 |
SpamapS | bug #1021471 | 17:30 |
uvirtbot | Launchpad bug 1021471 in linux "clone() hang when creating new network namespace (dmesg show unregister_netdevice: waiting for lo to become free. Usage count = 2)" [High,Confirmed] https://launchpad.net/bugs/1021471 | 17:30 |
SpamapS | hallyn: I posted the steps to reproduce using my juju branch | 17:31 |
hallyn | SpamapS: given that it conflicts with the bug which smb did fix, i'm not sure we can bisect now :) | 17:31 |
hallyn | still i could try. but if smb thinks he's about got it nailed... | 17:31 |
smb | hallyn, No not that one. Just got data for it to look trhough | 17:32 |
hallyn | ok. lemme finis up the libvirt one and then i'll see if i can reproduce and join the party | 17:32 |
smb | hallyn, Though if you want to run bisecting maybe from the other end (take our mainline kernels between 3.2 and 3.6 as a start) | 17:33 |
smb | Usually it turns out to be one -rc1 and then the fun begins | 17:33 |
smb | Err make that 3.5 | 17:34 |
hallyn | smb: happy to. i was going to do linus' tree, what do you mean by 'our mainline'? (i think i can find it from old emails, but if you have git url handy...) | 17:34 |
hallyn | right :) | 17:34 |
hallyn | 3.2..3.6 was going to be my range | 17:34 |
hallyn | uh 3.5 wtf? | 17:34 |
smb | http://kernel.ubuntu.com/~kernel-ppa/mainline/ | 17:35 |
hallyn | thanks. ttyl | 17:35 |
frojnd | Is there a manual or something how to change default ssh login prompt? Changing /etc/motd as root doesn't work unfortunatelly. | 17:36 |
smb | Well, we know 3.5 is broken (3.6 apparently not) So you could go that way as well, but at least for that I know that the removed the route cache for one thing | 17:36 |
hallyn | yeah, i meant 3.5, i don't knwo why i keep mistyping. ttyl :) | 17:36 |
=== n0ts_off is now known as n0ts | ||
SpamapS | hallyn: while you're still playing, I'm going to try 3.4.12-quantal | 17:49 |
SpamapS | hallyn: as in, http://kernel.ubuntu.com/~kernel-ppa/mainline/v3.4.12-quantal/ | 17:49 |
hallyn | sounds good. | 17:50 |
=== n0ts is now known as n0ts_off | ||
=== mcclurmc is now known as mcclurmc_away | ||
hallyn | smb: regression tests still running; built fine on fresh instance; intend to push this afternoon. | 18:09 |
hallyn | zul: you have no libvirt changes to push to quantal right? | 18:10 |
zul | nope | 18:10 |
smb | hallyn, ok, phew | 18:10 |
SpamapS | hallyn: I am unable to reproduce on 3.4.12 btw | 18:26 |
hallyn | SpamapS: in your recipe, should i just 'apt-get install juju'? | 18:27 |
SpamapS | hallyn: no | 18:27 |
SpamapS | hallyn: it shows, bzr branch the specific branch I'm using | 18:27 |
SpamapS | hallyn: it still happens with stock juju, but not as frequently for some reason | 18:27 |
hallyn | you don't show building/installing it | 18:27 |
SpamapS | bzr branch | 18:27 |
SpamapS | just run it from there | 18:27 |
hallyn | i did bzr branch, entered it, then you say run 'juju', but juju was not installed (and ./juju is a dir) | 18:28 |
hallyn | oh hm | 18:28 |
hallyn | so it should be coming from my new path. but it's not happy with it. why? | 18:29 |
hallyn | (checking) | 18:29 |
SpamapS | I set PATH :) | 18:29 |
hallyn | SpamapS: well i've since apt-get installed juju, and id o see juju in my path, so i'm guessing there was a dependency of juju which wasn't installed. (the error has scrolled off ths creen) | 18:30 |
hallyn | proceeding | 18:30 |
SpamapS | ahh right | 18:31 |
SpamapS | python-txzookeeper or python-txaws or twisted or something like that :-P | 18:31 |
=== three18t- is now known as three18ti | ||
SpamapS | hallyn: ok, so in the interest of narrowing your bisect ... 3.4.12 is fine.. what would be the other end of the spectrum to try? 3.5.5? | 18:33 |
uvirtbot | New bug: #1062474 in nova "Migrations do not populate volume_id_mappings and instance_id_mappings completely" [Undecided,New] https://launchpad.net/bugs/1062474 | 18:36 |
hallyn | SpamapS: not sure, hold on lemme check around which release the route cache was removed | 18:38 |
hallyn | btw, first attempt on c1.xlarge did NOT reproduce. trying again | 18:38 |
SpamapS | hallyn: if it fails again, perhaps let me login and I'll see if there are assumptions I've left out | 18:38 |
hallyn | heck, the commit "ipv4: reintroduce route cache garbage collector" sounds suspicious :) | 18:39 |
SpamapS | hallyn: hah yeah | 18:41 |
hallyn | SpamapS: 3.5-rc7 did not have commit 89aef8921bfbac22f00e04f8450f6e447db13e42. 3.6 did. (that being the commit removing the routing cache) | 18:41 |
SpamapS | hallyn: I haven't actually confirmed that 3.6 doesn't reproduce with my particular issue | 18:41 |
SpamapS | hallyn: ok, 3.5.5 does reproduce | 18:43 |
SpamapS | hallyn: also reproduces with the original reproducer, so no news there | 18:43 |
SpamapS | [ 191.483843] unregister_netdevice: waiting for lo to become free. Usage count = 2 | 18:44 |
SpamapS | thats 2, so route cache + "the other thing" | 18:44 |
hallyn | SpamapS: hrmph, juju seems to have messed with my ptys | 18:44 |
=== n0ts_off is now known as n0ts | ||
SpamapS | hallyn: ?? | 18:44 |
SpamapS | hallyn: ok, so I'll try 3.6-rc1 next | 18:45 |
hallyn | i ran 'juju bootstrap' in byobu, exited the shell, and now defunct java tasks are pinning /dev/pts/1, and re-starting byobu hangs | 18:46 |
hallyn | rebooting, not sure what state i just left things in... sigh | 18:46 |
SpamapS | hallyn: oh thats interesting.. zookeeper should have daemonized. | 18:47 |
SpamapS | hallyn: thats likely the culprit there | 18:47 |
SpamapS | hallyn: so i'll do 3.6-rc1, and then 3.5-rc7 | 18:49 |
SpamapS | hallyn: yeah looks like we need to use start-stop-daemon to run ZK so it will let go of the terminal | 18:52 |
SpamapS | hallyn: you should have just killed the java process | 18:52 |
=== n0ts is now known as n0ts_off | ||
SpamapS | hallyn: AHA! | 18:57 |
SpamapS | hallyn: 3.6 still has it | 18:57 |
SpamapS | hallyn: well, 3.6-rc1 | 18:58 |
SpamapS | hallyn: so, the route cache issue is separate | 18:58 |
SpamapS | [ 253.892068] unregister_netdevice: waiting for lo to become free. Usage count = 1 | 18:58 |
SpamapS | thats on 3.6 rc1 | 18:58 |
smb | SpamapS, Unfortunately there are a few additions to the whole mess that are (or where) not even in upstream linux when I looked last | 19:01 |
SpamapS | smb: well I think you were looking at the route cache issue and using stgraber's reproducer only | 19:02 |
hallyn | SpamapS: interesting | 19:02 |
SpamapS | trying 3.6-quantal now | 19:02 |
smb | I was also following a few hints from upstream, but they pointed to a follow up on things that happened after removing the route cache. So things may have gone jojo | 19:03 |
SpamapS | well if nothing else, its clear that this is a separate bug from the route cache issue | 19:04 |
hallyn | SpamapS: it might be worth opening a separate upstream kernel bugzilla | 19:04 |
hallyn | except, of course, we probably need a reproducer outside of juju :) | 19:04 |
SpamapS | yeah | 19:05 |
SpamapS | I think its something in the timing | 19:05 |
SpamapS | hallyn: lxc-stop returns immediately, right, you still need lxc-wait to know that it is stopped? | 19:06 |
=== cpg|away is now known as cpg | ||
hallyn | SpamapS: i think so, though the python api one can wait iirc | 19:08 |
hallyn | SpamapS: though lxc_stop "sort of" waits, by waiting for the monitor socket to close | 19:09 |
SpamapS | ah ok | 19:11 |
SpamapS | weird.. why do our cloud images have ntfs installed? :-P | 19:12 |
Daviey | SpamapS: ntfs ebs volume ! :) | 19:12 |
SpamapS | so much win | 19:13 |
SpamapS | hallyn: ok, confirmed still in 3.6 | 19:13 |
SpamapS | hallyn: I'll try to boil this down to at least just lxc commands to reproduce | 19:13 |
hallyn | SpamapS: gah, no, can't reproduce on amazon with juju. i'll just try quickly running lxc testsuite and see ifthat helps | 19:16 |
hallyn | smb: libvirt pushed, thanks. | 19:16 |
smb | hallyn, At least something that worked today. :) | 19:16 |
smb | thanks | 19:16 |
hallyn | haha no kidding. still no idea about psivaa's bug | 19:17 |
hallyn | mdeslaur: if i said that after psivaa installs a new quantal desktop and installs kvm, then virt-manager, virt-manager fails to create/start VMs, but once you manually create one with virsh define and start that, then virt-manager succeeds | 19:18 |
hallyn | mdeslaur: would that ring any bells for you? | 19:18 |
mdeslaur | hallyn: no...what does virt-manager do? | 19:18 |
hallyn | my only guess would be bad perms inside /var/lib/libvirt/images, but that doesn't really make sense either | 19:18 |
hallyn | it manages to start kvm, but then kvm bombs with with -EPERM trying to actually use kvm | 19:19 |
hallyn | (as seen in the /var/lib/libvirt/qemu/x.log) | 19:19 |
hallyn | the bug # is 1057024 fwiw | 19:19 |
mdeslaur | hallyn: trying on a fresh install now | 19:22 |
hallyn | mdeslaur: thanks. i've tried that 3 times no with no success, maybe i'm unconsciously doing something blatantly different... | 19:23 |
mdeslaur | hallyn: with no success? you mean you couldn't reproduce his issue, or you couldn't get it to work? | 19:23 |
hallyn | couldn't reproduce it | 19:24 |
hallyn | worked for me every time | 19:24 |
=== basz_afk is now known as basz | ||
mdeslaur | hallyn: worked fine for me | 19:29 |
mdeslaur | hallyn: not sure what's going wrong for him | 19:29 |
mdeslaur | hallyn: hardware problem? | 19:30 |
hallyn | mdeslaur: kvm does run fine by hand, or using 'virsh'. | 19:31 |
hallyn | mdeslaur: but there are funky /dev/sda errors in syslog | 19:32 |
hallyn | worse, psivaa has *two* machiens that do it | 19:32 |
mdeslaur | hallyn: hrm | 19:32 |
hallyn | so i started wondering if there is a corrupt archiv emirror... | 19:32 |
hallyn | i dunno | 19:32 |
mdeslaur | hallyn: I did a fresh install, dit "apt-get install virt-manager", did "apt-get install qemu-kvm", added myself to the libvirtd group, rebooted and everything worked | 19:33 |
hallyn | mdeslaur: thanks for trying. i'm out of ideas | 19:33 |
hallyn | psivaa: any chance you can post dmesg from the other machine that also does this for you? (to the bug) | 19:34 |
hallyn | are they the same kind of laptop? | 19:34 |
=== kInOzAwA2 is now known as kInOzAwA | ||
mdeslaur | hallyn: my iso was in my home directory, not sure that's relevant | 19:37 |
hallyn | mdeslaur: hm, so was mine. | 19:38 |
bgoliveira | Hello everyone. Could anyone can explain me why /etc/resolf.conf turns into a sym link to /run/resolvconf/resolv.conf ? | 19:40 |
bgoliveira | I'm having to upgrate my servers from 11.04 to 12.04 ...and I just realised this because a normal user can not use the host comand. | 19:41 |
SpamapS | hallyn: ok so here is how lxc-create is invoked.. | 19:41 |
SpamapS | 3742 execve("/usr/bin/lxc-create", ["lxc-create", "-n", "clint-local-ci-ubuntu-0", "-t", "ubuntu-cloud", "--", "--debug", "--hostid", "clint-local-ci-ubuntu-0", "-r", "precise", "--userdata", "/tmp/tmplp3Y7E"], [/* 13 vars */]) = 0 | 19:41 |
sarnold | bgoliveira: it's part of the 'resolvconf' package, which rewrites the resolv.conf file for different name servers based on dhcp results | 19:41 |
SpamapS | hallyn: and lxc-start.. | 19:42 |
SpamapS | 3816 execve("/usr/bin/sudo", ["sudo", "lxc-start", "--daemon", "-n", "clint-local-ci-ubuntu-0", "-l", "DEBUG", "-o", "/home/clint/.juju/data/clint-local-ci/units/ubuntu-0/container.log"], [/* 21 vars */]) = 0 | 19:42 |
SpamapS | followed by.. | 19:42 |
SpamapS | 3820 execve("/usr/bin/sudo", ["sudo", "lxc-wait", "-n", "clint-local-ci-ubuntu-0", "-s", "RUNNING"], [/* 21 vars */]) = 0 | 19:42 |
psivaa | hallyn, dmesg attached to the bug | 19:43 |
hallyn | psivaa: thanks | 19:43 |
mdeslaur | hallyn: this is definitely odd: Could not access KVM kernel module: Permission denied | 19:44 |
hallyn | SpamapS: hm, precise container, <shrug> maybe that makes a difference | 19:44 |
psivaa | hallyn, one of them is Hp pavilion g6 and the other is Dell Inspiron | 19:44 |
hallyn | mdeslaur: right. | 19:44 |
SpamapS | hallyn: sure, I'll try a quantal container | 19:44 |
=== n0ts_off is now known as n0ts | ||
hallyn | psivaa: and so where do you store the iso for virt-manager to pick up? | 19:45 |
mdeslaur | psivaa: what's "getfacl /dev/kvm" show? | 19:46 |
bgoliveira | sarnold for example, I can ping a website. But, ping only can resolv the name because it has a SUID bit active. Host doesn't have a SUID bit active ... any ideia? | 19:46 |
Daviey | woot, sbeattie extended his ~ubuntu-server membership.. Great to have you. :) | 19:47 |
mdeslaur | psivaa: ? | 19:49 |
hallyn | mdeslaur: i fear it's way past EOD for psivaa | 19:50 |
mdeslaur | hallyn: consolekit is supposed to set extended acls on the kvm device to your user | 19:50 |
psivaa | mdeslaur, hallyn, ohh the iso's are stored on in desktop in one machine and home/myname/iso/ubuntu in anouter | 19:50 |
mdeslaur | hallyn: maybe he's using some retro-grouch desktop environment | 19:51 |
hallyn | mdeslaur: interesting | 19:51 |
hallyn | i hate magic | 19:51 |
mdeslaur | psivaa: can you do a "getfacl /dev/kvm"? | 19:51 |
psivaa | mdeslaur, just added to the bug | 19:52 |
hallyn | mdeslaur: but, /dev/kvm is group kvm, group read-write, and libvirt-qemu user is in default group kvm... | 19:53 |
hallyn | so unless virt-manager manages to run kvm not in libvirt-qemu user, ... | 19:53 |
mdeslaur | hallyn: he has group::--- | 19:54 |
mdeslaur | hallyn: incorrect permissions, kvm group doesn't have proper rights | 19:54 |
=== n0ts is now known as n0ts_off | ||
mdeslaur | hallyn: (yeah, the extended acl isn't used) | 19:54 |
psivaa | hallyn, mdeslaur on the other machine which now works after following th wiki from hallyn i have group::rw- | 19:54 |
mdeslaur | psivaa: so something is breaking the permissions on that device | 19:55 |
mdeslaur | psivaa: do you have /lib/udev/rules.d/40-qemu-kvm.rules ? | 19:56 |
hallyn | SpamapS: say, ... you're using juju, so you're not using lxcbr0? are you using virbr0? | 19:57 |
hallyn | well you're not specifying -f, so i guess you must be using lxcbr0 | 19:57 |
psivaa | mdeslaur, yes | 19:57 |
SpamapS | hallyn: this branch does use lxcbr0 | 19:57 |
hallyn | SpamapS: sadly, i still can't reproduce. | 19:58 |
SpamapS | hallyn: yeah I'm working out a series of lxc-* commands to do it | 19:58 |
hallyn | doing a loop of while [ $c -lt 100 ]; do sudo lxc-start -n p1 -d & sudo lxc-wait -n p1 -s RUNNING; sudo lxc-stop -n p1; done | 19:58 |
hallyn | SpamapS: can you pastebin your userdata file? | 19:58 |
psivaa | mdeslaur, just to make sure, i did a fresh quantal install, i did apt-get install kvm and install virtual machine manager using software centre, addded the user to the group and logged out and in back | 19:59 |
SpamapS | hallyn: sure | 19:59 |
SpamapS | hallyn: thats part of what I'm working on :) | 19:59 |
mdeslaur | psivaa: quite odd, I'm not sure how your permissions get like that | 19:59 |
mdeslaur | psivaa: if you reboot, do they go back to normal? | 20:00 |
psivaa | mdeslaur, i'm not sure which permissions do you mean here, but rebooting alone did not solve the issue, it was running the sequence given in https://wiki.ubuntu.com/SergeHallyn_libvirtnest and then reboot solved it | 20:02 |
hallyn | mdeslaur: on the other machine of psivaa's, definately /dev/kvm was rw-rw---- when it was not working | 20:02 |
mdeslaur | hallyn: so, two different problems then | 20:02 |
Daviey | SpamapS: Hey, are we expecting another juju upload? | 20:03 |
mdeslaur | hallyn: it definitely can't work if the kvm group doesn't have access | 20:03 |
psivaa | mdeslaur, although i only tried once with the wiki sequence, i could try once more if you think its needed | 20:03 |
hallyn | mdeslaur: yeah... actually, psivaa, wasn't that other lapto (with bad kvm perms) one you had a bad upgrade on? | 20:03 |
Daviey | SpamapS: specifically, i want bug 1061286 included :) | 20:03 |
uvirtbot | Launchpad bug 1061286 in juju "juju bootstrap returned ERROR Invalid 'cpu_count' constraint '1.0'" [Medium,Fix committed] https://launchpad.net/bugs/1061286 | 20:03 |
gholms | smoser: Any idea why the resizefs cloud-init module opts to create a new device node? | 20:03 |
gholms | I *suspect* it's so it doesn't have to attempt to locate the one under /dev, but I'd like to be sure. | 20:04 |
SpamapS | hallyn: http://paste.ubuntu.com/1262631/ | 20:05 |
SpamapS | Daviey: yes I am, and it would include that | 20:05 |
SpamapS | Daviey: hoping on Tuesday or Wednesday | 20:05 |
hallyn | psivaa: if you reboot the second machine, which doesn't have group write perms on /dev/kvm (as shown by getfacl), does /dev/kvm *then* have group write perms? | 20:05 |
Daviey | SpamapS: do you care if i monkey that patch in now, with your upload superseeding it? | 20:06 |
smoser | gholms, yes. | 20:06 |
uvirtbot | New bug: #1062518 in maas (main) "[FFe] New upstream release" [Undecided,New] https://launchpad.net/bugs/1062518 | 20:06 |
smoser | how else would it figure out the device name ? | 20:06 |
SpamapS | Daviey: go ahead, we may not get our act together for this 0.6 release in time to squeeze it into quantal | 20:06 |
smoser | so it just doesn't bother. | 20:07 |
Daviey | SpamapS: ok, thanks | 20:07 |
gholms | smoser: By looking it up in /proc/partitions? | 20:07 |
gholms | smoser: The problem I'm running into here is that Fedora and friends mount /run with nodev. | 20:07 |
=== mcclurmc_away is now known as mcclurmc | ||
gholms | Same with /tmp and a host of other things. :-\ | 20:07 |
smoser | gholms, we could just make it in /dev/ | 20:10 |
smoser | that is likely to not be mounted without 'nodev' | 20:10 |
smoser | :) | 20:10 |
smoser | err.. wahtever . i meant that probably we can create device nodes in /dev/ | 20:10 |
smoser | just try that rather than /tmp | 20:10 |
smoser | it seems wasteful and pointless to find do a stat on /, get exactly the data we need, then to go looking in some other location for a name that doesnt matter. | 20:11 |
gholms | Heh | 20:13 |
gholms | smoser: Would you be against a patch that just grabs the device numbers from /, looks up the matching blockdev in /proc/partitions, and uses that directly? | 20:13 |
gholms | I mean, I get *why* it works the way it does. | 20:14 |
gholms | It just... adds more variables that can break. | 20:14 |
sbeattie | Daviey: I like to delude myself into believing I know something about servers. | 20:15 |
SpamapS | sbeattie: you know nothing john snow | 20:15 |
smoser | gholms, it is less variables. | 20:15 |
* SpamapS apologizes.. the GoT withdrawals are getting intense | 20:15 | |
smoser | you're suggesting more variables (parsing /proc/partitions incorrectly, or /proc not being mounted) | 20:16 |
SpamapS | hallyn: success! | 20:16 |
sbeattie | SpamapS: no worries... Winter is coming. :) | 20:16 |
SpamapS | hallyn: I have a simpler reproducer now | 20:16 |
gholms | smoser: touché | 20:16 |
smoser | just make the device in /dev/.tmp.cloudinit | 20:16 |
* gholms hrms | 20:16 | |
gholms | Yeah, that's probably the best bet at this point. | 20:16 |
Daviey | sbeattie: hah, get a grip.. you've probably done more server package uploads than me :) | 20:16 |
SpamapS | hallyn: http://paste.ubuntu.com/1262645/ | 20:17 |
SpamapS | hallyn: now, here's the tough part.. it only seems to happen if you wait for the system to fully boot | 20:17 |
SpamapS | hallyn: the way that script works, you have to press enter after cloud-init says "done booting" | 20:17 |
SpamapS | hallyn: interrupting it earlier seems to not reproduce | 20:17 |
SpamapS | hallyn: which should actually help narrow down what is causing the issue | 20:17 |
SpamapS | hallyn: I need to take a break and get some lunch, but hopefully that helps | 20:22 |
SpamapS | hallyn: I wonder if the proprietary WL driver has anything to do with this. | 20:22 |
psivaa | hallyn, yes the bad upgrade one is the one with wrong permissions, to check if rebooting recovers the group permissions, i need to reboot *this machine, | 20:25 |
Daviey | SpamapS: fancy reviewing http://pb.daviey.com/wTv8/ ? | 20:28 |
psivaa | hallyn, mdeslaur rebooting has made the getfacl /dev/kvm -> group::rw- and i am able to create vm's now | 20:33 |
mdeslaur | psivaa: interesting...maybe the qemu-kvm postinst isn't setting the device permissions as intended | 20:35 |
hallyn | mdeslaur: in the past psivaa has shown /dev/kvm perms to b eright when still unable to create vms | 20:37 |
mdeslaur | hallyn: ok | 20:37 |
mdeslaur | hallyn: I just confirmed postinst seems to work too, so it's not that | 20:38 |
mdeslaur | hallyn: I'm stumped now | 20:38 |
psivaa | mdeslaur, im doing a fresh quantal install now ill let you know if rebooting alone gives the req'd permission, iirc it did not, | 20:39 |
mdeslaur | psivaa: ok. I'm eod, but leave the info in the bug. | 20:39 |
psivaa | mdeslaur, ack | 20:40 |
Daviey | SpamapS: would also seem prudent to include http://pb.daviey.com/0DDk/ i guess | 20:40 |
uvirtbot | New bug: #1061286 in juju "juju bootstrap returned ERROR Invalid 'cpu_count' constraint '1.0'" [High,In progress] https://launchpad.net/bugs/1061286 | 20:41 |
uvirtbot | New bug: #1062538 in php5 (main) "package php5-dev 5.3.10-1ubuntu3.4 failed to install/upgrade: subprocess installed post-installation script returned error exit status 2" [Undecided,New] https://launchpad.net/bugs/1062538 | 20:41 |
=== n0ts_off is now known as n0ts | ||
SpamapS | Daviey: indeed, that one fixes canonistack | 20:48 |
Daviey | SpamapS: you don't make my life easier, you know. :P | 20:48 |
SpamapS | Daviey: why not just make a snapshot from trunk? | 20:48 |
Daviey | SpamapS: well it's getting that way... you know the CURRENT package FTBFS? | 20:49 |
SpamapS | no | 20:49 |
SpamapS | but I suspect its a race | 20:49 |
SpamapS | had to retry the PPA builds a few times | 20:49 |
SpamapS | anyway, I have not eaten yet, and I think if I delay any further my stomach will eat me | 20:49 |
Daviey | ok | 20:49 |
hallyn | SpamapS: can you use the laptop over wired and not load the proprietary wireless driver? | 20:51 |
uvirtbot | New bug: #1061678 in juju "OpenStack provider should fall back to local-ipv4 if public-ipv4 is not found" [Critical,Fix committed] https://launchpad.net/bugs/1061678 | 20:51 |
=== n0ts is now known as n0ts_off | ||
hallyn | SpamapS: nope, repro.sh still doesn't reproduce it here. jinkeys! lemme try that laptop over yonder | 21:03 |
hallyn | SpamapS: nope, not there either | 21:20 |
SpamapS | hallyn: this one has no wired conn.. but I can try w/ my pro | 21:44 |
=== n0ts_off is now known as n0ts | ||
SpamapS | hallyn: I suppose I can try w/ everything cached and just shut down wireless too | 21:46 |
hallyn | SpamapS: yup, should work. of course it's still possible that it's just your fast ssd making it happen | 21:53 |
=== n0ts is now known as n0ts_off | ||
SpamapS | hallyn: I've had it work quite a high percentage of the time on the spinning rust too | 21:57 |
SpamapS | hallyn: ok I forgot to rmmod wl .. but it did happen while on wired LAN and eth1 (wl) was down | 21:58 |
SpamapS | hallyn: ok, actually, no.. it was the route cache one | 22:02 |
SpamapS | hallyn: so indeed, it may only be my air that reproduces with the route cache fix smb made | 22:02 |
hallyn | utlemming: precise cloud images have empty /dev | 22:32 |
utlemming | hallyn: I think that might affect oneiric and quantal too | 22:33 |
utlemming | hallyn: I believe that it is excluded in the build since the initramfs populates it | 22:34 |
SpamapS | hallyn: progress! it only happens when wl is involved | 22:34 |
hallyn | utlemming: no, quantal has /dev populated | 22:35 |
utlemming | hallyn: right, I just checked.... | 22:35 |
hallyn | utlemming: ok, i just wanted to make sure it wasn't intentional | 22:35 |
utlemming | hallyn: nope...I'll check into what is happening | 22:36 |
SpamapS | hallyn: oo, and rmmod on wl makes the problem go away | 22:36 |
SpamapS | err no | 22:36 |
SpamapS | ignore that | 22:36 |
hallyn | utlemming: great, thanks | 22:37 |
basil60 | hi i've had complaints from isp about my home network sending unsolicited emails. I've run some "extensive" tests on my pc, and found nothing. I suspect it may come from my linux server (ubuntu 11.04). Any suggestions on how I may test it? | 22:40 |
sarnold | basil60: if you've got a wordpress or phpbb or something similar installed, that's a reasonable possibility | 22:41 |
sarnold | if your machine has been _rooted_, you may not be able to do much examining "from inside" the system. But if it is a simple / stupid exploit, it might be fixable from within... | 22:41 |
basil60 | not running Wordpress | 22:42 |
basil60 | I ran "rootkitcheck" yesterday - it found nothinh | 22:42 |
sarnold | basil60: does 'netstat -anp' show any unexpected connections? | 22:43 |
basil60 | can i export that netstat command to a text file? | 22:45 |
sarnold | netstat -anp > /tmp/file | 22:45 |
sarnold | :) | 22:45 |
basil60 | netstat -anp >/tmp/file/netstat | 22:46 |
sarnold | that'll fail unless you already have a /tmp/file directory | 22:46 |
sarnold | netstat -anp > /tmp/netstat would work alright | 22:46 |
basil60 | dovecot appears to be running - wasn't even sure i had that turned on | 22:50 |
SpamapS | smb`: so, good news, the problem is just in the wl drivers.. | 23:02 |
SpamapS | smb`: see bug report for more info | 23:02 |
hallyn | SpamapS: ok, i wasn't sure which order the bug and irc comments went in :) i thought after the lp comment you decided it was NOT jsut the drivers. | 23:03 |
hallyn | so, phew ;) | 23:03 |
SpamapS | it *is* | 23:03 |
hallyn | and, nice reminder about proprietary drivers, i guess | 23:03 |
SpamapS | just that you can't rmmod wl to clear it | 23:03 |
SpamapS | I thought wl was supposed to be fully free'd at some point | 23:03 |
hallyn | so, WOOT! smb might have it fixed in quantal release then? | 23:03 |
hallyn | <shrug> | 23:03 |
SpamapS | hallyn: yes, his last patch addresses the route cache problem that the reproducing C program shows perfectly | 23:04 |
hallyn | SpamapS: right, but we're past kernel freeze | 23:04 |
SpamapS | hallyn: ok, so worst case its in that 0-day SRU kernel that we always end up shipping anyway ;) | 23:04 |
* SpamapS adds a "its past beer-thirty" disclaimer to that comment | 23:05 | |
hallyn | all i see there is 'beer', and i'm outta here :) | 23:05 |
hallyn | tttyl | 23:05 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!