>On Tue, 15 May 2012, Josh Durgin wrote: > > On 05/15/2012 07:01 AM, Martin Wilderoth wrote: > > > Hello, > > > > > > I have a xenhost using rbd device for the guests. In the guest I have a > > > mounted ceph file system. > > > From time to time I get the guest hanging and I have the following error in > > > my logfiles on the guest. > > > > > > Maybe I should not use both rbd and ceph ? > > > > While it's not a well tested configuration, I don't see any reason this > > wouldn't work. Sage, Alex, are there any shared resources in libceph > > that would cause problems with this? There shouldn't be any problems with running rbd + ceph together. > > > May 15 14:13:18 lintx2 kernel: [ 3560.225095] Modules linked in: cryptd > > > aes_x86_64 aes_generic cbc ceph libceph crc32c libcrc32c evdev snd_pcm > > > snd_timer snd soundcore snd_page_alloc pcspkr ext3 jbd mbcache xen_netfront > > > xen_blkfront > > > May 15 14:13:18 lintx2 kernel: [ 3560.225140] Pid: 18, comm: kworker/0:1 > > > Tainted: G W 3.2.0-0.bpo.2-amd64 #1 > > > May 15 14:13:18 lintx2 kernel: [ 3560.225148] Call Trace: > > > May 15 14:13:18 lintx2 kernel: [ 3560.225155] [<ffffffff810497b4>] ? > > > warn_slowpath_common+0x78/0x8c > > > May 15 14:13:18 lintx2 kernel: [ 3560.225167] [<ffffffffa00db647>] ? > > > ceph_add_cap+0x38e/0x49e [ceph] > > > May 15 14:13:18 lintx2 kernel: [ 3560.225178] [<ffffffffa00d220a>] ? > > > fill_inode+0x4eb/0x602 [ceph] > > > May 15 14:13:18 lintx2 kernel: [ 3560.225186] [<ffffffff811157ad>] ? > > > __d_instantiate+0x8b/0xda > > > May 15 14:13:18 lintx2 kernel: [ 3560.225197] [<ffffffffa00d317d>] ? > > > ceph_readdir_prepopulate+0x2de/0x375 [ceph] > > > May 15 14:13:18 lintx2 kernel: [ 3560.225209] [<ffffffffa00e2d3f>] ? > > > dispatch+0xa35/0xef2 [ceph] > > > May 15 14:13:18 lintx2 kernel: [ 3560.225220] [<ffffffffa00ae841>] ? > > > ceph_tcp_recvmsg+0x43/0x4f [libceph] > > > May 15 14:13:18 lintx2 kernel: [ 3560.225231] [<ffffffffa00b0821>] ? > > > con_work+0x1070/0x13b8 [libceph] > > > May 15 14:13:18 lintx2 kernel: [ 3560.225240] [<ffffffff81006f7f>] ? > > > xen_restore_fl_direct_reloc+0x4/0x4 > > > May 15 14:13:18 lintx2 kernel: [ 3560.225248] [<ffffffff81044549>] ? > > > update_curr+0xbc/0x160 > > > May 15 14:13:18 lintx2 kernel: [ 3560.225259] [<ffffffffa00af7b1>] ? > > > try_write+0xbe1/0xbe1 [libceph] > > > May 15 14:13:18 lintx2 kernel: [ 3560.225268] [<ffffffff8105f897>] ? > > > process_one_work+0x1cc/0x2ea > > > May 15 14:13:18 lintx2 kernel: [ 3560.225277] [<ffffffff8105fae2>] ? > > > worker_thread+0x12d/0x247 > > > May 15 14:13:18 lintx2 kernel: [ 3560.225285] [<ffffffff8105f9b5>] ? > > > process_one_work+0x2ea/0x2ea > > > May 15 14:13:18 lintx2 kernel: [ 3560.225294] [<ffffffff810632ed>] ? > > > kthread+0x7a/0x82 > > > May 15 14:13:18 lintx2 kernel: [ 3560.225302] [<ffffffff8136b974>] ? > > > kernel_thread_helper+0x4/0x10 > > > May 15 14:13:18 lintx2 kernel: [ 3560.225311] [<ffffffff81369a33>] ? > > > int_ret_from_sys_call+0x7/0x1b > > > May 15 14:13:18 lintx2 kernel: [ 3560.225319] [<ffffffff8136453c>] ? > > > retint_restore_args+0x5/0x6 > > > May 15 14:13:18 lintx2 kernel: [ 3560.225328] [<ffffffff8136b970>] ? > > > gs_change+0x13/0x13 > > > May 15 14:13:18 lintx2 kernel: [ 3560.225335] ---[ end trace > > > 111652db8892cd8b ]--- > > > > The only warning in ceph_add_cap is when it can't lookup the snap > > realm. I'm not sure if this has any real consequences. Sage? > > Not really. It is a bug, but you're seeing the _guest_ hang, not the fs, > right? I suspect there is something else going on, and this is a red > herring. It's the guest that hangs or is looping I get some CPU#0 stuck for ??s on the console all the time. The only solution is to reset the guest. I can remount the ceph. So the filesystem is not hanging. Martin > > FWIW, I've seen two RBD kernel hangs in the last few days in our QA (under > xfstests workload). We're still looking into that. > -- To unsubscribe from this list: send the line "unsubscribe ceph-devel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
