XFS+DRBD hangs
We will try again with most recent Linus HEAD, but with a build of Linus e9208a4eec8acbde7ede6516c39dea05f3b700dc (3.1-rc4), we are seeing awesome VM performance (low memory fragmentation, good caching), followed quickly by a hard lockup. APEI and pstore stuff just doesn't seem to like this PowerEdge 2950 (ERST: Failed to get Error Log Address Range.), so this was captured over an old-fashioned serial cable:
[ 93.004364] block drbd0: Resync done (total 52 sec; paused 0 sec; 3884 K/sec) [ 93.008004] block drbd0: updated UUIDs D1D6A96AAE5E394E:0000000000000000:264787A8207F2376:7E86E22A1F92B7FB [ 93.008004] block drbd0: conn( SyncTarget -> Connected ) disk( Inconsistent -> UpToDate ) [ 93.030394] block drbd0: helper command: /sbin/drbdadm after-resync-target minor-0 [ 93.044631] block drbd0: helper command: /sbin/drbdadm after-resync-target minor-0 exit code 0 (0x0) [ 93.102981] block drbd0: bitmap WRITE of 3382 pages took 10 jiffies [ 93.109462] block drbd0: 0 KB (0 bits) marked out-of-sync by on disk bit-map. [ 1114.600102] block drbd0: peer( Primary -> Secondary ) [ 1114.904031] block drbd0: role( Secondary -> Primary ) [ 1115.284503] XFS (drbd0): Mounting Filesystem [ 1115.928766] XFS (drbd0): Ending clean mount [ 1116.324036] NFSD: Using /var/lib/nfs/v4recovery as the NFSv4 state recovery directory [ 1116.336050] NFSD: starting 90-second grace period [ 1532.427149] Kernel panic - not syncing: Watchdog detected hard LOCKUP on cpu 1 [ 1532.427149] Pid: 3590, comm: lmtp Not tainted 3.1.0-rc4-hw+ #28 [ 1532.427149] Call Trace: [ 1532.427149] <NMI> [<ffffffff816aa954>] panic+0xba/0x1f5 [ 1532.427149] [<ffffffff8108f6d1>] watchdog_overflow_callback+0xb1/0xc0 [ 1532.427149] [<ffffffff810bb442>] __perf_event_overflow+0xa2/0x1f0 [ 1532.427149] [<ffffffff810b4981>] ? perf_event_update_userpage+0x11/0xc0 [ 1532.427149] [<ffffffff810bb994>] perf_event_overflow+0x14/0x20 [ 1532.427149] [<ffffffff81014bcb>] intel_pmu_handle_irq+0x21b/0x460 [ 1532.427149] [<ffffffff816af286>] perf_event_nmi_handler+0x36/0xb0 [ 1532.427149] [<ffffffff816b132f>] notifier_call_chain+0x3f/0x80 [ 1532.427149] [<ffffffff816b1395>] atomic_notifier_call_chain+0x15/0x20 [ 1532.427149] [<ffffffff816b13ce>] notify_die+0x2e/0x30 [ 1532.427149] [<ffffffff816ae9e2>] do_nmi+0xa2/0x250 [ 1532.427149] [<ffffffff816ae3ca>] nmi+0x1a/0x20 [ 1532.427149] [<ffffffff8103d512>] ? try_to_wake_up+0xc2/0x270 [ 1532.427149] <<EOE>> <IRQ> [<ffffffff8103d6cd>] default_wake_function+0xd/0x10 [ 1532.427149] [<ffffffff81067041>] autoremove_wake_function+0x11/0x40 [ 1532.427149] [<ffffffff8103075a>] __wake_up_common+0x5a/0x90 [ 1532.427149] [<ffffffff81037c43>] __wake_up+0x43/0x70 [ 1532.427149] [<ffffffffa0060579>] drbd_al_complete_io+0x99/0xd0 [drbd] [ 1532.427149] [<ffffffffa005b9b5>] _req_may_be_done+0x325/0x560 [drbd] [ 1532.427149] [<ffffffff810fabe1>] ? kmem_cache_free+0xe1/0xf0 [ 1532.427149] [<ffffffffa005bc0d>] _req_may_be_done_not_susp+0x1d/0x30 [drbd] [ 1532.427149] [<ffffffffa005c0cb>] __req_mod+0x4ab/0xee0 [drbd] [ 1532.427149] [<ffffffffa004f22e>] drbd_endio_pri+0xae/0x120 [drbd] [ 1532.427149] [<ffffffff81130b6c>] bio_endio+0x1c/0x40 [ 1532.427149] [<ffffffff813737d3>] req_bio_endio+0x83/0xb0 [ 1532.427149] [<ffffffff81373900>] blk_update_request+0x100/0x470 [ 1532.427149] [<ffffffff81373c92>] blk_update_bidi_request+0x22/0x90 [ 1532.427149] [<ffffffff81374e5a>] blk_end_bidi_request+0x2a/0x80 [ 1532.427149] [<ffffffff81374efb>] blk_end_request+0xb/0x10 [ 1532.427149] [<ffffffff8146744d>] scsi_io_completion+0x11d/0x620 [ 1532.427149] [<ffffffff8145f2ff>] scsi_finish_command+0xbf/0x120 [ 1532.427149] [<ffffffff81467a9e>] scsi_softirq_done+0x13e/0x160 [ 1532.427149] [<ffffffff8137ab7d>] blk_done_softirq+0x6d/0x80 [ 1532.427149] [<ffffffff8104d0f0>] __do_softirq+0xd0/0x1c0 [ 1532.427149] [<ffffffff816b726c>] call_softirq+0x1c/0x30 [ 1532.427149] [<ffffffff81004035>] do_softirq+0x55/0x90 [ 1532.427149] [<ffffffff8104ce2d>] irq_exit+0xad/0xe0 [ 1532.427149] [<ffffffff81003824>] do_IRQ+0x64/0xe0 [ 1532.427149] [<ffffffff816adf6b>] common_interrupt+0x6b/0x6b [ 1532.427149] <EOI> [ 1545.171011] Kernel panic - not syncing: Watchdog detected hard LOCKUP on cpu 0 [ 1545.171011] Pid: 20283, comm: xfsbufd/drbd0 Not tainted 3.1.0-rc4-hw+ #28 [ 1545.171011] Call Trace: [ 1545.171011] <NMI> [<ffffffff816aa954>] panic+0xba/0x1f5 [ 1545.171011] [<ffffffff8108f6d1>] watchdog_overflow_callback+0xb1/0xc0 [ 1545.171011] [<ffffffff810bb442>] __perf_event_overflow+0xa2/0x1f0 [ 1545.171011] [<ffffffff810b4981>] ? perf_event_update_userpage+0x11/0xc0 [ 1545.171011] [<ffffffff810bb994>] perf_event_overflow+0x14/0x20 [ 1545.171011] [<ffffffff81014bcb>] intel_pmu_handle_irq+0x21b/0x460 [ 1545.171011] [<ffffffff816af286>] perf_event_nmi_handler+0x36/0xb0 [ 1545.171011] [<ffffffff816b132f>] notifier_call_chain+0x3f/0x80 [ 1545.171011] [<ffffffff816b1395>] atomic_notifier_call_chain+0x15/0x20 [ 1545.171011] [<ffffffff816b13ce>] notify_die+0x2e/0x30 [ 1545.171011] [<ffffffff816ae9e2>] do_nmi+0xa2/0x250 [ 1545.171011] [<ffffffff816ae3ca>] nmi+0x1a/0x20 [ 1545.171011] [<ffffffff816adb07>] ? _raw_spin_lock_irqsave+0x17/0x20 [ 1545.171011] <<EOE>> <IRQ> [<ffffffffa004f21c>] drbd_endio_pri+0x9c/0x120 [drbd] [ 1545.171011] [<ffffffff814302d2>] ? put_device+0x12/0x20 [ 1545.171011] [<ffffffff81130b6c>] bio_endio+0x1c/0x40 [ 1545.171011] [<ffffffff813737d3>] req_bio_endio+0x83/0xb0 [ 1545.171011] [<ffffffff81373900>] blk_update_request+0x100/0x470 [ 1545.171011] [<ffffffff81373c92>] blk_update_bidi_request+0x22/0x90 [ 1545.171011] [<ffffffff81374e5a>] blk_end_bidi_request+0x2a/0x80 [ 1545.171011] [<ffffffff81374efb>] blk_end_request+0xb/0x10 [ 1545.171011] [<ffffffff8146744d>] scsi_io_completion+0x11d/0x620 [ 1545.171011] [<ffffffff8145f2ff>] scsi_finish_command+0xbf/0x120 [ 1545.171011] [<ffffffff81467a9e>] scsi_softirq_done+0x13e/0x160 [ 1545.171011] [<ffffffff8137ab7d>] blk_done_softirq+0x6d/0x80 [ 1545.171011] [<ffffffff8104d0f0>] __do_softirq+0xd0/0x1c0 [ 1545.171011] [<ffffffff816b726c>] call_softirq+0x1c/0x30 [ 1545.171011] [<ffffffff81004035>] do_softirq+0x55/0x90 [ 1545.171011] [<ffffffff8104ce2d>] irq_exit+0xad/0xe0 [ 1545.171011] [<ffffffff8101cd0f>] smp_call_function_single_interrupt+0x2f/0x40 [ 1545.171011] [<ffffffff816b6d4b>] call_function_single_interrupt+0x6b/0x70 [ 1545.171011] <EOI> [<ffffffff8146635c>] ? scsi_request_fn+0xec/0x510 [ 1545.171011] [<ffffffff8136fd25>] __blk_run_queue+0x15/0x20 [ 1545.171011] [<ffffffff813876ec>] cfq_insert_request+0x49c/0x5c0 [ 1545.171011] [<ffffffff8136f658>] __elv_add_request+0x198/0x2a0 [ 1545.171011] [<ffffffff81376410>] blk_flush_plug_list+0x1d0/0x210 [ 1545.171011] [<ffffffff816ab70a>] schedule+0x62a/0xad0 [ 1545.171011] [<ffffffff810fb26e>] ? kmem_cache_alloc+0xee/0x100 [ 1545.171011] [<ffffffffa0042199>] ? lc_get+0x49/0x250 [lru_cache] [ 1545.171011] [<ffffffffa0061815>] drbd_al_begin_io+0x1c5/0x200 [drbd] [ 1545.171011] [<ffffffff81132784>] ? bio_alloc_bioset+0x54/0xe0 [ 1545.171011] [<ffffffff81067030>] ? wake_up_bit+0x40/0x40 [ 1545.171011] [<ffffffff81131591>] ? __bio_clone+0x21/0x60 [ 1545.171011] [<ffffffffa005d825>] drbd_make_request_common+0xd25/0x1650 [drbd] [ 1545.171011] [<ffffffff8136e916>] ? elv_set_request+0x16/0x40 [ 1545.171011] [<ffffffffa005e48a>] drbd_make_request+0x33a/0x1150 [drbd] [ 1545.171011] [<ffffffff8137a175>] ? blk_recount_segments+0x25/0x40 [ 1545.171011] [<ffffffff81131b81>] ? bio_phys_segments+0x21/0x30 [ 1545.171011] [<ffffffff8137410e>] ? drive_stat_acct+0x11e/0x180 [ 1545.171011] [<ffffffff8137665d>] ? __make_request+0x1bd/0x330 [ 1545.171011] [<ffffffff81374713>] generic_make_request+0x3f3/0x540 [ 1545.171011] [<ffffffff8138c327>] ? kobject_put+0x27/0x60 [ 1545.171011] [<ffffffff813748d6>] submit_bio+0x76/0xf0 [ 1545.171011] [<ffffffff81278a57>] _xfs_buf_ioapply+0xb7/0x1f0 [ 1545.171011] [<ffffffff8127a3a5>] ? xfs_bdstrat_cb+0x85/0xb0 [ 1545.171011] [<ffffffff8127a251>] xfs_buf_iorequest+0x71/0xc0 [ 1545.171011] [<ffffffff8127a3a5>] xfs_bdstrat_cb+0x85/0xb0 [ 1545.171011] [<ffffffff8127a4c6>] xfsbufd+0xf6/0x130 [ 1545.171011] [<ffffffff8127a3d0>] ? xfs_bdstrat_cb+0xb0/0xb0 [ 1545.171011] [<ffffffff81066b16>] kthread+0x96/0xb0 [ 1545.171011] [<ffffffff816b7174>] kernel_thread_helper+0x4/0x10 [ 1545.171011] [<ffffffff81066a80>] ? kthread_worker_fn+0x130/0x130 [ 1545.171011] [<ffffffff816b7170>] ? gs_change+0xb/0xb
We upgraded from 2.6.36 which seemed to have a page leak (file pages left on the LRU) and so would eventually perform very poorly. 2.6.37 and 2.6.38 seemed to have some unix socket issue that caused heartbeat to wedge. Shall we enable lock debugging or something here?
More bits at http://0x.ca/sim/ref/3.1-rc4/.
Simon-
_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs
Lars Ellenberg
maintainerQuoted 130 lines · Simon Kirby
We will try again with most recent Linus HEAD, but with a build of Linus
e9208a4eec8acbde7ede6516c39dea05f3b700dc(3.1-rc4), we are seeing awesome VM performance (low memory fragmentation, good caching), followed quickly by a hard lockup. APEI and pstore stuff just doesn't seem to like this PowerEdge 2950 (ERST: Failed to get Error Log Address Range.), so this was captured over an old-fashioned serial cable:[ 93.004364] block drbd0: Resync done (total 52 sec; paused 0 sec; 3884 K/sec) [ 93.008004] block drbd0: updated UUIDs D1D6A96AAE5E394E:0000000000000000:264787A8207F2376:7E86E22A1F92B7FB [ 93.008004] block drbd0: conn( SyncTarget -> Connected ) disk( Inconsistent -> UpToDate ) [ 93.030394] block drbd0: helper command: /sbin/drbdadm after-resync-target minor-0 [ 93.044631] block drbd0: helper command: /sbin/drbdadm after-resync-target minor-0 exit code 0 (0x0) [ 93.102981] block drbd0: bitmap WRITE of 3382 pages took 10 jiffies [ 93.109462] block drbd0: 0 KB (0 bits) marked out-of-sync by on disk bit-map. [ 1114.600102] block drbd0: peer( Primary -> Secondary ) [ 1114.904031] block drbd0: role( Secondary -> Primary ) [ 1115.284503] XFS (drbd0): Mounting Filesystem [ 1115.928766] XFS (drbd0): Ending clean mount [ 1116.324036] NFSD: Using /var/lib/nfs/v4recovery as the NFSv4 state recovery directory [ 1116.336050] NFSD: starting 90-second grace period [ 1532.427149] Kernel panic - not syncing: Watchdog detected hard LOCKUP on cpu 1 [ 1532.427149] Pid: 3590, comm: lmtp Not tainted 3.1.0-rc4-hw+ #28 [ 1532.427149] Call Trace: [ 1532.427149] <NMI> [<ffffffff816aa954>] panic+0xba/0x1f5 [ 1532.427149] [<ffffffff8108f6d1>] watchdog_overflow_callback+0xb1/0xc0 [ 1532.427149] [<ffffffff810bb442>] __perf_event_overflow+0xa2/0x1f0 [ 1532.427149] [<ffffffff810b4981>] ? perf_event_update_userpage+0x11/0xc0 [ 1532.427149] [<ffffffff810bb994>] perf_event_overflow+0x14/0x20 [ 1532.427149] [<ffffffff81014bcb>] intel_pmu_handle_irq+0x21b/0x460 [ 1532.427149] [<ffffffff816af286>] perf_event_nmi_handler+0x36/0xb0 [ 1532.427149] [<ffffffff816b132f>] notifier_call_chain+0x3f/0x80 [ 1532.427149] [<ffffffff816b1395>] atomic_notifier_call_chain+0x15/0x20 [ 1532.427149] [<ffffffff816b13ce>] notify_die+0x2e/0x30 [ 1532.427149] [<ffffffff816ae9e2>] do_nmi+0xa2/0x250 [ 1532.427149] [<ffffffff816ae3ca>] nmi+0x1a/0x20 [ 1532.427149] [<ffffffff8103d512>] ? try_to_wake_up+0xc2/0x270 [ 1532.427149] <<EOE>> <IRQ> [<ffffffff8103d6cd>] default_wake_function+0xd/0x10 [ 1532.427149] [<ffffffff81067041>] autoremove_wake_function+0x11/0x40 [ 1532.427149] [<ffffffff8103075a>] __wake_up_common+0x5a/0x90 [ 1532.427149] [<ffffffff81037c43>] __wake_up+0x43/0x70 [ 1532.427149] [<ffffffffa0060579>] drbd_al_complete_io+0x99/0xd0 [drbd] [ 1532.427149] [<ffffffffa005b9b5>] _req_may_be_done+0x325/0x560 [drbd] [ 1532.427149] [<ffffffff810fabe1>] ? kmem_cache_free+0xe1/0xf0 [ 1532.427149] [<ffffffffa005bc0d>] _req_may_be_done_not_susp+0x1d/0x30 [drbd] [ 1532.427149] [<ffffffffa005c0cb>] __req_mod+0x4ab/0xee0 [drbd] [ 1532.427149] [<ffffffffa004f22e>] drbd_endio_pri+0xae/0x120 [drbd] [ 1532.427149] [<ffffffff81130b6c>] bio_endio+0x1c/0x40 [ 1532.427149] [<ffffffff813737d3>] req_bio_endio+0x83/0xb0 [ 1532.427149] [<ffffffff81373900>] blk_update_request+0x100/0x470 [ 1532.427149] [<ffffffff81373c92>] blk_update_bidi_request+0x22/0x90 [ 1532.427149] [<ffffffff81374e5a>] blk_end_bidi_request+0x2a/0x80 [ 1532.427149] [<ffffffff81374efb>] blk_end_request+0xb/0x10 [ 1532.427149] [<ffffffff8146744d>] scsi_io_completion+0x11d/0x620 [ 1532.427149] [<ffffffff8145f2ff>] scsi_finish_command+0xbf/0x120 [ 1532.427149] [<ffffffff81467a9e>] scsi_softirq_done+0x13e/0x160 [ 1532.427149] [<ffffffff8137ab7d>] blk_done_softirq+0x6d/0x80 [ 1532.427149] [<ffffffff8104d0f0>] __do_softirq+0xd0/0x1c0 [ 1532.427149] [<ffffffff816b726c>] call_softirq+0x1c/0x30 [ 1532.427149] [<ffffffff81004035>] do_softirq+0x55/0x90 [ 1532.427149] [<ffffffff8104ce2d>] irq_exit+0xad/0xe0 [ 1532.427149] [<ffffffff81003824>] do_IRQ+0x64/0xe0 [ 1532.427149] [<ffffffff816adf6b>] common_interrupt+0x6b/0x6b [ 1532.427149] <EOI> [ 1545.171011] Kernel panic - not syncing: Watchdog detected hard LOCKUP on cpu 0 [ 1545.171011] Pid: 20283, comm: xfsbufd/drbd0 Not tainted 3.1.0-rc4-hw+ #28 [ 1545.171011] Call Trace: [ 1545.171011] <NMI> [<ffffffff816aa954>] panic+0xba/0x1f5 [ 1545.171011] [<ffffffff8108f6d1>] watchdog_overflow_callback+0xb1/0xc0 [ 1545.171011] [<ffffffff810bb442>] __perf_event_overflow+0xa2/0x1f0 [ 1545.171011] [<ffffffff810b4981>] ? perf_event_update_userpage+0x11/0xc0 [ 1545.171011] [<ffffffff810bb994>] perf_event_overflow+0x14/0x20 [ 1545.171011] [<ffffffff81014bcb>] intel_pmu_handle_irq+0x21b/0x460 [ 1545.171011] [<ffffffff816af286>] perf_event_nmi_handler+0x36/0xb0 [ 1545.171011] [<ffffffff816b132f>] notifier_call_chain+0x3f/0x80 [ 1545.171011] [<ffffffff816b1395>] atomic_notifier_call_chain+0x15/0x20 [ 1545.171011] [<ffffffff816b13ce>] notify_die+0x2e/0x30 [ 1545.171011] [<ffffffff816ae9e2>] do_nmi+0xa2/0x250 [ 1545.171011] [<ffffffff816ae3ca>] nmi+0x1a/0x20 [ 1545.171011] [<ffffffff816adb07>] ? _raw_spin_lock_irqsave+0x17/0x20 [ 1545.171011] <<EOE>> <IRQ> [<ffffffffa004f21c>] drbd_endio_pri+0x9c/0x120 [drbd] [ 1545.171011] [<ffffffff814302d2>] ? put_device+0x12/0x20 [ 1545.171011] [<ffffffff81130b6c>] bio_endio+0x1c/0x40 [ 1545.171011] [<ffffffff813737d3>] req_bio_endio+0x83/0xb0 [ 1545.171011] [<ffffffff81373900>] blk_update_request+0x100/0x470 [ 1545.171011] [<ffffffff81373c92>] blk_update_bidi_request+0x22/0x90 [ 1545.171011] [<ffffffff81374e5a>] blk_end_bidi_request+0x2a/0x80 [ 1545.171011] [<ffffffff81374efb>] blk_end_request+0xb/0x10 [ 1545.171011] [<ffffffff8146744d>] scsi_io_completion+0x11d/0x620 [ 1545.171011] [<ffffffff8145f2ff>] scsi_finish_command+0xbf/0x120 [ 1545.171011] [<ffffffff81467a9e>] scsi_softirq_done+0x13e/0x160 [ 1545.171011] [<ffffffff8137ab7d>] blk_done_softirq+0x6d/0x80 [ 1545.171011] [<ffffffff8104d0f0>] __do_softirq+0xd0/0x1c0 [ 1545.171011] [<ffffffff816b726c>] call_softirq+0x1c/0x30 [ 1545.171011] [<ffffffff81004035>] do_softirq+0x55/0x90 [ 1545.171011] [<ffffffff8104ce2d>] irq_exit+0xad/0xe0 [ 1545.171011] [<ffffffff8101cd0f>] smp_call_function_single_interrupt+0x2f/0x40 [ 1545.171011] [<ffffffff816b6d4b>] call_function_single_interrupt+0x6b/0x70 [ 1545.171011] <EOI> [<ffffffff8146635c>] ? scsi_request_fn+0xec/0x510 [ 1545.171011] [<ffffffff8136fd25>] __blk_run_queue+0x15/0x20 [ 1545.171011] [<ffffffff813876ec>] cfq_insert_request+0x49c/0x5c0 [ 1545.171011] [<ffffffff8136f658>] __elv_add_request+0x198/0x2a0 [ 1545.171011] [<ffffffff81376410>] blk_flush_plug_list+0x1d0/0x210 [ 1545.171011] [<ffffffff816ab70a>] schedule+0x62a/0xad0 [ 1545.171011] [<ffffffff810fb26e>] ? kmem_cache_alloc+0xee/0x100 [ 1545.171011] [<ffffffffa0042199>] ? lc_get+0x49/0x250 [lru_cache] [ 1545.171011] [<ffffffffa0061815>] drbd_al_begin_io+0x1c5/0x200 [drbd] [ 1545.171011] [<ffffffff81132784>] ? bio_alloc_bioset+0x54/0xe0 [ 1545.171011] [<ffffffff81067030>] ? wake_up_bit+0x40/0x40 [ 1545.171011] [<ffffffff81131591>] ? __bio_clone+0x21/0x60 [ 1545.171011] [<ffffffffa005d825>] drbd_make_request_common+0xd25/0x1650 [drbd] [ 1545.171011] [<ffffffff8136e916>] ? elv_set_request+0x16/0x40 [ 1545.171011] [<ffffffffa005e48a>] drbd_make_request+0x33a/0x1150 [drbd] [ 1545.171011] [<ffffffff8137a175>] ? blk_recount_segments+0x25/0x40 [ 1545.171011] [<ffffffff81131b81>] ? bio_phys_segments+0x21/0x30 [ 1545.171011] [<ffffffff8137410e>] ? drive_stat_acct+0x11e/0x180 [ 1545.171011] [<ffffffff8137665d>] ? __make_request+0x1bd/0x330 [ 1545.171011] [<ffffffff81374713>] generic_make_request+0x3f3/0x540 [ 1545.171011] [<ffffffff8138c327>] ? kobject_put+0x27/0x60 [ 1545.171011] [<ffffffff813748d6>] submit_bio+0x76/0xf0 [ 1545.171011] [<ffffffff81278a57>] _xfs_buf_ioapply+0xb7/0x1f0 [ 1545.171011] [<ffffffff8127a3a5>] ? xfs_bdstrat_cb+0x85/0xb0 [ 1545.171011] [<ffffffff8127a251>] xfs_buf_iorequest+0x71/0xc0 [ 1545.171011] [<ffffffff8127a3a5>] xfs_bdstrat_cb+0x85/0xb0 [ 1545.171011] [<ffffffff8127a4c6>] xfsbufd+0xf6/0x130 [ 1545.171011] [<ffffffff8127a3d0>] ? xfs_bdstrat_cb+0xb0/0xb0 [ 1545.171011] [<ffffffff81066b16>] kthread+0x96/0xb0 [ 1545.171011] [<ffffffff816b7174>] kernel_thread_helper+0x4/0x10 [ 1545.171011] [<ffffffff81066a80>] ? kthread_worker_fn+0x130/0x130 [ 1545.171011] [<ffffffff816b7170>] ? gs_change+0xb/0xb
Quoted 4 lines
We upgraded from 2.6.36 which seemed to have a page leak (file pages left on the LRU) and so would eventually perform very poorly. 2.6.37 and 2.6.38 seemed to have some unix socket issue that caused heartbeat to wedge. Shall we enable lock debugging or something here?
That could help us understand that stack trace.
It looks like cpu 1 blocks in
Quoted 2 lines
[ 1532.427149] [<ffffffff8103d512>] ? try_to_wake_up+0xc2/0x270 [ 1532.427149] <<EOE>> <IRQ> [<ffffffff8103d6cd>] default_wake_function+0xd/0x10
Which does not make sense to me at all.
Signature · 3 lines
: Lars Ellenberg : LINBIT | Your Way to High Availability : DRBD/HA support and consulting http://www.linbit.com
Lars Ellenberg
maintainerSorry for double posting on drbd-dev, I managed to strip the other lists from Cc.
Quoted 130 lines · Simon Kirby
We will try again with most recent Linus HEAD, but with a build of Linus
e9208a4eec8acbde7ede6516c39dea05f3b700dc(3.1-rc4), we are seeing awesome VM performance (low memory fragmentation, good caching), followed quickly by a hard lockup. APEI and pstore stuff just doesn't seem to like this PowerEdge 2950 (ERST: Failed to get Error Log Address Range.), so this was captured over an old-fashioned serial cable:[ 93.004364] block drbd0: Resync done (total 52 sec; paused 0 sec; 3884 K/sec) [ 93.008004] block drbd0: updated UUIDs D1D6A96AAE5E394E:0000000000000000:264787A8207F2376:7E86E22A1F92B7FB [ 93.008004] block drbd0: conn( SyncTarget -> Connected ) disk( Inconsistent -> UpToDate ) [ 93.030394] block drbd0: helper command: /sbin/drbdadm after-resync-target minor-0 [ 93.044631] block drbd0: helper command: /sbin/drbdadm after-resync-target minor-0 exit code 0 (0x0) [ 93.102981] block drbd0: bitmap WRITE of 3382 pages took 10 jiffies [ 93.109462] block drbd0: 0 KB (0 bits) marked out-of-sync by on disk bit-map. [ 1114.600102] block drbd0: peer( Primary -> Secondary ) [ 1114.904031] block drbd0: role( Secondary -> Primary ) [ 1115.284503] XFS (drbd0): Mounting Filesystem [ 1115.928766] XFS (drbd0): Ending clean mount [ 1116.324036] NFSD: Using /var/lib/nfs/v4recovery as the NFSv4 state recovery directory [ 1116.336050] NFSD: starting 90-second grace period [ 1532.427149] Kernel panic - not syncing: Watchdog detected hard LOCKUP on cpu 1 [ 1532.427149] Pid: 3590, comm: lmtp Not tainted 3.1.0-rc4-hw+ #28 [ 1532.427149] Call Trace: [ 1532.427149] <NMI> [<ffffffff816aa954>] panic+0xba/0x1f5 [ 1532.427149] [<ffffffff8108f6d1>] watchdog_overflow_callback+0xb1/0xc0 [ 1532.427149] [<ffffffff810bb442>] __perf_event_overflow+0xa2/0x1f0 [ 1532.427149] [<ffffffff810b4981>] ? perf_event_update_userpage+0x11/0xc0 [ 1532.427149] [<ffffffff810bb994>] perf_event_overflow+0x14/0x20 [ 1532.427149] [<ffffffff81014bcb>] intel_pmu_handle_irq+0x21b/0x460 [ 1532.427149] [<ffffffff816af286>] perf_event_nmi_handler+0x36/0xb0 [ 1532.427149] [<ffffffff816b132f>] notifier_call_chain+0x3f/0x80 [ 1532.427149] [<ffffffff816b1395>] atomic_notifier_call_chain+0x15/0x20 [ 1532.427149] [<ffffffff816b13ce>] notify_die+0x2e/0x30 [ 1532.427149] [<ffffffff816ae9e2>] do_nmi+0xa2/0x250 [ 1532.427149] [<ffffffff816ae3ca>] nmi+0x1a/0x20 [ 1532.427149] [<ffffffff8103d512>] ? try_to_wake_up+0xc2/0x270 [ 1532.427149] <<EOE>> <IRQ> [<ffffffff8103d6cd>] default_wake_function+0xd/0x10 [ 1532.427149] [<ffffffff81067041>] autoremove_wake_function+0x11/0x40 [ 1532.427149] [<ffffffff8103075a>] __wake_up_common+0x5a/0x90 [ 1532.427149] [<ffffffff81037c43>] __wake_up+0x43/0x70 [ 1532.427149] [<ffffffffa0060579>] drbd_al_complete_io+0x99/0xd0 [drbd] [ 1532.427149] [<ffffffffa005b9b5>] _req_may_be_done+0x325/0x560 [drbd] [ 1532.427149] [<ffffffff810fabe1>] ? kmem_cache_free+0xe1/0xf0 [ 1532.427149] [<ffffffffa005bc0d>] _req_may_be_done_not_susp+0x1d/0x30 [drbd] [ 1532.427149] [<ffffffffa005c0cb>] __req_mod+0x4ab/0xee0 [drbd] [ 1532.427149] [<ffffffffa004f22e>] drbd_endio_pri+0xae/0x120 [drbd] [ 1532.427149] [<ffffffff81130b6c>] bio_endio+0x1c/0x40 [ 1532.427149] [<ffffffff813737d3>] req_bio_endio+0x83/0xb0 [ 1532.427149] [<ffffffff81373900>] blk_update_request+0x100/0x470 [ 1532.427149] [<ffffffff81373c92>] blk_update_bidi_request+0x22/0x90 [ 1532.427149] [<ffffffff81374e5a>] blk_end_bidi_request+0x2a/0x80 [ 1532.427149] [<ffffffff81374efb>] blk_end_request+0xb/0x10 [ 1532.427149] [<ffffffff8146744d>] scsi_io_completion+0x11d/0x620 [ 1532.427149] [<ffffffff8145f2ff>] scsi_finish_command+0xbf/0x120 [ 1532.427149] [<ffffffff81467a9e>] scsi_softirq_done+0x13e/0x160 [ 1532.427149] [<ffffffff8137ab7d>] blk_done_softirq+0x6d/0x80 [ 1532.427149] [<ffffffff8104d0f0>] __do_softirq+0xd0/0x1c0 [ 1532.427149] [<ffffffff816b726c>] call_softirq+0x1c/0x30 [ 1532.427149] [<ffffffff81004035>] do_softirq+0x55/0x90 [ 1532.427149] [<ffffffff8104ce2d>] irq_exit+0xad/0xe0 [ 1532.427149] [<ffffffff81003824>] do_IRQ+0x64/0xe0 [ 1532.427149] [<ffffffff816adf6b>] common_interrupt+0x6b/0x6b [ 1532.427149] <EOI> [ 1545.171011] Kernel panic - not syncing: Watchdog detected hard LOCKUP on cpu 0 [ 1545.171011] Pid: 20283, comm: xfsbufd/drbd0 Not tainted 3.1.0-rc4-hw+ #28 [ 1545.171011] Call Trace: [ 1545.171011] <NMI> [<ffffffff816aa954>] panic+0xba/0x1f5 [ 1545.171011] [<ffffffff8108f6d1>] watchdog_overflow_callback+0xb1/0xc0 [ 1545.171011] [<ffffffff810bb442>] __perf_event_overflow+0xa2/0x1f0 [ 1545.171011] [<ffffffff810b4981>] ? perf_event_update_userpage+0x11/0xc0 [ 1545.171011] [<ffffffff810bb994>] perf_event_overflow+0x14/0x20 [ 1545.171011] [<ffffffff81014bcb>] intel_pmu_handle_irq+0x21b/0x460 [ 1545.171011] [<ffffffff816af286>] perf_event_nmi_handler+0x36/0xb0 [ 1545.171011] [<ffffffff816b132f>] notifier_call_chain+0x3f/0x80 [ 1545.171011] [<ffffffff816b1395>] atomic_notifier_call_chain+0x15/0x20 [ 1545.171011] [<ffffffff816b13ce>] notify_die+0x2e/0x30 [ 1545.171011] [<ffffffff816ae9e2>] do_nmi+0xa2/0x250 [ 1545.171011] [<ffffffff816ae3ca>] nmi+0x1a/0x20 [ 1545.171011] [<ffffffff816adb07>] ? _raw_spin_lock_irqsave+0x17/0x20 [ 1545.171011] <<EOE>> <IRQ> [<ffffffffa004f21c>] drbd_endio_pri+0x9c/0x120 [drbd] [ 1545.171011] [<ffffffff814302d2>] ? put_device+0x12/0x20 [ 1545.171011] [<ffffffff81130b6c>] bio_endio+0x1c/0x40 [ 1545.171011] [<ffffffff813737d3>] req_bio_endio+0x83/0xb0 [ 1545.171011] [<ffffffff81373900>] blk_update_request+0x100/0x470 [ 1545.171011] [<ffffffff81373c92>] blk_update_bidi_request+0x22/0x90 [ 1545.171011] [<ffffffff81374e5a>] blk_end_bidi_request+0x2a/0x80 [ 1545.171011] [<ffffffff81374efb>] blk_end_request+0xb/0x10 [ 1545.171011] [<ffffffff8146744d>] scsi_io_completion+0x11d/0x620 [ 1545.171011] [<ffffffff8145f2ff>] scsi_finish_command+0xbf/0x120 [ 1545.171011] [<ffffffff81467a9e>] scsi_softirq_done+0x13e/0x160 [ 1545.171011] [<ffffffff8137ab7d>] blk_done_softirq+0x6d/0x80 [ 1545.171011] [<ffffffff8104d0f0>] __do_softirq+0xd0/0x1c0 [ 1545.171011] [<ffffffff816b726c>] call_softirq+0x1c/0x30 [ 1545.171011] [<ffffffff81004035>] do_softirq+0x55/0x90 [ 1545.171011] [<ffffffff8104ce2d>] irq_exit+0xad/0xe0 [ 1545.171011] [<ffffffff8101cd0f>] smp_call_function_single_interrupt+0x2f/0x40 [ 1545.171011] [<ffffffff816b6d4b>] call_function_single_interrupt+0x6b/0x70 [ 1545.171011] <EOI> [<ffffffff8146635c>] ? scsi_request_fn+0xec/0x510 [ 1545.171011] [<ffffffff8136fd25>] __blk_run_queue+0x15/0x20 [ 1545.171011] [<ffffffff813876ec>] cfq_insert_request+0x49c/0x5c0 [ 1545.171011] [<ffffffff8136f658>] __elv_add_request+0x198/0x2a0 [ 1545.171011] [<ffffffff81376410>] blk_flush_plug_list+0x1d0/0x210 [ 1545.171011] [<ffffffff816ab70a>] schedule+0x62a/0xad0 [ 1545.171011] [<ffffffff810fb26e>] ? kmem_cache_alloc+0xee/0x100 [ 1545.171011] [<ffffffffa0042199>] ? lc_get+0x49/0x250 [lru_cache] [ 1545.171011] [<ffffffffa0061815>] drbd_al_begin_io+0x1c5/0x200 [drbd] [ 1545.171011] [<ffffffff81132784>] ? bio_alloc_bioset+0x54/0xe0 [ 1545.171011] [<ffffffff81067030>] ? wake_up_bit+0x40/0x40 [ 1545.171011] [<ffffffff81131591>] ? __bio_clone+0x21/0x60 [ 1545.171011] [<ffffffffa005d825>] drbd_make_request_common+0xd25/0x1650 [drbd] [ 1545.171011] [<ffffffff8136e916>] ? elv_set_request+0x16/0x40 [ 1545.171011] [<ffffffffa005e48a>] drbd_make_request+0x33a/0x1150 [drbd] [ 1545.171011] [<ffffffff8137a175>] ? blk_recount_segments+0x25/0x40 [ 1545.171011] [<ffffffff81131b81>] ? bio_phys_segments+0x21/0x30 [ 1545.171011] [<ffffffff8137410e>] ? drive_stat_acct+0x11e/0x180 [ 1545.171011] [<ffffffff8137665d>] ? __make_request+0x1bd/0x330 [ 1545.171011] [<ffffffff81374713>] generic_make_request+0x3f3/0x540 [ 1545.171011] [<ffffffff8138c327>] ? kobject_put+0x27/0x60 [ 1545.171011] [<ffffffff813748d6>] submit_bio+0x76/0xf0 [ 1545.171011] [<ffffffff81278a57>] _xfs_buf_ioapply+0xb7/0x1f0 [ 1545.171011] [<ffffffff8127a3a5>] ? xfs_bdstrat_cb+0x85/0xb0 [ 1545.171011] [<ffffffff8127a251>] xfs_buf_iorequest+0x71/0xc0 [ 1545.171011] [<ffffffff8127a3a5>] xfs_bdstrat_cb+0x85/0xb0 [ 1545.171011] [<ffffffff8127a4c6>] xfsbufd+0xf6/0x130 [ 1545.171011] [<ffffffff8127a3d0>] ? xfs_bdstrat_cb+0xb0/0xb0 [ 1545.171011] [<ffffffff81066b16>] kthread+0x96/0xb0 [ 1545.171011] [<ffffffff816b7174>] kernel_thread_helper+0x4/0x10 [ 1545.171011] [<ffffffff81066a80>] ? kthread_worker_fn+0x130/0x130 [ 1545.171011] [<ffffffff816b7170>] ? gs_change+0xb/0xb
Quoted 4 lines
We upgraded from 2.6.36 which seemed to have a page leak (file pages left on the LRU) and so would eventually perform very poorly. 2.6.37 and 2.6.38 seemed to have some unix socket issue that caused heartbeat to wedge. Shall we enable lock debugging or something here?
That could help us understand that stack trace.
It looks like cpu 1 blocks in
Quoted 2 lines
[ 1532.427149] [<ffffffff8103d512>] ? try_to_wake_up+0xc2/0x270 [ 1532.427149] <<EOE>> <IRQ> [<ffffffff8103d6cd>] default_wake_function+0xd/0x10
Which does not make sense to me at all.
Signature · 3 lines
: Lars Ellenberg : LINBIT | Your Way to High Availability : DRBD/HA support and consulting http://www.linbit.com
Quoted 7 lines · Lars Ellenberg
Sorry for double posting on drbd-dev, I managed to strip the other lists from Cc.
Quoted 4 lines
We upgraded from 2.6.36 which seemed to have a page leak (file pages left on the LRU) and so would eventually perform very poorly. 2.6.37 and 2.6.38 seemed to have some unix socket issue that caused heartbeat to wedge. Shall we enable lock debugging or something here?
Quoted 8 lines
That could help us understand that stack trace.
It looks like cpu 1 blocks in
Quoted 2 lines
[ 1532.427149] [<ffffffff8103d512>] ? try_to_wake_up+0xc2/0x270 [ 1532.427149] <<EOE>> <IRQ> [<ffffffff8103d6cd>] default_wake_function+0xd/0x10Which does not make sense to me at all.
Well, good news, I think.. I believe this may be related to "PCI: Set PCI-E Max Payload Size on fabric", added by b03e7495a862b02829. 3.1-rc5 is running now with a patch to basically disable those changes, and has been stable for 12 hours. It usually hung in a few minutes before.
The XFS peoples say it was very likely not 58d84c4ee0389ddeb86238d5 which is the only other thing that changed between these versions that seems to be at all in the hang path.
Also, when the thing hangs, it stops pinging immediately, and with the PCI-E max payload thing active, the device that raises a bus error is actually the PCI-E to PCI-X bridge chip used to support the BCM5708 NICs, so that all seems related.
Simon-
Quoted 24 lines · Simon Kirby
Quoted 7 lines · Lars Ellenberg
Sorry for double posting on drbd-dev, I managed to strip the other lists from Cc.
Quoted 4 lines
We upgraded from 2.6.36 which seemed to have a page leak (file pages left on the LRU) and so would eventually perform very poorly. 2.6.37 and 2.6.38 seemed to have some unix socket issue that caused heartbeat to wedge. Shall we enable lock debugging or something here?
Quoted 8 lines
That could help us understand that stack trace.
It looks like cpu 1 blocks in
Quoted 2 lines
[ 1532.427149] [<ffffffff8103d512>] ? try_to_wake_up+0xc2/0x270 [ 1532.427149] <<EOE>> <IRQ> [<ffffffff8103d6cd>] default_wake_function+0xd/0x10Which does not make sense to me at all.
Well, good news, I think.. I believe this may be related to "PCI: Set PCI-E Max Payload Size on fabric", added by
b03e7495a862b02829. 3.1-rc5 is running now with a patch to basically disable those changes, and has been stable for 12 hours. It usually hung in a few minutes before.
Quoted 8 lines
The XFS peoples say it was very likely not
58d84c4ee0389ddeb86238d5which is the only other thing that changed between these versions that seems to be at all in the hang path.Also, when the thing hangs, it stops pinging immediately, and with the PCI-E max payload thing active, the device that raises a bus error is actually the PCI-E to PCI-X bridge chip used to support the BCM5708 NICs, so that all seems related.
Except that I accidentally git reset out the patch, and so it's been running unmodified 79016f648872549392d232cd648bd02298c2d2bb (past -rc5), and still hasn't crashed, so I guess it _was_ the XFS changes, or something else. Boggle. In any event, it's still running well. :)
Simon-