=== BenC__ is now known as BenC [06:39] moin [07:21] morning === tjaalton_ is now known as tjaalton === jussio1 is now known as jussi [08:04] * ppisati notices that if you don't read lkml on a daily basis, the associated folder tends to explode... [09:02] morning ... [09:57] apw: morning, i'm trying to add 'ln -s' workaround in lxc-start-ephemeral [09:59] apw: i don't fully understand where shall i add this 'ln -s', upper or lower. i think it should be lower directory [09:59] cooloney, well if it is in the lower then its a one off thing [09:59] cooloney, though as its a bodge, in the upper seems fine [10:03] apw: cd ${upper}; ln -s rootfs / ? [10:04] ln -s . rootfs [10:04] i'd guess, as we want /rootfs to mean / [10:14] apw: oh, i tried that. actually it doesn't work, [10:14] and why does it not work ? [10:14] apw: rootfs will be an endless loop point [10:14] ? [10:14] if you ln -s . rootfs it cannot be an endless loop [10:17] i assume we end up with /rootfs in the names because somewhere you have a rootfs directory (a real directory) which is used to build the fake namespace [10:17] inside that directory we need to add the link, not replace it [10:18] as a test i would try and run "ln -s . rootfs; ls -ld /etc /rootfs/etc" as the command payload for an ephemeral lxc start thing [10:18] cooloney, ^^ [10:19] and see what that produces [10:19] so in upper there is rootfs -> . [10:19] ok that sounds right [10:19] so what does ls -ld /etc /rootfs/etc do in that [10:19] it looks like overwrite the real directory in lower, [10:20] oh, the lxc hangs in the middle [10:20] whats in rootfs in lower ? [10:20] $ ls /var/lib/lxc/p1 [10:20] config fstab rootfs [10:20] this is lower [10:21] rootfs is a directory containing everything for lxc container [10:21] ok get rid of the symlink in upper [10:21] and ln -s . /var/lib/lx/p1/rootfs/rootfs [10:21] and try that [10:21] after i added a link in upper named rootfs, it probably overwrite this real one [10:22] yeah i don't want the top rootfs replaced i want to add rootfs/rootfs effectivly as a link [10:22] so remove the upper one for now, and add the lower one and tell me waht that does [10:22] note the double rootfs at the end [10:24] cooloney, ^^ [10:25] cd ${upper} sudo ln -s . ${lower}/rootfs/rootfs cd - [10:25] ? [10:25] apw: i added this [10:25] no that is in upper [10:25] oh, sorry [10:26] actually that might be in lower, as its absolute [10:26] so you don't need the cd [10:26] ln -s . ${lower}/rootfs/rootfs [10:26] looks right [10:27] cooloney, and ? does it work / [10:27] ? [10:28] apw: oh, no, i got the same error with /rootfs [10:28] ok [10:28] actually which error [10:29] this directory /var/lib/lx/p1/rootfs, is it persistant? always there? [10:29] http://pastebin.ubuntu.com/1144614/ [10:29] apw: yeah, always there [10:29] since lower is a real directory [10:30] ok rathan than putting it in the ephermal scripts like you are [10:30] it might be ln -s rootfs ${lower}/rootfs/rootfs [10:30] remove that change, and run: ln -s . /var/lib/lx/p1/rootfs/rootfs in the real filesystem [10:30] outside of the container [10:31] then paste me the ls -l /var/lib/lx/p1/rootfs === yofel_ is now known as yofel [10:34] roc@roc-desktop:~$ sudo ln -s . /var/lib/lxc/p1/rootfs/rootfs [10:34] roc@roc-desktop:~$ sudo ls -l /var/lib/lxc/p1/rootfs/ [10:34] total 80 [10:34] drwxr-xr-x 2 root root 4096 Aug 2 11:52 bin [10:34] drwxr-xr-x 2 root root 4096 Jul 4 02:01 boot [10:34] drwxr-xr-x 3 root root 4096 Aug 2 11:51 dev [10:34] drwxr-xr-x 63 root root 4096 Aug 2 11:54 etc [10:34] drwxr-xr-x 3 root root 4096 Aug 2 11:54 home [10:34] drwxr-xr-x 12 root root 4096 Aug 2 11:52 lib [10:34] drwxr-xr-x 2 root root 4096 Aug 2 11:49 lib64 [10:34] drwxr-xr-x 2 root root 4096 Aug 2 11:48 media [10:34] drwxr-xr-x 2 root root 4096 Jul 4 02:01 mnt [10:34] drwxr-xr-x 2 root root 4096 Aug 2 11:48 opt [10:34] drwxr-xr-x 2 root root 4096 Jul 4 02:01 proc [10:35] drwx------ 2 root root 4096 Aug 2 11:48 root [10:35] lrwxrwxrwx 1 root root 1 Aug 13 18:33 rootfs -> . [10:35] drwxr-xr-x 6 root root 4096 Aug 2 11:52 run [10:35] drwxr-xr-x 2 root root 4096 Aug 2 11:53 sbin [10:35] drwxr-xr-x 2 root root 4096 Jun 12 02:36 selinux [10:35] drwxr-xr-x 2 root root 4096 Aug 2 11:48 srv [10:35] drwxr-xr-x 2 root root 4096 Jul 21 10:42 sys [10:35] drwxrwxrwt 2 root root 4096 Aug 2 11:53 tmp [10:35] drwxr-xr-x 10 root root 4096 Aug 2 11:48 usr [10:35] drwxr-xr-x 11 root root 4096 Aug 2 11:48 var [10:35] apw: ^^^ [10:37] looks like it is right. [10:37] * ppisati hates debian packaging... [10:38] apw: but the output still has /rootfs/ prefix [10:45] cooloney, the output matters not, as long as the path whatever it is points to the right place [10:46] so as long as ls -l /proc/$$/fds [10:46] as long as those paths wrong though they are with /rootfs on the front now validly point to the right file [10:46] which they should as the symlink effectivly strips the prefix (makes it point to the same as /) we should be safe [10:48] does the container work right with it like that [10:48] * apw has to nip out for an hour ... will check back in when he gets back [10:51] apw: http://pastebin.ubuntu.com/1144648/ [11:49] * henrix -> lunch [14:13] hi [14:14] can someone enlighten me why umask 0022 is not working as expected? [14:14] see this with linux-{3.5.1,3.6-rc1,next-201213} [14:14] http://nopaste.snit.ch/156629 [14:15] chmod 755 looks like the workaround [14:15] * dileks wants to understand [14:22] looks like I have to umask on ${MOD_ROOT_FILES_DIR} [14:24] formerly I could do umask 0022 $local-git-repo [14:52] * ogasawara back in 20 [15:30] bjf, henrix, herton possible regression with 3.2.0-29. bug 1035590 [15:30] Launchpad bug 1035590 in linux "Kernel update on 10 Aug 2012 affects nm-applet signal strength" [Medium,Confirmed] https://launchpad.net/bugs/1035590 [15:35] ogra_: there? [15:36] jsalisbury, ack. Only direct change to ipw drivers was "net/wireless: ipw2x00: add supported cipher suites to wiphy initialization" [15:37] herton, I also added the network-manager package to the bug for their review [15:39] I wonder if nm is trying to use now nl80211 because of that change, and that doesn't work well with this driver [15:40] yeah, it could be just a gui thing. it's not clear from the bug if the card actually works or not [15:40] (gui or nm) [16:11] hggdh, whats up ? [16:21] ogra_: have you tried today's image for quantal-server armhf+omap4? On one machine (the lab, with the uboot) it loops on 'detecting hardware to find CDROM', on another (no preseed, manual) goes thru the language selection, then nothing [16:21] on the loop: I ^C, see a message about seg fault, and hardware detection restarts [16:22] same thing on the manual, it seems [16:26] well, i just tested desktop an that explodes as well but in ubiquity [16:27] hggdh, bug 1028905 for cdrom-detect ... try if wiping the partition table on the target disk helps [16:27] Launchpad bug 1028905 in cdrom-detect "cdrom-detect in quantal omap4 hangs trying to look for install media on an extended partition" [High,Confirmed] https://launchpad.net/bugs/1028905 [16:27] oh crap, forgot about this one [16:29] i'll try to nail it down this week [16:29] butu first i need a usable desktop install to test a kernel [16:29] *but [16:31] ogra_: yes indeed, it is the extended partition === skaet_ is now known as skaet [16:51] hi apw , why was aufs patched but not enabled? [16:52] Kano, for the same reason we patch it every cycle, because if we don't need to we do not want to enable it [16:52] i would like it to be enabled by default [16:52] where is the dsc for the -10 package? [16:58] Kano, in the archive ? [16:58] yes [16:58] i only see 9 [16:59] ok, found it on https://launchpad.net/ubuntu/+source/linux/ [17:37] hey, do we have quantal kernels for precise yet at all? even in testing stages [17:37] I'm trying debs from quantal directly and not having a huge amount of luck [17:40] elmo: https://launchpad.net/~ubuntu-x-swat/+archive/q-lts-backport [17:41] elmo: there be kernels there [17:41] bjf: thanks, will try that [17:49] bjf: interesting [17:49] bjf: so, the linux-image-3.5.0-9-generic in quantal doesn't have a hpsa.ko despite having it as =m in the /boot/config file [17:49] bjf: but the PPA deb *does* have the hpsa.ko [17:50] and hpsa is what gives me my root file system, so ... [17:57] elmo, so you are golden because you are using the ppa which has the module you need right? [17:57] bjf: sure sure [17:57] bjf: but I suspect if I was using quantal (userspace + kernel); the machine wouldn't boot [17:58] bjf: but I don't have time to (dis)prove that theory right now, so I'll wave my hands and assume you guys have that under control [17:58] elmo, ah, that was your point [17:58] elmo, a bug on that would be nice [17:59] bjf: ok, will file one in a sec [17:59] just got to quickly see if quantal fixes all my OCZ SSD woes === tyhicks` is now known as tyhicks [18:03] elmo: by chance, when you were installing the .debs from quantal directly, did you make sure to install the linux-image-extra deb? [18:03] $ dpkg -c linux-image-extra-3.5.0-9-generic_3.5.0-9.9_amd64.deb | grep "hpsa" [18:03] -rw-r--r-- root/root 97080 2012-08-09 07:16 ./lib/modules/3.5.0-9-generic/kernel/drivers/scsi/hpsa.ko [18:05] elmo: what ocz problems? which drive [18:06] ogasawara: I was, I didn't [18:08] ogasawara: forgive my ignorance, but why are drivers split out into -extra? [18:09] elmo: we slit the -generic package so that we could eliminate maintaining a separate -virtual flavor. so the linux-image is a paired down offering for vm's, linux-image-extra provides everything else, so most will want both installed. [18:09] ogasawara: ah, I see. so PEBKAC, then, no bug needed. thanks for explaining [18:10] 0b:00.0 SCSI storage controller: OCZ Technology Group, Inc. Device 1044 (rev 02) [18:10] Kano: ^-- those; they spass out when more than one is in a given box at the same time [18:10] unfortunately, the OCZ binary driver works in that case, but the mvsas driver does not [18:10] I was hoping quantal's 3.5.xx might fix it, but it doesn't [18:11] elmo: a pci-e ocz? [18:11] Kano: yes [18:11] i only have got ocz via sata [18:11] whats the pci-id? lspci -vnn [18:12] Kano: http://paste.ubuntu.com/1145316/ [18:13] or http://paste.ubuntu.com/1145320/ (as root) [18:14] http://www.spinics.net/lists/linux-scsi/msg54932.html [18:14] hmm thats the part you could write a bug report [18:14] do you use latest firmware? [18:15] mvsas is the driver [18:16] btw. did you look at modinfo mysas [18:16] that driver has got 2 modes [18:24] jsalisbury: you marked bug 1036063 as triaged, do you need more information from me or not? [18:24] Launchpad bug 1036063 in linux "BUG: soft lockup - CPU#4 stuck for 22s! [rm:30630]" [High,Triaged] https://launchpad.net/bugs/1036063 [19:29] apw: i reenabled the aufs module, now it fails on hash check (after renameing 9.9 to 10.10 for as 10.10kanotix1) [19:30] apw: how to fix that [19:30] Kano, hash check? [19:31] apw: http://paste.debian.net/183416/ [19:33] Kano, what does debian/rules printenv say [19:33] * apw suspects you have not pulled a real 10.10 ABI into your package [19:33] no i just renamed 9.9 to 10.10 because it failed [19:34] wget -qO- 'http://kernel.ubuntu.com/git?p=ubuntu/ubuntu-quantal.git;a=patch;h=eddf18835fe46235ed6800ff80b88251a390b84b'|sed s/8.8/9.9/g|patch -Rp1 [19:34] then its saying what it should that the ABI is changed, you either need a real abi or to disable it for your build 'skipabi=true' [19:34] i did that, then moved 9.9 to 10.10 [19:34] abi is not disabled it its failing [19:35] http://paste.debian.net/183417/ [19:35] if you just renames 9.9 to 10.10 thats not going to work [19:35] you need to turn of the abi check or get accurate abi files [19:35] so how to do it? [19:35] skipabi=true [19:36] and then? [19:36] then it won't do the check [19:36] and you need to build it again [19:36] and when i want to update the check? [19:36] update the check? [19:36] well the way you do it [19:37] how would you enable aufs? [19:39] i would use getabis to pull the abi information, but as you are going to then end up with a failed abi check because you need it to be called 10.10xxx1 you need to disable the check [19:39] but when i compile it twice? [19:40] no just add skipabi=true [19:40] and build it just once [19:41] i want to build it with pbuilder [19:41] so add skipabi in the arch.mk files [19:41] and how do you update it? [19:42] update what [19:42] the abi check [19:42] if you mean the abi, if you update it you will have to make your version 11.10 [19:43] and the you will be newer than our 11.11 [19:43] and your versioning will go to hell [19:43] hmm [19:43] cant you just enable aufs by default ;) [19:44] heh === rsalveti` is now known as rsalveti