[07:20] moin === fmasi_afk is now known as fmasi === fmasi is now known as fmasi_afk === fmasi_afk is now known as fmasi === smb` is now known as smb === fmasi is now known as fmasi_afk === fmasi_afk is now known as fmasi [08:42] apw: nvidia and fglrx are ready for 3.11 now (rtg said that maybe you could upload 3.11 since he's travelling) === fmasi is now known as fmasi_afk === fmasi_afk is now known as fmasi [08:50] ppisati, Goood morning. [09:46] apw: not sure if you noticed bug #1208401, impacting saucy lvm installations. probably triggered by dependency issues? [09:46] Launchpad bug 1208401 in lvm2 (Ubuntu) "saucy server lvm installations fail to boot " [Undecided,New] https://launchpad.net/bugs/1208401 [10:20] psivaa, will have a poke thanks [10:25] jodh, yo ... this nested thing ... i just spun one up, how can i tell if it is broke, as it seems to be running, and making my machine fan itself ... [10:25] jodh, and indeed it has now 'completed' and exit 0 at the end [10:26] jodh, this is with the 3.11 kernel which is about to drop into the archive... [10:26] jodh, would it have broken if your issue occured, or is it more subtle [10:28] apw: the problem occurs when you run a nested kvm instance from within a kvm instance. The host is using kvm_intel with nested=Y, but both guests seem to have nested=N. [10:28] apw: a fresh kernel panic for you from my 32-bit host system. The panic is from the "2nd-level" guest: http://paste.ubuntu.com/5958383/ [10:28] jodh, so i don't see any panic and the test command listed completes exit 0 is that correct ? [10:29] I am really not sure that nested goes more than one level... [10:29] apw: can you confirm what test you are running ? [10:29] $ sudo modprobe kvm_intel nested=1 [10:29] $ bzr branch lp:auto-package-testing [10:29] $ cd auto-package-testing [10:29] $ bin/prepare-testbed -r saucy -d i386 [10:29] jodh, the test set in the bug [10:29] that 4th command runs to completion without error [10:31] ok - that bug is rather confused I'm afraid as I was seeing crashes locally and also strange behaviour in canonistack. I think I'm also getting panics under canonistack but can't access the 2nd-level vm as it's spinning at 200% cpu. [10:31] to see the panic, after you're run prepare-testbed, you need to run: [10:32] bin/run-adt-test -r saucy -a i386 -l -k -b lp:~jamesodhunt/ubuntu/saucy/upstart/dep-8 upstart [10:32] then, when that vm fires up, "cd upstart && adt-run --debug --tmp-dir=/tmp/logs --log-file=$HOME/adt-run.log --unbuilt-tree . --- adt-virt-null --debug" [10:33] I missed a "mkdir /tmp/logs" above too :) [10:33] jodh, hmmm, ok this is going to be hard to know i have done it right ;) [10:33] a question though is whether *all 3* systems (host+2 guests) should have "kvm_intel nested=1" loaded? [10:34] jodh, but i would say if you can repro it, then one might suggest using 3.11 to confirm it is there as well as some kvm issues seemed to be solved by the update [10:34] apw: it takes a while to run adt-run locally I'm afraid, hence my attempt at migrating to canonistack, but that didn't go so well either :) [10:34] jodh, and is this then something new which has never worked ? [10:34] is kvm nesting something we even expect to next infinitly? [10:34] apw: I think I might be trail-blazing here yes [10:35] it may not even be somting you can expect to work [10:35] this is the first time I've tried all this so it has never worked for me atleast :) [10:36] Support for it at least has been around for a while now [10:36] for reference: http://paste.ubuntu.com/5958403/, vm.cfg = http://paste.ubuntu.com/5958409/ [10:36] (so one could expect it to work) [10:37] ... and /proc/cpuinfo on my host: http://paste.ubuntu.com/5958342/ [10:43] jodh, so this second command 'cd upstart &&' etc am i running that _in the byobu session that the previous command runs ? [10:43] as i don't have an upstart in here [10:43] ah - sorry, you probably have a "dep-8" directory right? [10:44] if so, "cd dep8 && adt-run ..." [10:44] dep-8 even :) [10:44] ok that is doing something [10:45] jodh, no that all blows up with dependancy issues [10:45] or is this a sudo job [10:45] dependency issues? [10:45] -> Considering build-dep build-essential [10:45] -> Trying build-essential [10:45] -> Installing build-essential [10:46] E: Could not open lock file /var/lib/dpkg/lock - open (13: Permission denied) [10:46] E: Unable to lock the administration directory (/var/lib/dpkg/), are you root? [10:46] -> Trying to fix apt error [10:46] bear in mind if you send me command lines i am runing them as they are not apply other intelligence [10:46] as clearly your commands are what you do [10:47] * apw tries it as root [10:47] on sec.. [10:47] too late [10:47] gah [10:47] I was going so say: http://paste.ubuntu.com/5958446/ [10:48] which isn't quite the same now is it ... sigh, users [10:48] so anyhow it is runnig something [10:48] what am i looking for for the symptoms and how do i see it [10:50] apw: Is 3.11 meta happening today? [10:50] infinity, i need to confirm the graphics stuff is built, but it should [10:51] apw: Kay, just curious if we need to do d-i things. Shame you're not in the office today, we could have tried to clean all that up. [10:51] jodh, So on a 64bit guest at least, nested=1 is passed on [10:51] once it's built upstart, it will eventually run the "run-tests" command. You'll see a ton of cryptic shell code but the last line will be running that command with stdout redirected to a file [10:51] apw: But I didn't finish cleaning up the seeds yet anyway. [10:51] infinity, yeah ... [10:52] just tail -f that file until you see "Timeout, server localhost not responding". At that point the kernel will have panicked. [10:56] jodh, didn't occur for me, ran to completion [10:57] I can try the 3.11 kernel if you point me to where I can grab it, but it would be good to know exactly how kvm_intel's nested= param should be set for all 3 systems. [10:57] VirtSubproc>: debug: executing quit [10:57] can I see the *stdout and *stderr logs in /tmp/logs/? [10:57] smb, ^^ [10:57] jodh, Mine is at least Y in the first guest [10:57] jodh, but i would assum you want =1 in each? [10:58] apw: ok, but could that account for the panic? [10:58] to say you can nest further, it is only a true value ? [10:58] jodh, i wpuld expect if you can nest you can nest infinitly, but perhaps i don't understnad nesting as well as smb [10:58] apw, So the nested=1 as modules option says it should be used if possible [10:59] apw, If the parm is actually true/y/1 in sysfs it is using it [10:59] And my guest (though 64bit) seems to be ok [10:59] apw, Is your first level 32 or 64bit? [10:59] smb, 64 [11:00] who would run 32, ... that limits one from having 64 bit guests goesn [11:00] odesn't it ? [11:00] apw, So following the other instructions, I see higher cpu usage and some messages about timeout [11:01] jodh, the 3.11 kernel is sitting in proposed, so you can get it from the pool or from https://launchpad.net/ubuntu/+source/linux/3.11.0-0.2 [11:04] apw, Ok, probably did not mean much.... apparently it says "installation sucessful" now [11:04] smb, so my first level kvm doesn't seem to have kvm nesting available / [11:10] jodh, Btw, did you ever reboot your 1rst level guest after setting up the packages for kvm et.all [11:12] Cause the modprobe kvm-intel after logging out and logging in may be useless as the modules have been loaded already before (without nested=1) [11:23] smb: no, I'm not rebooting the 1st level VM. That is provided by the autopkgtest QA environment. In fact, a test can't reboot it as its run with "qemu -no-reboot ...". [11:24] jodh, So I would try (when no 2nd level guest runs) to rmmod kvm_intel and modprobe it again [11:24] and then check nested [11:35] smb, (and jodh) just confirmed here that just installing KVM inside does not help, as kvm-intel is loaded on boot of the VM and not with kvm options [11:36] so one would need to rmmod and mobprobe it to get nested [11:36] and confirm that sudo rmmod kvm-intel; sudo modprobe kvm-intel [11:47] apw: ack. I'll try that, then do the same with 3.11 if it doesn't help. [11:47] * ppisati feels a bit tired... [11:50] Seems for me and a 64bit host and 1rst level guest the prepare-testbed thing seemed to have completed even with nested turned off in the guest [11:55] ppisati, are you in london still? [11:56] cking: nope, back home [11:56] ppisati, wow, that was quick [11:56] cking: just 2 hours flight [11:57] ppisati, and a few hours either end travelling to/from airports + waiting for plane ... [11:57] Wasn't that acutally yesterday [11:57] yeah [11:58] cking, I am flying today (fwiw) [11:58] :) [12:00] * cking got back home by 15:30.. a travel win at last [12:12] * smb leaves [12:37] dude, what's the smart way to filter LP bugs email? [12:38] * ppisati recalls there was a way... [12:41] filtering on the subj "^[Bug" is ok, but can definitely be better using one of those X-fields [12:44] ppisati, https://wiki.ubuntu.com/Bugs/HowToFilter [12:45] ogra_: eh, but those are procmail rules, how do we do that in the brand new gmail world? [12:45] no idea, i dont touch gmail [12:46] life sucks, and then you gmail [12:46] * ogra_ has his own mail server [12:46] ogra_: poser [12:46] i route some mail through gmail for its good spam filter though :) [12:46] ppisati: afaik, gmail won't let you filter on '[' in the subject either [12:46] but after all thats just a loop that points back to my server [12:47] josepht: that's hat i'm doing right now [12:48] *what [12:51] ppisati: https://support.google.com/mail/answer/6593?hl=en [12:55] * ppisati shrugs [12:55] The following filters are applied to all incoming mail: Matches: subject:(^[Bug) [12:55] Do this: Skip Inbox, Apply label "LP-bugs" [12:55] from my gmail settings [13:01] ppisati: for me; subject:Bug, subject:(^[Bug), and subject:(Bug) all return the same results and some are not [Bug xxx] messages [13:33] psivaa, that lvm2 issue is very odd because lvm2 directly dpeends on the seemingly missing package so it really should be installed. can we get the installer and apt logs out of a broken instance ? [13:35] apw: installer logs are attached. i'll try to get the apt logs [13:35] https://launchpadlibrarian.net/146822202/utah-12463-install.log [13:36] psivaa, thanks ... it is a server install, so it could be a udeb issue [13:52] psivaa, i am slightly confused because the package in question is shown as installed as far as i can see: [13:52] Aug 3 08:30:13 in-target: libdevmapper1.02.1 is already the newest version. [13:53] apw, libdevmapper-event is not installed though [13:54] psivaa, point, perhaps it has fallen out of the udeb ... will check [13:56] apw: on a good run this is what we see when the above packages gets installed: [13:56] http://pastebin.ubuntu.com/5958992/ [14:00] psivaa, hmm indeed, so it is a higher level issue somehow, we are not even trying to install lvm2 in this way [14:09] psivaa, do we know when this started failing ? [14:10] apw: images of 20130803 onwords [14:11] psivaa, ta [14:28] psivaa, ok i think xnox has a handle on that bug, and will update it and hopefully fix :) [14:29] apw: thanks :) === jhenke_ is now known as jhenke [17:26] hi all [17:28] buenos dias a todos [20:04] what exactly is 'something.isra' in context of ftrace? === fmasi is now known as fmasi_afk === rsalveti_ is now known as rsalveti