[09:22] <verwilst> hello, i would like to test a small patch to the kernel that fixes a kernel oops: https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1020207/comments/11 but my kernel doesnt seem to build with make-kpkg :(
[09:22] <ubot2`> Ubuntu bug 1020207 in linux "gfs2 kernel oops when deleting file as first action after mounting" [Medium,Confirmed]
[09:22] <verwilst> fsam7400.c:(.text+0x3b0): undefined reference to `populate_rootfs_wait'
[09:22] <verwilst> maybe somebody here feels like creating a new kernel deb with this patch that i can test? :)
[09:22] <apw> verwilst, i'll have a look in a minutre
[09:23] <verwilst> apw, awesome thanks! just gimme a yell
[09:27] <apw> verwilst, so is this against precise ?
[09:27] <verwilst> yip
[09:27] <verwilst> the 12.04 kernel oops'es as soon as you remove a file on gfs2 :)
[09:27] <verwilst> it's fixed in 3.3-rc6, which this mini patch is a backport of
[09:29] <verwilst> if i can test that this patch works, the redhat guy is willing to give his ACK for -stable inclusion mainline
[09:36] <apw> verwilst, whats in the bug is all mangled, and if you want a specific patch included so you can send in a Tested-by it might make sense to email me the patch so i can apply the exact patch
[09:36] <apw> verwilst, email in pm
[09:38] <apw> verwilst, if there isn't a patch i can make one out of that hunk .... let me know
[09:39] <verwilst> apw, well, i just pasted the thing i made :)
[09:39] <verwilst> i can mail the patch itself
[09:39] <apw> which ever you want
[09:39] <verwilst> it's the one in the comments, right? #11, the 4 lines
[09:39] <apw> i am assuming thats the one you wanted
[09:40] <apw> but it is you who is asking, so i derfer to you :)
[09:42] <verwilst> apw, you have mail :)
[09:43] <verwilst> it's done in fs/gfs2 btw
[09:43] <verwilst> maybe i should have done it from the root
[09:47] <apw> verwilst, i'll cope ... which commit in the upstream kernel is this a backport of?  am just doing the commentary
[09:47] <apw> verwilst, and i am going to take the liberty of adding your signon
[09:47] <apw> signoff
[09:48] <verwilst> apw, http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=718b97bd6b03445be53098e3c8f896aeebc304aa
[09:49] <apw> verwilst, and we don't need the other hunk ?
[09:50] <verwilst> apw, normally no, since out_inodes doesnt exist in 3.2
[09:51] <verwilst> apw, https://www.redhat.com/archives/linux-cluster/2012-August/msg00151.html
[09:51] <apw> ok.  /me shoves it on a buildder
[10:00] <apw> verwilst, ok looking at this if we take a previous fixup patch we can take the patch you want unaltered
[10:00] <apw> verwilst, so that might be a better solution as the missing error case (which your version doesn't have to fix) looks like something we want to fix also
[10:00] <verwilst> the -EROFS?
[10:01] <verwilst> that would be VERY nice apw :)
[10:02] <verwilst> apw, this patch is actually part of a series with some more handy-looking patches: http://git.kernel.org/?p=linux/kernel/git/torvalds/linux.git;a=commitdiff;h=891003abb0db6bfffd61b76ad0ed39bb7c3db8e1
[10:02] <apw> verwilst, ok i'll build a kernel with that as well
[10:03] <verwilst> apw, not sure about the possibility to get those in, just wanted to show you as it might  make gfs2 a better experience on precise :)
[10:03] <apw> hmmm, a tricker proposition if people are not hitting the issue
[10:03] <verwilst> the one i pasted is (for now) the only showstopper for me
[10:04] <apw> from an SRU perspective
[10:04] <verwilst> yeah
[10:04] <verwilst> apw, it was just FYI :)
[10:13] <apw> verwilst, ok here is the first one: http://people.canonical.com/~apw/lp1020207-precise/ with just your one hunk
[10:14] <apw> verwilst, second one with the unaltered patch is building now
[10:30] <apw> verwilst, ok they are both there, the earlier timestamp is your patch, the later timestamp is the two cherry-picks ... test them if you could and let me know
[10:40]  * apw pokes verwilst
[10:48] <verwilst> apw, i was out for f00d :)
[10:49] <verwilst> apw, should i test them both?
[10:49] <ogra_> eat cake then ...
[10:49] <apw> verwilst, i would suggest testing the newer one, and if that works we can suggest that to dave woodhouse
[10:51] <verwilst> alright great
[10:51] <verwilst> so there are more gfs2-related fixes apw?
[10:52] <apw> verwilst, this contains just the foundation patch and the full cherrypick of the patch you were porting part of
[10:52] <apw> we might also suggest separatly that one at least of the other two you mentioned be pushed to stable, but that is a separate transaction me thinks
[10:52] <verwilst> yeah idd
[10:53] <verwilst> the gfs2_grow thing for example
[10:53] <verwilst> oh well, they all look like nice fixes :)
[11:01] <verwilst> apw, with that other patch included, backporting gfs2 fixes should be much more straightforward i guess in the future
[11:19] <ppisati> http://www.cnx-software.com/2012/08/20/codethink-launches-the-baserock-slab-arm-server/
[11:19] <ppisati> ogra_: ^
[11:19] <ppisati> ARMv5? no, i mean, really?
[11:26] <apw> ppisati, ? core ARMv7-A processors
[11:27] <apw> ppisati, i note they all have massive fans on them, not sure thats the idea :)
[11:36] <ogra_> the v5 system is the management 
[11:40] <apw> oh thats dumb, so you have to have a different distro on management, awsome
[11:53] <ppisati> but those could be candidates for the arm builders
[12:08] <verwilst> apw, well, it certainly boots in kvm.. :)
[12:09]  * cking reboots
[12:09] <verwilst> apw, will test a normal server with a cluster now :)
[12:09]  * ppisati -> out for lunch
[12:13]  * henrix -> lunch
[12:28] <verwilst> apw, great success!!
[12:31] <apw> verwilst, good news
[12:31] <apw> verwilst, so i'll reply to that original thread with the two patches together
[12:31] <verwilst> yip
[12:32] <verwilst> apw, so, do you decide what goes into the new precise kernel?
[12:32] <apw> verwilst, not on my own no, but a bug with a fix is something which is SRUable
[12:33] <apw> verwilst, and if it goes into stable then it will hit precise without any action anyhow
[12:33] <verwilst> apw, i would certainly think so :) thanks a lot for the help btw!
[12:33] <apw> verwilst, np
[12:33] <verwilst> apw, will you push it towards stable?
[12:35] <apw> verwilst, i think the appropriate thing is for me to add your tested-by (if thats ok) to these two patches and email them as a set in reply to Steven
[12:35] <verwilst> okido
[12:35] <verwilst> apw, i kinda sent a mail to him as well though, but without the tested-by thing.. feel free to reply and take credit where needed :D
[12:35] <verwilst> in all my excitement.. :D
[12:36] <verwilst> ( linux cluster ML )
[12:36] <apw> verwilst, heh just want him to know you tested it, so i'd add Tested-by: your email in the bottom
[12:37] <verwilst> well, i pointed him to the link you gave me, so if you add it to the patches there... :)
[12:38] <rtg> apw, are you guys working on overlayfs ?
[12:38] <apw> rtg, i am working on it yes, but not related to this discussion above; crashing heap of junk
[12:38] <rtg> apw, ah, bummer
[12:39] <verwilst> apw, i'm going to install your kernel on every node of my cluster
[12:40] <apw> verwilst, that is the *1111* one that you tested right ...
[12:40] <verwilst> Linux vm02-test 3.2.0-31-generic #48~lp1020207v201208221111 SMP Wed Aug 22 10:19:27 UTC 2012 x86_64 x86_64 x86_64 GNU/Linux
[12:40] <apw> rtg, i have an id patch accepted, an NFS fix pending, and another i am investigating
[12:41] <apw> verwilst, ok i have cleaned up that directory then so only the relevant files are in there, and added the tested-by:, if you want to check its the email you don't mind being in the public record, and if not tell me in PM what you want :)
[12:42] <verwilst> it's fine apw :)
[12:43] <verwilst> nothing my dspam can't handle ;)
[12:50] <apw> verwilst, heh good, as i am not subscribed to that list, i'll let you keep an eye and poke me if i need to do something with those patches, i am hoping Steve will pass them on, if not then i should
[12:51] <verwilst> he usually replies pretty quickly
[12:51] <verwilst> precise is really starting to get somewhere imo
[12:51] <verwilst> 2 months ago i was on the brink of switching everything to centos :)
[12:51] <verwilst> because it felt more enterprisy
[12:52] <verwilst> every step of the way to a working cluster was riddled with bugs :)
[12:52] <verwilst> but now i chased enough people to fix my last hurdle, it's started to look very nice :)
[12:52] <verwilst> my faith has been restored
[12:52] <verwilst> ;)
[12:53] <verwilst> well, 1 hurdle left in lvm2, but hey
[12:54] <xnox> verwilst: clvm?
[12:54] <verwilst> yip
[12:55] <verwilst> the no-monitoring patch thing xnox :)
[12:55] <xnox> verwilst: yes. indeed. Do you think we should not have it? TBH I don't know what that monitoring does.
[12:55] <verwilst> it works for me now because i manually removed the patch from the deb and recompiled, but hey, it will be a warm fuzzy feeling when it will be really fixed
[12:56] <verwilst> xnox, it's only used to build udeb's iirc
[12:56] <xnox> verwilst: also in quantal we had clvm init script borked for a little while. Now fixed.
[12:56] <verwilst> xnox, because dmeventd isnt running or something similar i think
[12:56] <verwilst> xnox, it's the same in debian though
[12:56] <verwilst> xnox, ah, is the init script something backportable to precise?
[12:57] <xnox> verwilst: it should be applied the whole time, it's just in the running system you can specify monitor=1 or whatever the option is.
[12:57] <xnox> verwilst: precise has it. quantal dropped it for couple of months. So if you recompiled from quantal, beware ;-)
[12:57] <verwilst> xnox, yeah, but that's the part that's broken :) monitor=xx isnt read anymore
[12:57] <xnox> verwilst: ah =) interesting.
[12:58] <verwilst> xnox, i think i took it from debian
[12:58] <verwilst> the .95
[12:58] <xnox> verwilst: well quantal now has .95 merge with bug-fixes on top of debian. there are a few things different on debian. 
[12:58] <verwilst> xnox, https://bugs.launchpad.net/ubuntu/+source/lvm2/+bug/833368
[12:58] <ubot2`> Ubuntu bug 833368 in resource-agents "clustered lvm commands fail with "activation/monitoring=0 is incompatible with clustered Volume Group" error" [Undecided,In progress]
[12:59] <verwilst> xnox, well, it's currently a test setup, i hope to reinstall them once i go production with all the 'native' precise packages :)
[12:59] <verwilst> except like libvirt 0.9.13 etc ;)
[13:00] <xnox> verwilst: well precise has old lvm2 (missed the feature freeze), so you might have to backport lvm2 from quantal.
[13:00] <xnox> verwilst: but I probably should hunt that patch from .96 and apply it in quantal, possibly precise as well.
[13:00] <verwilst> xnox, is there a .96?
[13:01] <verwilst> ah
[13:01] <verwilst> in the bug you mean
[13:01] <xnox> yeah.
[13:01] <verwilst> bugreport*
[13:01] <xnox> from upstream *gasp*
[13:01] <verwilst> that would be awesome idd, my coding foo surely isnt strong enough :)
[13:01] <verwilst> but for the rest, it's working very well
[13:02] <verwilst> now that the gfs2 thing is solved
[13:02] <verwilst> i lost many evenings on that problem :P
[13:02] <xnox> well, we are here to help ;-)
[13:02] <verwilst> xnox, 2.02.95-4ppa1 is my version
[13:03] <verwilst> 2.02.95-4 simply recompiled
[13:03] <verwilst> what did you 'fix' in the init script?
[13:04] <xnox> verwilst: init stayed the same. But for example the debian version is broke some symbols & the multiarch was done fully. Fixed those.
[13:04] <xnox> also applied the usual ubuntu patches.
[13:04] <xnox> I would recomend you to use $ backportpackage to rebuild quantal's lvm2 for precise into your ppa
[13:04] <verwilst> backportpackage?
[13:05] <verwilst> ubuntu-dev-tools.. xnox, am i missing out on some good stuff here? :)
[13:05] <verwilst> i always do apt-get source, dch -v ... debuild -S -sa; dput
[13:05] <verwilst> :)
[13:06] <xnox> verwilst: well it does that... automated... and it can also upload it into the ppa for you or build locally =)
[13:07] <xnox> verwilst: also it has the handy $ pull-lp-source and $ pull-debian source
[13:07] <xnox> verwilst: it takes these arguments: pkgname [exact-package-version | $distro-series]
[13:07] <xnox> by default grab latest ;-)
[13:08] <verwilst> where has that been all my life! :)
[13:19] <cking> bah, setting up chroots on various boxen takes forever
[13:53] <cking> hrm, seemed to have hacked around my blasted Virgin Media Super Hub TCP timeout issue. stupid device
[13:56] <apw> cking, vhub> what did you do?
[13:56] <cking> apw, http://smackerelofopinion.blogspot.co.uk/2012/08/virgin-media-super-hub-and-ssh-timeouts.html
[13:56] <apw> silly me
[13:57] <cking> apw, it is really a ghastly device
[13:57] <apw> cking, it sounds lame indeed, and this is the V2 device the first ones were utter balls i believe
[13:57] <cking> yep
[13:58] <cking> well, I've not had to reboot it like the previous kit I had, so that's a bonus I suppose
[13:58] <apw> heh swings and roundabouts perhaps
[13:59] <cking> apw, well, it is fast, so mustn't grumble really
[13:59] <apw> cking, its like 10x what i have i recon
[13:59] <ppisati> rsalveti: how was kernel testing?
[14:17] <verwilst> apw, you have another mail
[14:43]  * ogasawara back in 20
[14:55] <rsalveti> ppisati: working fine, built a bunch of packages, used glmark for a while, stable still
[14:55] <rsalveti> ppisati: seems we got a winner
[14:57] <brendand> bjf - some forewarning. this DC move has screwed up our server testing capacity, so the next SRU runs may be sans a bunch of servers
[14:58] <bjf> brendand: ack, thanks for the heads up.
[14:58] <ppisati> rsalveti: cool
[14:58] <ppisati> ogra_: ^
[15:00] <ogra_> :)
[15:00] <ppisati> correct me if i'm mistaken but:
[15:01] <ogra_> ppisati, convince ndec, he didnt belive yours is more stable ;)
[15:01] <ppisati> new kernel + pvr-omap4 = unity3d working, right?
[15:08] <ppisati> ogra_: tsk.../me neither, actually... :)
[16:18] <xnox> do we have any bleeding edge / recent kernels backported for lucid?
[16:31] <bjf> xnox, no
[16:31] <xnox> bjf: ok, thanks.
[16:32] <cking> sconklin, ping
[16:35] <sconklin> cking: yes? I'm replying to your email now . . .
[16:36] <cking> sconklin, ack
[16:59] <cking> sconklin, I'm still unsure what kind of "clamping" tool I'm expected to use on that splitter. it is quite chunky and I've only got weedy alligator clips
[17:00] <sconklin> sorry, perhaps that bit of kit hasn't arrived yet
[17:00] <sconklin> stand by for a link/photo
[17:01] <sconklin> oh, and stand by for a revision to my email - I made a mistake in the chart
[17:01] <sconklin> cking ^^
[17:01] <sconklin> http://www.amazon.com/Fluke-i400-AC-Current-Clamp/dp/B000EA1ETC/ref=sr_1_5?ie=UTF8&qid=1345654831&sr=8-5&keywords=fluke+clamp+current+probe
[17:02] <cking> sconklin, OK - did anyone order that for me?
[17:02] <sconklin> yes, you should receive one. I received my later than the splitter, so I suspect it's on the way
[17:02] <sconklin> s/my/mine/
[17:03] <cking> sconklin, ack, do we have a tracking number? I'm having a bad week for deliveries mysteriously not making to my door
[17:03] <sconklin> cking: ask pete, He handled that during the QA sprint in Lexington, and I wasn't there
[17:05]  * cking notes that clamp looks like it has some powerful grip, ideal for keeping the kids under control ;-)
[17:06] <sconklin> cking: it doesn't actually make contact with the wire under test - it measures the sum of the current passing through the clamp loop.
[17:06] <cking> sconklin, yep, I comprehend that. just looks like a mean grip to it
[17:06] <sconklin> cking: yep, watch yoru fingers
[17:08] <sconklin> cking: revised email sent
[17:08] <cking> sconklin, ta
[17:16]  * ppisati goes for some groceries
[17:23]  * rtg -> lunch
[18:08] <jdstrand> smb`: hey, on bug #1031090, am I supposed to be able to load the kvm_intel module in a precise guest on a *quantal* host?
[18:08] <ubot2`> Launchpad bug 1031090 in linux "kvm_intel not loadable in a quantal guest" [High,In progress] https://launchpad.net/bugs/1031090
[18:09] <rtg> jdstrand, smb is on his way to san diego
[18:09] <rtg> apw, ^^
[18:09] <jdstrand> smb`: I updated to 3.2.0-30.47-generic in the guest and am running 3.5.0-10.10-generic on the host
[18:09] <jdstrand> hmm
[18:10]  * apw looks
[18:10] <apw> jdstrand, that was being fixed indeed, i don't think i have seen final resolution on the right fix from upstream
[18:11] <jdstrand> apw: I see. so my quantal host doesn't have the necessary bits either?
[18:11] <apw> well your precise guest is playing host
[18:12] <apw> jdstrand, so i am not 100% sure if that would tickle the same issue or not
[18:12] <jdstrand> apw: this is where it gets confusing-- so, I am running an i7 quantal host (my laptop). I start a precise openstack vm. in the precise vm, loading kvm_intel fails
[18:13] <jdstrand> $ sudo modprobe kvm_intel
[18:13] <jdstrand> FATAL: Error inserting kvm_intel (/lib/modules/3.2.0-30-generic/kernel/arch/x86/kvm/kvm-intel.ko): Operation not supported
[18:14] <apw> is there anything at the bottom of dmesg from that invocation ?
[18:14] <jdstrand> [  550.189256] kvm: no hardware support
[18:14] <jdstrand> the kvm module is loaded
[18:15] <apw> this fix was slated to fix he opposite combination so i am not sure if he would have tested the other way round
[18:16] <jdstrand> right
[18:16] <apw> so your test isn't going to test the verification requested there
[18:16] <apw> though being a valid combination we'd expect it to do something
[18:16] <jdstrand> it just happens to be that I upgraded in the meantime and am trying to test an openstack CVE on precise :)
[18:17] <jdstrand> apw: the bit I was concerned about was that the linux task was marked "Won't Fix", but this seems very related
[18:17] <jdstrand> apw: I'm happy to file another bug, or to comment in the existing one
[18:18] <apw> linux should be invalid as in theory invalid as that bug is specific to the other combination
[18:18] <apw> i think we should file a new one, as its as likely to be unrelated as not
[18:18] <apw> in cause if not symptoms, but do put the new bug number in the old bug 
[18:19] <jdstrand> apw: interesting. If I do: sudo rmmod kvm_intel ; sudo modprobe kvm_intel nested=1 I get this in dmesg:
[18:19] <jdstrand> [  972.774373] kvm: VM_EXIT_LOAD_IA32_PERF_GLOBAL_CTRL does not work properly. Using workaround
[18:19] <apw> jdstrand, you were loading it without nested=1 ?  isn't that required ?
[18:20] <apw> for your use combination?
[18:21] <jdstrand> apw: I was trusting /etc/init/qemu-kvm.conf to dtrt since /etc/default/qemu-kvm has KVM_NESTED=" nested=1"
[18:21] <jdstrand> (the default)
[18:21] <jdstrand> I just wanted to be *sure* nested=1 was set
[18:21] <apw> jdstrand, hmmm ... so i guess can we test nested=1 and see if it makes things work for starters
[18:21] <apw> as yeah indeed what you said should be true, but perhaps is not
[18:22] <jdstrand> ha!
[18:22] <jdstrand> that was it
[18:22] <jdstrand> the upstart job isn't working right
[18:22] <jdstrand> apw: ok, thanks
[18:23] <apw> jdstrand, phew :)
[18:24] <jdstrand> indeed
[18:26] <apw> sconklin, bjf, fyi i hacked on the cve-autotriager to wack armadaxp appropriatly when released, which should clean out the cve matrix some
[18:29]  * henrix -> EOD
[18:29] <bjf> apw, very nice, thanks
[18:31]  * bjf -> lunch
[18:50]  * cking --> EOD
[18:56] <rtg> sforshee, can you subscribe me to bug #1040215
[18:59] <sforshee> rtg, done
[19:00] <rtg> sforshee, thanks