[LU-7329] sanity test_60a timeouts with “* invoking oom-killer” Created: 22/Oct/15  Updated: 24/Oct/17  Resolved: 30/Nov/15

Status: Resolved
Project: Lustre
Component/s: None
Affects Version/s: Lustre 2.8.0
Fix Version/s: Lustre 2.8.0

Type: Bug Priority: Critical
Reporter: James Nunez (Inactive) Assignee: Bruno Faccini (Inactive)
Resolution: Fixed Votes: 0
Labels: None
Environment:

autotest


Issue Links:
Related
is related to LU-10056 sanity test_60a invokes oom-killer in... Open
is related to LU-6556 changelog catalog corruption if all p... Resolved
Severity: 3
Rank (Obsolete): 9223372036854775807

 Description   

We’ve seen four recent hangs on the MDS during sanity test 60a due to OOM.

For https://testing.hpdd.intel.com/test_sets/383ea288-787c-11e5-b04c-5254006e85c2, we see the following from the console on the MDS:

23:58:40:Lustre: 1823:0:(llog_test.c:962:llog_test_7_sub()) 7_sub: records are not aligned, written 64452 from 64767
23:58:40:Lustre: 1823:0:(llog_test.c:1079:llog_test_7()) 7g: test llog_gen_rec
23:58:40:hald invoked oom-killer: gfp_mask=0x200da, order=0, oom_adj=0, oom_score_adj=0
23:58:40:hald cpuset=/ mems_allowed=0
23:58:40:Pid: 1967, comm: hald Tainted: P           -- ------------    2.6.32-573.7.1.el6_lustre.gd359461.x86_64 #1
23:58:40:Call Trace:
23:58:40: [<ffffffff810d71a1>] ? cpuset_print_task_mems_allowed+0x91/0xb0
23:58:40: [<ffffffff8112a9a0>] ? dump_header+0x90/0x1b0
23:58:40: [<ffffffff8112ab0e>] ? check_panic_on_oom+0x4e/0x80
23:58:40: [<ffffffff8112b1fb>] ? out_of_memory+0x1bb/0x3c0
23:58:40: [<ffffffff81137c3c>] ? __alloc_pages_nodemask+0x93c/0x950
23:58:40: [<ffffffff815395e2>] ? io_schedule+0x92/0xc0
23:58:40: [<ffffffff811279c0>] ? sync_page_killable+0x0/0x40
23:58:40: [<ffffffff8117074a>] ? alloc_pages_vma+0x9a/0x150
23:58:40: [<ffffffff81163c72>] ? read_swap_cache_async+0xf2/0x160
23:58:40: [<ffffffff811647c9>] ? valid_swaphandles+0x69/0x160
23:58:40: [<ffffffff81163d67>] ? swapin_readahead+0x87/0xc0
23:58:40: [<ffffffff81152ead>] ? handle_pte_fault+0x6dd/0xb20
23:58:40: [<ffffffff81153589>] ? handle_mm_fault+0x299/0x3d0
23:58:40: [<ffffffff81155fa6>] ? find_vma+0x46/0x80
23:58:40: [<ffffffff8104f156>] ? __do_page_fault+0x146/0x500
23:58:40: [<ffffffff81043f28>] ? pvclock_clocksource_read+0x58/0xd0
23:58:40: [<ffffffff81042fbc>] ? kvm_clock_read+0x1c/0x20
23:58:40: [<ffffffff81042fc9>] ? kvm_clock_get_cycles+0x9/0x10
23:58:40: [<ffffffff8153f3de>] ? do_page_fault+0x3e/0xa0
23:58:40: [<ffffffff8153c785>] ? page_fault+0x25/0x30
23:58:40:Mem-Info:
23:58:40:Node 0 DMA per-cpu:
23:58:40:CPU    0: hi:    0, btch:   1 usd:   0
23:58:40:CPU    1: hi:    0, btch:   1 usd:   0
23:58:40:Node 0 DMA32 per-cpu:
23:58:40:CPU    0: hi:  186, btch:  31 usd:  30
23:58:40:CPU    1: hi:  186, btch:  31 usd:   4
23:58:40:active_anon:97 inactive_anon:877 isolated_anon:0
23:58:40: active_file:3 inactive_file:55 isolated_file:0
23:58:40: unevictable:0 dirty:0 writeback:876 unstable:0
23:58:40: free:13242 slab_reclaimable:1788 slab_unreclaimable:442860
23:58:40: mapped:1 shmem:1 pagetables:681 bounce:0
23:58:40:Node 0 DMA free:8356kB min:332kB low:412kB high:496kB active_anon:0kB inactive_anon:0kB active_file:0kB inactive_file:0kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:15348kB mlocked:0kB dirty:0kB writeback:0kB mapped:0kB shmem:0kB slab_reclaimable:0kB slab_unreclaimable:7388kB kernel_stack:0kB pagetables:0kB unstable:0kB bounce:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? yes
23:58:40:lowmem_reserve[]: 0 2004 2004 2004
23:58:40:Node 0 DMA32 free:44612kB min:44720kB low:55900kB high:67080kB active_anon:388kB inactive_anon:3508kB active_file:12kB inactive_file:220kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:2052308kB mlocked:0kB dirty:0kB writeback:3504kB mapped:4kB shmem:4kB slab_reclaimable:7152kB slab_unreclaimable:1764052kB kernel_stack:4112kB pagetables:2724kB unstable:0kB bounce:0kB writeback_tmp:0kB pages_scanned:137468 all_unreclaimable? yes
23:58:40:lowmem_reserve[]: 0 0 0 0
23:58:40:Node 0 DMA: 5*4kB 0*8kB 1*16kB 0*32kB 0*64kB 1*128kB 0*256kB 0*512kB 0*1024kB 2*2048kB 1*4096kB = 8356kB
23:58:40:Node 0 DMA32: 527*4kB 271*8kB 147*16kB 109*32kB 69*64kB 47*128kB 20*256kB 11*512kB 7*1024kB 1*2048kB 1*4096kB = 44612kB
23:58:40:957 total pagecache pages
23:58:40:898 pages in swap cache
23:58:40:Swap cache stats: add 5911, delete 5013, find 54/85
23:58:40:Free swap  = 4106092kB
23:58:40:Total swap = 4128764kB
23:58:40:524284 pages RAM
23:58:40:43736 pages reserved
23:58:40:88 pages shared
23:58:40:438370 pages non-shared
23:58:40:[ pid ]   uid  tgid total_vm      rss cpu oom_adj oom_score_adj name
23:58:40:[  520]     0   520     2709        0   0     -17         -1000 udevd
23:58:40:[ 1706]     0  1706     6902        1   0     -17         -1000 auditd
23:58:40:[ 1740]     0  1740    62272       13   0       0             0 rsyslogd
23:58:40:[ 1774]     0  1774     4561       18   0       0             0 irqbalance
23:58:40:[ 1792]    32  1792     4744        1   0       0             0 rpcbind
23:58:40:[ 1814]    29  1814     5837        1   1       0             0 rpc.statd
23:58:40:[ 1848]    81  1848     6418        1   1       0             0 dbus-daemon
23:58:40:[ 1868]     0  1868    53919        1   1       0             0 ypbind
23:58:40:[ 1898]     0  1898    47233        1   1       0             0 cupsd
23:58:40:[ 1955]     0  1955     1020        0   1       0             0 acpid
23:58:40:[ 1967]    68  1967    10517       37   1       0             0 hald
23:58:40:[ 1968]     0  1968     5099        1   1       0             0 hald-runner
23:58:40:[ 2000]     0  2000     5629        1   1       0             0 hald-addon-inpu
23:58:40:[ 2010]    68  2010     4501        1   1       0             0 hald-addon-acpi
23:58:40:[ 2029]     0  2029     7646        1   1       0             0 zed
23:58:40:[ 2067]     0  2067    26827        0   0       0             0 rpc.rquotad
23:58:40:[ 2072]     0  2072     5417        0   0       0             0 rpc.mountd
23:58:40:[ 2116]     0  2116     6292        1   0       0             0 rpc.idmapd
23:58:40:[ 2163]   498  2163    57325        1   0       0             0 munged
23:58:40:[ 2274]     0  2274    16555        0   0     -17         -1000 sshd
23:58:40:[ 2285]     0  2285     5429        1   1       0             0 xinetd
23:58:40:[ 2317]     0  2317    22211       33   1       0             0 sendmail
23:58:40:[ 2326]    51  2326    20074        0   0       0             0 sendmail
23:58:40:[ 2354]     0  2354    29216        1   0       0             0 crond
23:58:40:[ 2369]     0  2369     5276        0   0       0             0 atd
23:58:40:[ 2382]     0  2382     1020        1   1       0             0 agetty
23:58:40:[ 2384]     0  2384     1016        1   0       0             0 mingetty
23:58:40:[ 2386]     0  2386     1016        1   1       0             0 mingetty
23:58:40:[ 2388]     0  2388     1016        1   1       0             0 mingetty
23:58:40:[ 2390]     0  2390     1016        1   1       0             0 mingetty
23:58:40:[ 2392]     0  2392     1016        1   1       0             0 mingetty
23:58:40:[ 2394]     0  2394     1016        1   1       0             0 mingetty
23:58:40:[ 2396]     0  2396     2664        0   0     -17         -1000 udevd
23:58:40:[ 2397]     0  2397     2662        0   0     -17         -1000 udevd
23:58:40:[ 2963]    38  2963     7689        1   1       0             0 ntpd
23:58:40:[ 1713]     0  1713    14747        1   1       0             0 in.mrshd
23:58:40:[ 1714]     0  1714    26515        1   1       0             0 bash
23:58:40:[ 1744]     0  1744    26515        0   0       0             0 bash
23:58:40:[ 1745]     0  1745    27594        1   1       0             0 sh
23:58:40:[ 1823]     0  1823    28024        1   0       0             0 lctl
23:58:40:Kernel panic - not syncing: Out of memory: system-wide panic_on_oom is enabled
23:58:40:
23:58:40:Pid: 1967, comm: hald Tainted: P           -- ------------    2.6.32-573.7.1.el6_lustre.gd359461.x86_64 #1
23:58:40:Call Trace:
23:58:40: [<ffffffff815386d4>] ? panic+0xa7/0x16f
23:58:40: [<ffffffff8112aaa1>] ? dump_header+0x191/0x1b0
23:58:40: [<ffffffff8112ab3c>] ? check_panic_on_oom+0x7c/0x80
23:58:40: [<ffffffff8112b1fb>] ? out_of_memory+0x1bb/0x3c0
23:58:40: [<ffffffff81137c3c>] ? __alloc_pages_nodemask+0x93c/0x950
23:58:40: [<ffffffff815395e2>] ? io_schedule+0x92/0xc0
23:58:40: [<ffffffff811279c0>] ? sync_page_killable+0x0/0x40
23:58:40: [<ffffffff8117074a>] ? alloc_pages_vma+0x9a/0x150
23:58:40: [<ffffffff81163c72>] ? read_swap_cache_async+0xf2/0x160
23:58:40: [<ffffffff811647c9>] ? valid_swaphandles+0x69/0x160
23:58:40: [<ffffffff81163d67>] ? swapin_readahead+0x87/0xc0
23:58:40: [<ffffffff81152ead>] ? handle_pte_fault+0x6dd/0xb20
23:58:40: [<ffffffff81153589>] ? handle_mm_fault+0x299/0x3d0
23:58:40: [<ffffffff81155fa6>] ? find_vma+0x46/0x80
23:58:40: [<ffffffff8104f156>] ? __do_page_fault+0x146/0x500
23:58:40: [<ffffffff81043f28>] ? pvclock_clocksource_read+0x58/0xd0
23:58:40: [<ffffffff81042fbc>] ? kvm_clock_read+0x1c/0x20
23:58:40: [<ffffffff81042fc9>] ? kvm_clock_get_cycles+0x9/0x10
23:58:40: [<ffffffff8153f3de>] ? do_page_fault+0x3e/0xa0
23:58:40: [<ffffffff8153c785>] ? page_fault+0x25/0x30

For https://testing.hpdd.intel.com/test_sets/0ad70b9c-7858-11e5-9a01-5254006e85c2, we see the following from the console on the MDS:

17:40:08:Lustre: 20451:0:(llog_test.c:1692:llog_test_10()) 10g: Cancel 65536 records, see one log zapped
17:40:08:irqbalance invoked oom-killer: gfp_mask=0x201da, order=0, oom_adj=0, oom_score_adj=0
17:40:08:irqbalance cpuset=/ mems_allowed=0
17:40:08:Pid: 1349, comm: irqbalance Not tainted 2.6.32-573.7.1.el6_lustre.gd359461.x86_64 #1
17:40:08:Call Trace:
17:40:08: [<ffffffff810d71a1>] ? cpuset_print_task_mems_allowed+0x91/0xb0
17:40:08: [<ffffffff8112a9a0>] ? dump_header+0x90/0x1b0
17:40:08: [<ffffffff8112ab0e>] ? check_panic_on_oom+0x4e/0x80
17:40:08: [<ffffffff8112b1fb>] ? out_of_memory+0x1bb/0x3c0
17:40:08: [<ffffffff81137c3c>] ? __alloc_pages_nodemask+0x93c/0x950
17:40:08: [<ffffffff8117064a>] ? alloc_pages_current+0xaa/0x110
17:40:08: [<ffffffff81127d97>] ? __page_cache_alloc+0x87/0x90
17:40:08: [<ffffffff8112777e>] ? find_get_page+0x1e/0xa0
17:40:08: [<ffffffff81128d37>] ? filemap_fault+0x1a7/0x500
17:40:08: [<ffffffff811522f4>] ? __do_fault+0x54/0x530
17:40:08: [<ffffffff811528c7>] ? handle_pte_fault+0xf7/0xb20
17:40:08: [<ffffffff81066053>] ? perf_event_task_sched_out+0x33/0x70
17:40:08: [<ffffffff81043f28>] ? pvclock_clocksource_read+0x58/0xd0
17:40:08: [<ffffffff810098a5>] ? __switch_to+0x285/0x340
17:40:08: [<ffffffff81042fbc>] ? kvm_clock_read+0x1c/0x20
17:40:11: [<ffffffff81153589>] ? handle_mm_fault+0x299/0x3d0
17:40:11: [<ffffffff81538dce>] ? thread_return+0x4e/0x7d0
17:40:11: [<ffffffff8104f156>] ? __do_page_fault+0x146/0x500
17:40:11: [<ffffffff810a677d>] ? hrtimer_try_to_cancel+0x3d/0xd0
17:40:12: [<ffffffff810a6832>] ? hrtimer_cancel+0x22/0x30
17:40:12: [<ffffffff8153b463>] ? do_nanosleep+0x93/0xc0
17:40:12: [<ffffffff810a6904>] ? hrtimer_nanosleep+0xc4/0x180
17:40:12: [<ffffffff810a5730>] ? hrtimer_wakeup+0x0/0x30
17:40:12: [<ffffffff8153f3de>] ? do_page_fault+0x3e/0xa0
17:40:12: [<ffffffff8153c785>] ? page_fault+0x25/0x30
17:40:12:Mem-Info:
17:40:12:Node 0 DMA per-cpu:
17:40:12:CPU    0: hi:    0, btch:   1 usd:   0
17:40:12:CPU    1: hi:    0, btch:   1 usd:   0
17:40:12:Node 0 DMA32 per-cpu:
17:40:12:CPU    0: hi:  186, btch:  31 usd:  13
17:40:12:CPU    1: hi:  186, btch:  31 usd:  55
17:40:12:active_anon:1008 inactive_anon:1027 isolated_anon:0
17:40:12: active_file:862 inactive_file:834 isolated_file:0
17:40:12: unevictable:0 dirty:0 writeback:0 unstable:0
17:40:12: free:13265 slab_reclaimable:2472 slab_unreclaimable:437874
17:40:12: mapped:1 shmem:1 pagetables:709 bounce:0
17:40:12:Node 0 DMA free:8336kB min:332kB low:412kB high:496kB active_anon:0kB inactive_anon:0kB active_file:128kB inactive_file:0kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:15348kB mlocked:0kB dirty:0kB writeback:0kB mapped:0kB shmem:0kB slab_reclaimable:4kB slab_unreclaimable:7276kB kernel_stack:0kB pagetables:0kB unstable:0kB bounce:0kB writeback_tmp:0kB pages_scanned:157 all_unreclaimable? yes
17:40:12:lowmem_reserve[]: 0 2004 2004 2004
17:40:12:Node 0 DMA32 free:44724kB min:44720kB low:55900kB high:67080kB active_anon:4032kB inactive_anon:4108kB active_file:3320kB inactive_file:3336kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:2052308kB mlocked:0kB dirty:0kB writeback:0kB mapped:4kB shmem:4kB slab_reclaimable:9884kB slab_unreclaimable:1744220kB kernel_stack:3808kB pagetables:2836kB unstable:0kB bounce:0kB writeback_tmp:0kB pages_scanned:430063 all_unreclaimable? yes
17:40:12:lowmem_reserve[]: 0 0 0 0
17:40:12:Node 0 DMA: 2*4kB 1*8kB 2*16kB 1*32kB 1*64kB 2*128kB 3*256kB 2*512kB 2*1024kB 2*2048kB 0*4096kB = 8336kB
17:40:13:Node 0 DMA32: 1993*4kB 954*8kB 510*16kB 271*32kB 72*64kB 8*128kB 4*256kB 1*512kB 1*1024kB 0*2048kB 1*4096kB = 44724kB
17:40:13:2723 total pagecache pages
17:40:13:1026 pages in swap cache
17:40:13:Swap cache stats: add 3678, delete 2652, find 0/0
17:40:13:Free swap  = 4114052kB
17:40:13:Total swap = 4128764kB
17:40:13:524284 pages RAM
17:40:13:43736 pages reserved
17:40:13:1750 pages shared
17:40:13:438911 pages non-shared
17:40:13:[ pid ]   uid  tgid total_vm      rss cpu oom_adj oom_score_adj name
17:40:13:[  509]     0   509     2663        0   0     -17         -1000 udevd
17:40:13:[ 1281]     0  1281     6899        1   0     -17         -1000 auditd
17:40:13:[ 1315]     0  1315    62273       54   1       0             0 rsyslogd
17:40:13:[ 1349]     0  1349     4561        1   0       0             0 irqbalance
17:40:13:[ 1367]    32  1367     4744        1   1       0             0 rpcbind
17:40:13:[ 1389]    29  1389     5837        1   1       0             0 rpc.statd
17:40:13:[ 1423]    81  1423     6418        1   1       0             0 dbus-daemon
17:40:13:[ 1443]     0  1443    53919        1   1       0             0 ypbind
17:40:13:[ 1477]     0  1477    47233        1   1       0             0 cupsd
17:40:13:[ 1534]     0  1534     1020        0   1       0             0 acpid
17:40:13:[ 1546]    68  1546    10521        1   1       0             0 hald
17:40:13:[ 1547]     0  1547     5099        1   1       0             0 hald-runner
17:40:13:[ 1579]     0  1579     5629        1   1       0             0 hald-addon-inpu
17:40:13:[ 1586]    68  1586     4501        1   1       0             0 hald-addon-acpi
17:40:13:[ 1635]     0  1635    26827        0   0       0             0 rpc.rquotad
17:40:13:[ 1640]     0  1640     5417        0   0       0             0 rpc.mountd
17:40:13:[ 1684]     0  1684     6292        1   0       0             0 rpc.idmapd
17:40:13:[ 1721]   498  1721    57325        1   0       0             0 munged
17:40:13:[ 1832]     0  1832    16555        0   0     -17         -1000 sshd
17:40:13:[ 1843]     0  1843     5429       16   1       0             0 xinetd
17:40:13:[ 1875]     0  1875    22211        6   0       0             0 sendmail
17:40:13:[ 1884]    51  1884    20074        0   0       0             0 sendmail
17:40:13:[ 1912]     0  1912    29215        1   0       0             0 crond
17:40:13:[ 1927]     0  1927     5276        0   0       0             0 atd
17:40:13:[ 1940]     0  1940     1020        1   1       0             0 agetty
17:40:13:[ 1942]     0  1942     1016        1   0       0             0 mingetty
17:40:13:[ 1944]     0  1944     1016        1   0       0             0 mingetty
17:40:13:[ 1946]     0  1946     1016        1   0       0             0 mingetty
17:40:13:[ 1948]     0  1948     1016        1   1       0             0 mingetty
17:40:13:[ 1950]     0  1950     2730        0   0     -17         -1000 udevd
17:40:13:[ 1951]     0  1951     2728        0   0     -17         -1000 udevd
17:40:13:[ 1952]     0  1952     1016        1   0       0             0 mingetty
17:40:13:[ 1954]     0  1954     1016        1   0       0             0 mingetty
17:40:13:[ 2706]    38  2706     7689        1   1       0             0 ntpd
17:40:13:[11506]     0 11506     4763        1   1       0             0 anacron
17:40:13:[20349]     0 20349    14749      161   1       0             0 in.mrshd
17:40:13:[20350]     0 20350    26524       58   0       0             0 bash
17:40:13:[20375]     0 20375    26524       58   1       0             0 bash
17:40:13:[20376]     0 20376    27603      617   1       0             0 sh
17:40:13:[20451]     0 20451    28026       64   0       0             0 lctl
17:40:13:Kernel panic - not syncing: Out of memory: system-wide panic_on_oom is enabled
17:40:14:
17:40:14:Pid: 1349, comm: irqbalance Not tainted 2.6.32-573.7.1.el6_lustre.gd359461.x86_64 #1
17:40:14:Call Trace:
17:40:14: [<ffffffff815386d4>] ? panic+0xa7/0x16f
17:40:14: [<ffffffff8112aaa1>] ? dump_header+0x191/0x1b0
17:40:14: [<ffffffff8112ab3c>] ? check_panic_on_oom+0x7c/0x80
17:40:14: [<ffffffff8112b1fb>] ? out_of_memory+0x1bb/0x3c0
17:40:14: [<ffffffff81137c3c>] ? __alloc_pages_nodemask+0x93c/0x950
17:40:14: [<ffffffff8117064a>] ? alloc_pages_current+0xaa/0x110
17:40:14: [<ffffffff81127d97>] ? __page_cache_alloc+0x87/0x90
17:40:14: [<ffffffff8112777e>] ? find_get_page+0x1e/0xa0
17:40:14: [<ffffffff81128d37>] ? filemap_fault+0x1a7/0x500
17:40:14: [<ffffffff811522f4>] ? __do_fault+0x54/0x530
17:40:14: [<ffffffff811528c7>] ? handle_pte_fault+0xf7/0xb20
17:40:14: [<ffffffff81066053>] ? perf_event_task_sched_out+0x33/0x70
17:40:14: [<ffffffff81043f28>] ? pvclock_clocksource_read+0x58/0xd0
17:40:14: [<ffffffff810098a5>] ? __switch_to+0x285/0x340
17:40:14: [<ffffffff81042fbc>] ? kvm_clock_read+0x1c/0x20
17:40:14: [<ffffffff81153589>] ? handle_mm_fault+0x299/0x3d0
17:40:14: [<ffffffff81538dce>] ? thread_return+0x4e/0x7d0
17:40:14: [<ffffffff8104f156>] ? __do_page_fault+0x146/0x500
17:40:14: [<ffffffff810a677d>] ? hrtimer_try_to_cancel+0x3d/0xd0
17:40:14: [<ffffffff810a6832>] ? hrtimer_cancel+0x22/0x30
17:40:14: [<ffffffff8153b463>] ? do_nanosleep+0x93/0xc0
17:40:14: [<ffffffff810a6904>] ? hrtimer_nanosleep+0xc4/0x180
17:40:14: [<ffffffff810a5730>] ? hrtimer_wakeup+0x0/0x30
17:40:14: [<ffffffff8153f3de>] ? do_page_fault+0x3e/0xa0
17:40:14: [<ffffffff8153c785>] ? page_fault+0x25/0x30

For https://testing.hpdd.intel.com/test_sets/aee14c08-769a-11e5-b71e-5254006e85c2, we see the following from the console on the MDS:

13:02:02:Lustre: 20522:0:(llog_test.c:1443:llog_test_10()) 10c: write 131072 more log records
13:02:02:sendmail invoked oom-killer: gfp_mask=0x201da, order=0, oom_adj=0, oom_score_adj=0
13:02:02:sendmail cpuset=/ mems_allowed=0
13:02:02:Pid: 1827, comm: sendmail Not tainted 2.6.32-573.7.1.el6_lustre.gd359461.x86_64 #1
13:02:02:Call Trace:
13:02:02: [<ffffffff810d71a1>] ? cpuset_print_task_mems_allowed+0x91/0xb0
13:02:02: [<ffffffff8112a9a0>] ? dump_header+0x90/0x1b0
13:02:02: [<ffffffff8112ab0e>] ? check_panic_on_oom+0x4e/0x80
13:02:02: [<ffffffff8112b1fb>] ? out_of_memory+0x1bb/0x3c0
13:02:02: [<ffffffff81137c3c>] ? __alloc_pages_nodemask+0x93c/0x950
13:02:02: [<ffffffff8117064a>] ? alloc_pages_current+0xaa/0x110
13:02:02: [<ffffffff81127d97>] ? __page_cache_alloc+0x87/0x90
13:02:02: [<ffffffff8112777e>] ? find_get_page+0x1e/0xa0
13:02:02: [<ffffffff81128d37>] ? filemap_fault+0x1a7/0x500
13:02:02: [<ffffffff811522f4>] ? __do_fault+0x54/0x530
13:02:02: [<ffffffff811528c7>] ? handle_pte_fault+0xf7/0xb20
13:02:02: [<ffffffff811b4901>] ? save_mount_options+0x21/0x40
13:02:02: [<ffffffff811b7b22>] ? seq_vprintf+0x32/0x60
13:02:02: [<ffffffff81012bbe>] ? copy_user_generic+0xe/0x20
13:02:02: [<ffffffff81153589>] ? handle_mm_fault+0x299/0x3d0
13:02:02: [<ffffffff8104f156>] ? __do_page_fault+0x146/0x500
13:02:02: [<ffffffff81042fc9>] ? kvm_clock_get_cycles+0x9/0x10
13:02:02: [<ffffffff810ad40f>] ? ktime_get_ts+0xbf/0x100
13:02:02: [<ffffffff811a9778>] ? poll_select_copy_remaining+0xf8/0x150
13:02:02: [<ffffffff8153f3de>] ? do_page_fault+0x3e/0xa0
13:02:02: [<ffffffff8153c785>] ? page_fault+0x25/0x30
13:02:02:Mem-Info:
13:02:02:Node 0 DMA per-cpu:
13:02:02:CPU    0: hi:    0, btch:   1 usd:   0
13:02:02:Node 0 DMA32 per-cpu:
13:02:02:CPU    0: hi:  186, btch:  31 usd: 115
13:02:02:active_anon:1685 inactive_anon:1752 isolated_anon:0
13:02:02: active_file:1244 inactive_file:1252 isolated_file:0
13:02:02: unevictable:0 dirty:0 writeback:0 unstable:0
13:02:02: free:13240 slab_reclaimable:2343 slab_unreclaimable:436182
13:02:02: mapped:1 shmem:1 pagetables:676 bounce:0
13:02:02:Node 0 DMA free:8352kB min:332kB low:412kB high:496kB active_anon:0kB inactive_anon:0kB active_file:0kB inactive_file:4kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:15348kB mlocked:0kB dirty:0kB writeback:0kB mapped:0kB shmem:0kB slab_reclaimable:4kB slab_unreclaimable:7384kB kernel_stack:0kB pagetables:0kB unstable:0kB bounce:0kB writeback_tmp:0kB pages_scanned:3 all_unreclaimable? yes
13:02:02:lowmem_reserve[]: 0 2004 2004 2004
13:02:02:Node 0 DMA32 free:44608kB min:44720kB low:55900kB high:67080kB active_anon:6740kB inactive_anon:7008kB active_file:4976kB inactive_file:5004kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:2052308kB mlocked:0kB dirty:0kB writeback:0kB mapped:4kB shmem:4kB slab_reclaimable:9368kB slab_unreclaimable:1737344kB kernel_stack:3200kB pagetables:2704kB unstable:0kB bounce:0kB writeback_tmp:0kB pages_scanned:374310 all_unreclaimable? yes
13:02:02:lowmem_reserve[]: 0 0 0 0
13:02:02:Node 0 DMA: 2*4kB 3*8kB 2*16kB 3*32kB 2*64kB 3*128kB 2*256kB 2*512kB 2*1024kB 2*2048kB 0*4096kB = 8352kB
13:02:02:Node 0 DMA32: 2062*4kB 1079*8kB 497*16kB 232*32kB 79*64kB 11*128kB 1*256kB 1*512kB 1*1024kB 0*2048kB 1*4096kB = 44608kB
13:02:02:4249 total pagecache pages
13:02:02:1752 pages in swap cache
13:02:02:Swap cache stats: add 2908, delete 1156, find 0/0
13:02:02:Free swap  = 4117132kB
13:02:02:Total swap = 4128764kB
13:02:02:524284 pages RAM
13:02:02:43736 pages reserved
13:02:02:2548 pages shared
13:02:02:438391 pages non-shared
13:02:02:[ pid ]   uid  tgid total_vm      rss cpu oom_adj oom_score_adj name
13:02:02:[  482]     0   482     2716       23   0     -17         -1000 udevd
13:02:02:[ 1235]     0  1235     6902       14   0     -17         -1000 auditd
13:02:02:[ 1269]     0  1269    62273      201   0       0             0 rsyslogd
13:02:02:[ 1320]    32  1320     4744        1   0       0             0 rpcbind
13:02:02:[ 1342]    29  1342     5837        1   0       0             0 rpc.statd
13:02:02:[ 1376]    81  1376     6444        1   0       0             0 dbus-daemon
13:02:02:[ 1398]     0  1398    53919        1   0       0             0 ypbind
13:02:02:[ 1430]     0  1430    47233        1   0       0             0 cupsd
13:02:02:[ 1486]     0  1486     1020        0   0       0             0 acpid
13:02:02:[ 1498]    68  1498    10523        1   0       0             0 hald
13:02:02:[ 1499]     0  1499     5099        1   0       0             0 hald-runner
13:02:02:[ 1531]     0  1531     5629        1   0       0             0 hald-addon-inpu
13:02:02:[ 1541]    68  1541     4501        1   0       0             0 hald-addon-acpi
13:02:02:[ 1587]     0  1587    26827        0   0       0             0 rpc.rquotad
13:02:02:[ 1592]     0  1592     5417        0   0       0             0 rpc.mountd
13:02:02:[ 1636]     0  1636     6292        1   0       0             0 rpc.idmapd
13:02:02:[ 1673]   498  1673    57325      139   0       0             0 munged
13:02:02:[ 1784]     0  1784    16555        0   0     -17         -1000 sshd
13:02:02:[ 1795]     0  1795     5429       18   0       0             0 xinetd
13:02:02:[ 1827]     0  1827    22210       23   0       0             0 sendmail
13:02:02:[ 1836]    51  1836    20074       13   0       0             0 sendmail
13:02:02:[ 1864]     0  1864    29216       10   0       0             0 crond
13:02:02:[ 1879]     0  1879     5276        0   0       0             0 atd
13:02:02:[ 1893]     0  1893     1020        2   0       0             0 agetty
13:02:02:[ 1894]     0  1894     1016        2   0       0             0 mingetty
13:02:02:[ 1896]     0  1896     1016       12   0       0             0 mingetty
13:02:02:[ 1898]     0  1898     1016       21   0       0             0 mingetty
13:02:02:[ 1900]     0  1900     1016       22   0       0             0 mingetty
13:02:02:[ 1902]     0  1902     2715       24   0     -17         -1000 udevd
13:02:02:[ 1903]     0  1903     2714       31   0     -17         -1000 udevd
13:02:02:[ 1904]     0  1904     1016       21   0       0             0 mingetty
13:02:02:[ 1906]     0  1906     1016       22   0       0             0 mingetty
13:02:02:[ 2658]    38  2658     7689      154   0       0             0 ntpd
13:02:02:[20420]     0 20420    14749      161   0       0             0 in.mrshd
13:02:02:[20421]     0 20421    26524       59   0       0             0 bash
13:02:02:[20446]     0 20446    26524       59   0       0             0 bash
13:02:02:[20447]     0 20447    27603      617   0       0             0 sh
13:02:02:[20522]     0 20522    28026       65   0       0             0 lctl
13:02:02:Kernel panic - not syncing: Out of memory: system-wide panic_on_oom is enabled
13:02:02:
13:02:02:Pid: 1827, comm: sendmail Not tainted 2.6.32-573.7.1.el6_lustre.gd359461.x86_64 #1
13:02:02:Call Trace:
13:02:02: [<ffffffff815386d4>] ? panic+0xa7/0x16f
13:02:02: [<ffffffff8112aaa1>] ? dump_header+0x191/0x1b0
13:02:02: [<ffffffff8112ab3c>] ? check_panic_on_oom+0x7c/0x80
13:02:02: [<ffffffff8112b1fb>] ? out_of_memory+0x1bb/0x3c0
13:02:02: [<ffffffff81137c3c>] ? __alloc_pages_nodemask+0x93c/0x950
13:02:02: [<ffffffff8117064a>] ? alloc_pages_current+0xaa/0x110
13:02:02: [<ffffffff81127d97>] ? __page_cache_alloc+0x87/0x90
13:02:02: [<ffffffff8112777e>] ? find_get_page+0x1e/0xa0
13:02:02: [<ffffffff81128d37>] ? filemap_fault+0x1a7/0x500
13:02:02: [<ffffffff811522f4>] ? __do_fault+0x54/0x530
13:02:02: [<ffffffff811528c7>] ? handle_pte_fault+0xf7/0xb20
13:02:02: [<ffffffff811b4901>] ? save_mount_options+0x21/0x40
13:02:02: [<ffffffff811b7b22>] ? seq_vprintf+0x32/0x60
13:02:02: [<ffffffff81012bbe>] ? copy_user_generic+0xe/0x20
13:02:02: [<ffffffff81153589>] ? handle_mm_fault+0x299/0x3d0
13:02:02: [<ffffffff8104f156>] ? __do_page_fault+0x146/0x500
13:02:02: [<ffffffff81042fc9>] ? kvm_clock_get_cycles+0x9/0x10
13:02:02: [<ffffffff810ad40f>] ? ktime_get_ts+0xbf/0x100
13:02:02: [<ffffffff811a9778>] ? poll_select_copy_remaining+0xf8/0x150
13:02:02: [<ffffffff8153f3de>] ? do_page_fault+0x3e/0xa0
13:02:02: [<ffffffff8153c785>] ? page_fault+0x25/0x30

For https://testing.hpdd.intel.com/test_sets/4790f2ba-767c-11e5-86b7-5254006e85c2, we see the following from the console on the MDS:

08:00:56:Lustre: 20445:0:(llog_test.c:1443:llog_test_10()) 10c: write 131072 more log records
08:00:56:sendmail invoked oom-killer: gfp_mask=0x201da, order=0, oom_adj=0, oom_score_adj=0
08:00:56:sendmail cpuset=/ mems_allowed=0
08:00:56:Pid: 1875, comm: sendmail Not tainted 2.6.32-573.7.1.el6_lustre.gd359461.x86_64 #1
08:00:56:Call Trace:
08:00:56: [<ffffffff810d71a1>] ? cpuset_print_task_mems_allowed+0x91/0xb0
08:00:56: [<ffffffff8112a9a0>] ? dump_header+0x90/0x1b0
08:00:56: [<ffffffff8112ab0e>] ? check_panic_on_oom+0x4e/0x80
08:00:56: [<ffffffff8112b1fb>] ? out_of_memory+0x1bb/0x3c0
08:00:56: [<ffffffff81137c3c>] ? __alloc_pages_nodemask+0x93c/0x950
08:00:56: [<ffffffff8117064a>] ? alloc_pages_current+0xaa/0x110
08:00:56: [<ffffffff81127d97>] ? __page_cache_alloc+0x87/0x90
08:00:56: [<ffffffff8112777e>] ? find_get_page+0x1e/0xa0
08:00:56: [<ffffffff81128d37>] ? filemap_fault+0x1a7/0x500
08:00:56: [<ffffffff811522f4>] ? __do_fault+0x54/0x530
08:00:56: [<ffffffff811528c7>] ? handle_pte_fault+0xf7/0xb20
08:00:56: [<ffffffff811b4901>] ? save_mount_options+0x21/0x40
08:00:56: [<ffffffff811b7b22>] ? seq_vprintf+0x32/0x60
08:00:56: [<ffffffff81012bbe>] ? copy_user_generic+0xe/0x20
08:00:56: [<ffffffff81153589>] ? handle_mm_fault+0x299/0x3d0
08:00:56: [<ffffffff8104f156>] ? __do_page_fault+0x146/0x500
08:00:56: [<ffffffff81042fc9>] ? kvm_clock_get_cycles+0x9/0x10
08:00:56: [<ffffffff810ad40f>] ? ktime_get_ts+0xbf/0x100
08:00:56: [<ffffffff811a9778>] ? poll_select_copy_remaining+0xf8/0x150
08:00:56: [<ffffffff8153f3de>] ? do_page_fault+0x3e/0xa0
08:00:56: [<ffffffff8153c785>] ? page_fault+0x25/0x30
08:00:56:Mem-Info:
08:00:56:Node 0 DMA per-cpu:
08:00:56:CPU    0: hi:    0, btch:   1 usd:   0
08:00:56:CPU    1: hi:    0, btch:   1 usd:   0
08:00:56:Node 0 DMA32 per-cpu:
08:00:56:CPU    0: hi:  186, btch:  31 usd:  94
08:00:56:CPU    1: hi:  186, btch:  31 usd:   0
08:00:56:active_anon:2321 inactive_anon:2319 isolated_anon:0
08:00:56: active_file:1224 inactive_file:1294 isolated_file:0
08:00:56: unevictable:0 dirty:13 writeback:2015 unstable:0
08:00:56: free:13240 slab_reclaimable:2534 slab_unreclaimable:434564
08:00:56: mapped:18 shmem:14 pagetables:709 bounce:0
08:00:56:Node 0 DMA free:8336kB min:332kB low:412kB high:496kB active_anon:0kB inactive_anon:0kB active_file:0kB inactive_file:0kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:15348kB mlocked:0kB dirty:0kB writeback:0kB mapped:0kB shmem:0kB slab_reclaimable:4kB slab_unreclaimable:7328kB kernel_stack:0kB pagetables:0kB unstable:0kB bounce:0kB writeback_tmp:0kB pages_scanned:1 all_unreclaimable? yes
08:00:56:lowmem_reserve[]: 0 2004 2004 2004
08:00:56:Node 0 DMA32 free:44624kB min:44720kB low:55900kB high:67080kB active_anon:9284kB inactive_anon:9276kB active_file:4896kB inactive_file:5176kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:2052308kB mlocked:0kB dirty:52kB writeback:8060kB mapped:72kB shmem:56kB slab_reclaimable:10132kB slab_unreclaimable:1730928kB kernel_stack:3680kB pagetables:2836kB unstable:0kB bounce:0kB writeback_tmp:0kB pages_scanned:300165 all_unreclaimable? yes
08:00:56:lowmem_reserve[]: 0 0 0 0
08:00:56:Node 0 DMA: 2*4kB 2*8kB 2*16kB 2*32kB 3*64kB 3*128kB 2*256kB 2*512kB 2*1024kB 2*2048kB 0*4096kB = 8376kB
08:00:56:Node 0 DMA32: 2124*4kB 1102*8kB 573*16kB 241*32kB 57*64kB 7*128kB 1*256kB 1*512kB 1*1024kB 0*2048kB 1*4096kB = 44624kB
08:00:56:4866 total pagecache pages
08:00:56:2340 pages in swap cache
08:00:56:Swap cache stats: add 2340, delete 0, find 0/0
08:00:56:Free swap  = 4119404kB
08:00:56:Total swap = 4128764kB
08:00:56:524284 pages RAM
08:00:56:43736 pages reserved
08:00:56:2566 pages shared
08:00:56:438172 pages non-shared
08:00:56:[ pid ]   uid  tgid total_vm      rss cpu oom_adj oom_score_adj name
08:00:56:[  506]     0   506     2708       25   0     -17         -1000 udevd
08:00:56:[ 1281]     0  1281     6902       13   0     -17         -1000 auditd
08:00:56:[ 1315]     0  1315    62272      176   1       0             0 rsyslogd
08:00:56:[ 1349]     0  1349     4561        1   1       0             0 irqbalance
08:00:56:[ 1367]    32  1367     4744        1   0       0             0 rpcbind
08:00:56:[ 1389]    29  1389     5837        1   1       0             0 rpc.statd
08:00:56:[ 1423]    81  1423     6418        1   0       0             0 dbus-daemon
08:00:56:[ 1443]     0  1443    53919        1   1       0             0 ypbind
08:00:56:[ 1477]     0  1477    47233        1   1       0             0 cupsd
08:00:56:[ 1534]     0  1534     1020        0   1       0             0 acpid
08:00:56:[ 1546]    68  1546    10517        1   1       0             0 hald
08:00:56:[ 1547]     0  1547     5099        1   0       0             0 hald-runner
08:00:56:[ 1579]     0  1579     5629        1   1       0             0 hald-addon-inpu
08:00:56:[ 1589]    68  1589     4501        1   1       0             0 hald-addon-acpi
08:00:56:[ 1635]     0  1635    26827        0   0       0             0 rpc.rquotad
08:00:56:[ 1640]     0  1640     5417        0   0       0             0 rpc.mountd
08:00:56:[ 1684]     0  1684     6292        1   0       0             0 rpc.idmapd
08:00:56:[ 1721]   498  1721    57325      139   0       0             0 munged
08:00:56:[ 1832]     0  1832    16555        0   0     -17         -1000 sshd
08:00:56:[ 1843]     0  1843     5429       17   1       0             0 xinetd
08:00:56:[ 1875]     0  1875    22211       31   0       0             0 sendmail
08:00:56:[ 1884]    51  1884    20074      365   0       0             0 sendmail
08:00:56:[ 1912]     0  1912    29215      151   1       0             0 crond
08:00:56:[ 1927]     0  1927     5276       45   0       0             0 atd
08:00:56:[ 1940]     0  1940     1020       22   1       0             0 agetty
08:00:56:[ 1942]     0  1942     1016       21   1       0             0 mingetty
08:00:56:[ 1944]     0  1944     1016       21   1       0             0 mingetty
08:00:56:[ 1946]     0  1946     1016       21   1       0             0 mingetty
08:00:56:[ 1948]     0  1948     1016       22   1       0             0 mingetty
08:00:56:[ 1951]     0  1951     2710       35   0     -17         -1000 udevd
08:00:56:[ 1952]     0  1952     2708       32   0     -17         -1000 udevd
08:00:56:[ 1953]     0  1953     1016       21   0       0             0 mingetty
08:00:56:[ 1955]     0  1955     1016       21   0       0             0 mingetty
08:00:56:[ 2711]    38  2711     7689      156   1       0             0 ntpd
08:00:56:[21148]     0 21148     4763       55   1       0             0 anacron
08:00:56:[20343]     0 20343    14749      161   0       0             0 in.mrshd
08:00:56:[20344]     0 20344    26524       58   0       0             0 bash
08:00:56:[20369]     0 20369    26524       58   1       0             0 bash
08:00:56:[20370]     0 20370    27603      619   0       0             0 sh
08:00:56:[20445]     0 20445    28026       65   1       0             0 lctl
08:00:56:Kernel panic - not syncing: Out of memory: system-wide panic_on_oom is enabled
08:00:56:
08:00:56:Pid: 1875, comm: sendmail Not tainted 2.6.32-573.7.1.el6_lustre.gd359461.x86_64 #1
08:00:56:Call Trace:
08:00:56: [<ffffffff815386d4>] ? panic+0xa7/0x16f
08:00:56: [<ffffffff8112aaa1>] ? dump_header+0x191/0x1b0
08:00:56: [<ffffffff8112ab3c>] ? check_panic_on_oom+0x7c/0x80
08:00:56: [<ffffffff8112b1fb>] ? out_of_memory+0x1bb/0x3c0
08:00:56: [<ffffffff81137c3c>] ? __alloc_pages_nodemask+0x93c/0x950
08:00:56: [<ffffffff8117064a>] ? alloc_pages_current+0xaa/0x110
08:00:56: [<ffffffff81127d97>] ? __page_cache_alloc+0x87/0x90
08:00:56: [<ffffffff8112777e>] ? find_get_page+0x1e/0xa0
08:00:56: [<ffffffff81128d37>] ? filemap_fault+0x1a7/0x500
08:00:56: [<ffffffff811522f4>] ? __do_fault+0x54/0x530
08:00:56: [<ffffffff811528c7>] ? handle_pte_fault+0xf7/0xb20
08:00:56: [<ffffffff811b4901>] ? save_mount_options+0x21/0x40
08:00:56: [<ffffffff811b7b22>] ? seq_vprintf+0x32/0x60
08:00:56: [<ffffffff81012bbe>] ? copy_user_generic+0xe/0x20
08:00:56: [<ffffffff81153589>] ? handle_mm_fault+0x299/0x3d0
08:00:56: [<ffffffff8104f156>] ? __do_page_fault+0x146/0x500
08:00:56: [<ffffffff81042fc9>] ? kvm_clock_get_cycles+0x9/0x10
08:00:56: [<ffffffff810ad40f>] ? ktime_get_ts+0xbf/0x100
08:00:56: [<ffffffff811a9778>] ? poll_select_copy_remaining+0xf8/0x150
08:00:56: [<ffffffff8153f3de>] ? do_page_fault+0x3e/0xa0
08:00:56: [<ffffffff8153c785>] ? page_fault+0x25/0x30


 Comments   
Comment by James Nunez (Inactive) [ 22/Oct/15 ]

Another failure (ppc): https://testing.hpdd.intel.com/test_sets/82b87298-78f0-11e5-9aa5-5254006e85c2

Comment by Andreas Dilger [ 22/Oct/15 ]

I don't think the actual process being OOM killed matters, since there are relatively few user-space processes on the MDS, and one will be selected at random to kill, since Lustre kernel threads cannot be killed.

I suspect this is related to Bruno's recent patch to add llog wrap testing to llog_test.c, http://review.whamcloud.com/14912 "LU-6556 obdclass: re-allow catalog to wrap around".

Comment by Bruno Faccini (Inactive) [ 23/Oct/15 ]

Strange is that the whole memory seems to be consumed as slab_unreclaimable !
Will have a look to the crash-dump of one of these autotests sessions, if available, and try to reproduce if not.

Comment by Bob Glossman (Inactive) [ 23/Oct/15 ]

another seen with sles11sp3 on master:
https://testing.hpdd.intel.com/test_sets/9b7b7a52-793e-11e5-80c8-5254006e85c2

Comment by Bruno Faccini (Inactive) [ 23/Oct/15 ]

Ok, found a crash-dump from https://testing.hpdd.intel.com/test_sets/4790f2ba-767c-11e5-86b7-5254006e85c2 failed session reported by James.
MGS/MDS is running in a VM of 2GB memory, and in crash-dump I find that there are 1.7GB consumed in Slabs, and the main one is "size-512", itself consuming:

CACHE                NAME          OBJSIZE  ALLOCATED     TOTAL  SLABS  SSIZE
ffff88007f8b02c0 size-512          512               2415216   2415272 301909     4k

Interesting is that in the 2 active threads on the 2 virtual cores :

PID: 1875   TASK: ffff880037497520  CPU: 0   COMMAND: "sendmail"
 #0 [ffff88007dbf77a0] machine_kexec at ffffffff8103d1ab
 #1 [ffff88007dbf7800] crash_kexec at ffffffff810cc8d2
 #2 [ffff88007dbf78d0] panic at ffffffff815386db
 #3 [ffff88007dbf7950] check_panic_on_oom at ffffffff8112ab3c
 #4 [ffff88007dbf7980] out_of_memory at ffffffff8112b1fb
 #5 [ffff88007dbf7a00] __alloc_pages_nodemask at ffffffff81137c3c
 #6 [ffff88007dbf7b50] alloc_pages_current at ffffffff8117064a
 #7 [ffff88007dbf7b80] __page_cache_alloc at ffffffff81127d97
 #8 [ffff88007dbf7bb0] filemap_fault at ffffffff81128d37
 #9 [ffff88007dbf7c20] __do_fault at ffffffff811522f4
#10 [ffff88007dbf7cb0] handle_pte_fault at ffffffff811528c7
#11 [ffff88007dbf7d90] handle_mm_fault at ffffffff81153589
#12 [ffff88007dbf7e00] __do_page_fault at ffffffff8104f156
#13 [ffff88007dbf7f20] do_page_fault at ffffffff8153f3de
#14 [ffff88007dbf7f50] page_fault at ffffffff8153c785
    RIP: 00007f1d296db393  RSP: 00007ffe4a334528  RFLAGS: 00010246
    RAX: 0000000000000000  RBX: 0000000000000000  RCX: 00007f1d296db393
    RDX: 0000000000000000  RSI: 00007ffe4a334620  RDI: 0000000000000005
    RBP: 0000000000000001   R8: 00007ffe4a3346a0   R9: 00007ffe4a338efc
    R10: 0000000000000000  R11: 0000000000000246  R12: 00007f1d2b9b9fb0
    R13: 0000000000000001  R14: 0000000000000004  R15: 00007f1d2b977b28
    ORIG_RAX: ffffffffffffffff  CS: 0033  SS: 002b

PID: 20445  TASK: ffff880077389520  CPU: 1   COMMAND: "lctl"
 #0 [ffff880002306e90] crash_nmi_callback at ffffffff81033cf6
 #1 [ffff880002306ea0] notifier_call_chain at ffffffff8153f495
 #2 [ffff880002306ee0] atomic_notifier_call_chain at ffffffff8153f4fa
 #3 [ffff880002306ef0] notify_die at ffffffff810a783e
 #4 [ffff880002306f20] do_nmi at ffffffff8153d153
 #5 [ffff880002306f50] nmi at ffffffff8153ca10
    [exception RIP: idr_get_next+108]
    RIP: ffffffff812911bc  RSP: ffff8800656af388  RFLAGS: 00000206
    RAX: 0000000000000000  RBX: ffffffff81c1ab68  RCX: 0000000000000000
    RDX: ffff8800656af390  RSI: ffff8800656af3e4  RDI: 0000000000000040
    RBP: ffff8800656af3b8   R8: 0000000000000012   R9: 00000000ffffffff
    R10: 0000000000000001  R11: 0000000000000012  R12: 0000000000000000
    R13: ffff8800656af3e4  R14: ffffffff81c1ab48  R15: 0000000000000001
    ORIG_RAX: ffffffffffffffff  CS: 0010  SS: 0000
--- <DOUBLEFAULT exception stack> ---
 #6 [ffff8800656af388] idr_get_next at ffffffff812911bc
 #7 [ffff8800656af3c0] css_get_next at ffffffff810d2077
 #8 [ffff8800656af420] mem_cgroup_iter at ffffffff811837ad
 #9 [ffff8800656af470] shrink_zone at ffffffff81142bc8
#10 [ffff8800656af4f0] do_try_to_free_pages at ffffffff81142de5
#11 [ffff8800656af590] try_to_free_pages at ffffffff811434a2
#12 [ffff8800656af630] __alloc_pages_nodemask at ffffffff811377cd
#13 [ffff8800656af780] kmem_getpages at ffffffff81177282
#14 [ffff8800656af7b0] fallback_alloc at ffffffff81177e9a
#15 [ffff8800656af830] ____cache_alloc_node at ffffffff81177c19
#16 [ffff8800656af890] kmem_cache_alloc_trace at ffffffff81178517
#17 [ffff8800656af8e0] lu_context_init at ffffffffa05f1efe [obdclass]
#18 [ffff8800656af910] osd_trans_start at ffffffffa0d108ee [osd_ldiskfs]
#19 [ffff8800656af950] llog_cat_add at ffffffffa05abb2d [obdclass]
#20 [ffff8800656af9a0] llog_test_10 at ffffffffa024910a [llog_test]
#21 [ffff8800656afa90] llog_run_tests at ffffffffa024da62 [llog_test]
#22 [ffff8800656afb10] llog_test_setup at ffffffffa024e499 [llog_test]
#23 [ffff8800656afbd0] obd_setup at ffffffffa05d9f8b [obdclass]
#24 [ffff8800656afc90] class_setup at ffffffffa05da2d9 [obdclass]
#25 [ffff8800656afce0] class_process_config at ffffffffa05e1481 [obdclass]
#26 [ffff8800656afd90] class_handle_ioctl at ffffffffa05b9cf2 [obdclass]
#27 [ffff8800656afe40] obd_class_ioctl at ffffffffa059f2ab [obdclass]
#28 [ffff8800656afe60] vfs_ioctl at ffffffff811a77a2
#29 [ffff8800656afea0] do_vfs_ioctl at ffffffff811a7944
#30 [ffff8800656aff30] sys_ioctl at ffffffff811a7ec1
#31 [ffff8800656aff80] system_call_fastpath at ffffffff8100b0d2
    RIP: 00000036680e0907  RSP: 00007ffd31d15f58  RFLAGS: 00010246
    RAX: 0000000000000010  RBX: ffffffff8100b0d2  RCX: 0000000000000000
    RDX: 00007ffd31d13c40  RSI: 00000000c00866b8  RDI: 0000000000000003
    RBP: 0000000000000001   R8: 0000000000000240   R9: 0000000000000240
    R10: 0000000000000040  R11: 0000000000000246  R12: 00000000c00866b8
    R13: 00007ffd31d13c40  R14: 000000000064f980  R15: 0000000000000003
    ORIG_RAX: 0000000000000010  CS: 0033  SS: 002b

one is the lctl cmd that started the in-kernel LLOG tests, which is trying to allocate a new object in the "size-512" Slab (requested size is 320 bytes) and the concerned code is :

static int keys_init(struct lu_context *ctx)
{
>>>  OBD_ALLOC(ctx->lc_value, ARRAY_SIZE(lu_keys) * sizeof ctx->lc_value[0]);
        if (likely(ctx->lc_value != NULL))
                return keys_fill(ctx);
         
        return -ENOMEM;
}        

These allocated things, should be freed in keys_fini() routine, itself being called by lu_context_fini(), to be called by osd_trans_commit_cb(). This last one is being run as a callback posted via ldiskfs_journal_callback_add() in osd_trans_stop().
Checking the number of currently queued of these callbacks in the crash-dump, I found that there are 1206674!!!

So seems that something has gone wrong/delayed in the journaling process here, and I wonder if this could not be due to too much work being done in a raw and in kernel space by llog_test() code, aggravated with my recent add-on of test_10() sub-test for LU-6556 fix verifications...

What is unclear for me now, is how/when these callbacks are executed by [ext4,ldiskfs]_journal_commit_callback() ??

Comment by Alex Zhuravlev [ 23/Oct/15 ]

this might be due to lots of transactions in the transaction groups. though 241K/sec looks very serious.. I've got a patch (still under development) which allows to get rid of per-tx context: http://review.whamcloud.com/#/c/4307/

Comment by Alex Zhuravlev [ 23/Oct/15 ]

> What is unclear for me now, is how/when these callbacks are executed by [ext4,ldiskfs]_journal_commit_callback() ??

it's called when the transaction group holding this transaction is committed. it's done every 5s roughly.

Comment by Bruno Faccini (Inactive) [ 23/Oct/15 ]

Ok, this is kjournald2 thread's job to execute these callbacks, via [ext4,ldiskfs]_journal_commit_callback() set in j_commit_callback field, and to be called by jbd2_journal_commit_transaction().

But according to crash-dump content the kjournald2 thread associated to the MGS/MDT device has not been scheduled since about 11s, so may be we just need to add some tempos in llog_test() code to allow the thread to be scheduled and run the callbacks?

Comment by Alex Zhuravlev [ 23/Oct/15 ]

no, this shouldn't be needed - JBD does this internally in journal_start() unless there was an open transaction. if the transaction group becomes very large or when running time is over (5s by default), journal_start() blocks and essentially passes control to the commit thread.

Comment by Bruno Faccini (Inactive) [ 23/Oct/15 ]

Ok, but then how can we explain that the kjournald2/commit thread has not been running since about 11s after calling schedule() ?
Don't you think we may face some scheduling issue inside a VM, where the thread running llog_test() will post a huge number of callbacks in a raw when the commit thread will not be allowed to run ?
Why not just add frequent schedule() calls, instead of delays, in llog_test() code to help for callbacks to be drained ?

Comment by nasf (Inactive) [ 25/Oct/15 ]

Another failure instance on master:
https://testing.hpdd.intel.com/test_sets/cd830060-7aba-11e5-b618-5254006e85c2

Comment by Alex Zhuravlev [ 25/Oct/15 ]

we do not control commit thread directly. llog_test() generates new transactions and at some of them JBD will block everything and commit thread get control. plus we're doing lots of allocations where the kernel can schedule (especially when we're low on memory). IOW, I doubt schedule() can help here.

Comment by James Nunez (Inactive) [ 26/Oct/15 ]

Another failure at https://testing.hpdd.intel.com/test_sets/04224472-7b2a-11e5-9650-5254006e85c2

Comment by Bob Glossman (Inactive) [ 26/Oct/15 ]

another on master:
https://testing.hpdd.intel.com/test_sessions/c0a745da-7c19-11e5-9ca1-5254006e85c2

Comment by nasf (Inactive) [ 28/Oct/15 ]

More failures on master:
https://testing.hpdd.intel.com/test_sets/f6566f6a-7cc0-11e5-ba8b-5254006e85c2

Comment by James Nunez (Inactive) [ 28/Oct/15 ]

More failures on master:
https://testing.hpdd.intel.com/test_sets/35483020-7c79-11e5-ba8b-5254006e85c2

Comment by James Nunez (Inactive) [ 29/Oct/15 ]

This test is failing about twice a day. Raised priority to Critical.

More failures:
https://testing.hpdd.intel.com/test_sets/2c0ea00c-7e13-11e5-9c23-5254006e85c2
https://testing.hpdd.intel.com/test_sets/cf2f6532-7e22-11e5-ae2b-5254006e85c2
https://testing.hpdd.intel.com/test_sets/cdee61c8-7db9-11e5-82ee-5254006e85c2
https://testing.hpdd.intel.com/test_sets/cc130ef8-7d91-11e5-8bcc-5254006e85c2
2015-11-01 19:42:15 - https://testing.hpdd.intel.com/test_sets/852de9f4-80ed-11e5-b644-5254006e85c2
2015-11-02 09:51:21 - https://testing.hpdd.intel.com/test_sets/d1b550dc-8192-11e5-9c69-5254006e85c2

Comment by nasf (Inactive) [ 03/Nov/15 ]

Another failure instance on master:
https://testing.hpdd.intel.com/test_sets/4f309ed0-81c6-11e5-af7b-5254006e85c2

Comment by Bob Glossman (Inactive) [ 03/Nov/15 ]

another on master:
https://testing.hpdd.intel.com/test_sets/04310f4a-821c-11e5-aa47-5254006e85c2

Comment by Gerrit Updater [ 05/Nov/15 ]

Faccini Bruno (bruno.faccini@intel.com) uploaded a new patch: http://review.whamcloud.com/17052
Subject: LU-7329 obdclass: sync device to flush journal callbacks
Project: fs/lustre-release
Branch: master
Current Patch Set: 1
Commit: 3fd3976139838bf60c87bc44be8996ce13b3a1e3

Comment by Bruno Faccini (Inactive) [ 05/Nov/15 ]

Patch #17052 is an attempt to sync LLOG device frequently in order to force flushing of journal commit callbacks and avoid OOM situations on VMs of auto-test.

Comment by Alex Zhuravlev [ 05/Nov/15 ]

i still don't understand.. i'm pretty sure that we don't hit anything like this running something silly like while true; do mkdir /mnt/lustre/d0; rmdir /mnt/lustre/d0; done

Comment by Jian Yu [ 05/Nov/15 ]

One more instance on master branch:
https://testing.hpdd.intel.com/test_sets/9c26010e-838c-11e5-a390-5254006e85c2

Comment by James Nunez (Inactive) [ 10/Nov/15 ]

Another failure on master
2015-11-09 21:05:46 - https://testing.hpdd.intel.com/test_sets/b91c2110-8767-11e5-ba88-5254006e85c2
2015-11-11 20:02:30 - https://testing.hpdd.intel.com/test_sets/04d585fe-8927-11e5-9de3-5254006e85c2
2015-11-12 01:20:50 - https://testing.hpdd.intel.com/test_sets/57ec926a-88fe-11e5-8174-5254006e85c2
2015-11-12 17:09:10 - https://testing.hpdd.intel.com/test_sets/4b8642fc-897e-11e5-9de3-5254006e85c2
2015-11-13 14:49:02 - https://testing.hpdd.intel.com/test_sets/6105fd0a-8a5c-11e5-ba42-5254006e85c2

Comment by Gerrit Updater [ 30/Nov/15 ]

Oleg Drokin (oleg.drokin@intel.com) merged in patch http://review.whamcloud.com/17052/
Subject: LU-7329 obdclass: sync device to flush journal callbacks
Project: fs/lustre-release
Branch: master
Current Patch Set:
Commit: 795e13aeed68c08a72884c7db9df35b114d972b1

Comment by Peter Jones [ 30/Nov/15 ]

Landed for 2.8

Comment by Gerrit Updater [ 18/Dec/15 ]

Andreas Dilger (andreas.dilger@intel.com) uploaded a new patch: http://review.whamcloud.com/17670
Subject: LU-7329 tests: increase sanity test_60b llog_test line count
Project: fs/lustre-release
Branch: master
Current Patch Set: 1
Commit: bcfb2ca8fe2a62d6a5ea3c722e3e314e6ed27a4f

Comment by Gerrit Updater [ 25/Jan/16 ]

Oleg Drokin (oleg.drokin@intel.com) merged in patch http://review.whamcloud.com/17670/
Subject: LU-7329 tests: increase sanity test_60b llog_test line count
Project: fs/lustre-release
Branch: master
Current Patch Set:
Commit: 03ee31f594d6c70b3c08a0910c16403591bb99d1

Generated at Sat Feb 10 02:07:57 UTC 2024 using Jira 9.4.14#940014-sha1:734e6822bbf0d45eff9af51f82432957f73aa32c.