[LU-10134] LBUG lfsck_namespace_double_scan()) ASSERTION( list_empty(&lad->lad_req_list) ) failed:

• Type: Bug
• Resolution: Fixed
• Priority: Critical
• Fix Version/s: Lustre 2.11.0, Lustre 2.10.3
• Affects Version/s: Lustre 2.11.0, Lustre 2.10.2
• Labels: soak
• Environment: Soak cluster, lustre-master build 3654
• Severity: 3
• Assignee: nasf (Inactive)
• Reporter: Cliff White (Inactive)

Description

Soak mds_failover test run, soak-9 powered off, MDT0001 failed over to soak-8

2017-10-17 22:18:47,423:fsmgmt.fsmgmt:INFO     Mounting soaked-MDT0001 on soak-8
Soak-8 log:
Oct 17 22:19:43 soak-8 kernel: Lustre: soaked-MDT0001: Will be in recovery for at least 2:30, or until 36 clients reconnect
Oct 17 22:19:51 soak-8 kernel: Lustre: soaked-MDT0001: Connection restored to 7553a398-a79b-c096-e685-e7284e2b17df (at 172.16.1.47@o2ib1)

Recovery completes, system hits LBUG and dies:

Oct 17 22:21:03 soak-8 kernel: Lustre: soaked-MDT0001: Recovery over after 1:20, of 36 clients 36 recovered and 0 were evicted.
Oct 17 22:21:08 soak-8 kernel: LustreError: 3745:0:(lfsck_namespace.c:4571:lfsck_namespace_double_scan()) ASSERTION( list_empty(&lad->lad_req_list) ) failed:
Oct 17 22:21:08 soak-8 kernel: LustreError: 3745:0:(lfsck_namespace.c:4571:lfsck_namespace_double_scan()) LBUG
Oct 17 22:21:08 soak-8 kernel: Pid: 3745, comm: lfsck
Oct 17 22:21:08 soak-8 kernel: #012Call Trace:
Oct 17 22:21:08 soak-8 kernel: [<ffffffffc0dc37ae>] libcfs_call_trace+0x4e/0x60 [libcfs]
Oct 17 22:21:08 soak-8 kernel: [<ffffffffc0dc383c>] lbug_with_loc+0x4c/0xb0 [libcfs]
Oct 17 22:21:08 soak-8 kernel: [<ffffffffc14ef398>] lfsck_namespace_double_scan+0x108/0x140 [lfsck]
Oct 17 22:21:08 soak-8 kernel: [<ffffffffc14e65a9>] lfsck_double_scan+0x59/0x200 [lfsck]
Oct 17 22:21:08 soak-8 kernel: [<ffffffffc143550a>] ? osd_otable_it_fini+0xca/0x240 [osd_ldiskfs]
Oct 17 22:21:09 soak-8 kernel: [<ffffffff811deec3>] ? kfree+0x103/0x140
Oct 17 22:21:09 soak-8 kernel: [<ffffffffc14eb134>] lfsck_master_engine+0x494/0x12b0 [lfsck]
Oct 17 22:21:09 soak-8 kernel: [<ffffffff810c4810>] ? default_wake_function+0x0/0x20
Oct 17 22:21:09 soak-8 kernel: [<ffffffffc14eaca0>] ? lfsck_master_engine+0x0/0x12b0 [lfsck]
Oct 17 22:21:09 soak-8 kernel: [<ffffffff810b098f>] kthread+0xcf/0xe0
Oct 17 22:21:09 soak-8 kernel: [<ffffffff810b08c0>] ? kthread+0x0/0xe0
Oct 17 22:21:09 soak-8 kernel: [<ffffffff816b4f18>] ret_from_fork+0x58/0x90
Oct 17 22:21:09 soak-8 kernel: [<ffffffff810b08c0>] ? kthread+0x0/0xe0
Oct 17 22:21:09 soak-8 kernel:
Oct 17 22:21:09 soak-8 kernel: Kernel panic - not syncing: LBUG

Crash dump from soak-8 is available on spirit cluster at: /scratch/dumps/soak-8.spirit.hpdd.intel.com/10.10.1.108-2017-10-17-22:21:34
vmcore-dmesg.txt attached

Issue Links

• is related to: LU-8647 lfsck_namespace_double_scan()) ASSERTION( list_empty(&lad->lad_req_list) ) failed

Comments

[Gerrit Updater]

John L. Hammond ([e-mail]) merged in patch https://review.whamcloud.com/30375/
Subject: LU-10134 lfsck: not add requests if engine out of work
Project: fs/lustre-release
Branch: b2_10
Current Patch Set:
Commit: 02065756c83057d3acc35792b744078f2739025d

[Peter Jones]

Landed for 2.11

[Gerrit Updater]

Oleg Drokin ([e-mail]) merged in patch https://review.whamcloud.com/30165/
Subject: LU-10134 lfsck: not add requests if engine out of work
Project: fs/lustre-release
Branch: master
Current Patch Set:
Commit: f22a0ab6c37db2d983451ec01e869ed8d3226cb2

[James Nunez (Inactive)]

I ported Fan Yong's patch to b2_10 to see if this will help with the issues seen on the soak cluster.

[Gerrit Updater]

James Nunez ([e-mail]) uploaded a new patch: https://review.whamcloud.com/30375
Subject: LU-10134 lfsck: not add requests if engine out of work
Project: fs/lustre-release
Branch: b2_10
Current Patch Set: 1
Commit: 90df962c58fb18367291dfff8698fdb1e70d9188

[Cliff White (Inactive)]

After the first assert hits, attempting to restart the systems results in further asserts. Can't start soak at this time, due to this. 3 out of our 4 MDS are now halted.

soak-10.log:Dec  5 16:44:33 soak-10 kernel: LustreError: 2593:0:(lfsck_namespace.c:4582:lfsck_namespace_double_scan()) LBUG
soak-10.log:Dec  5 16:44:34 soak-10 kernel: Kernel panic - not syncing: LBUG
soak-8.log:Dec  5 16:44:31 soak-8 kernel: LustreError: 2529:0:(lfsck_namespace.c:4582:lfsck_namespace_double_scan()) LBUG
soak-8.log:Dec  5 16:44:32 soak-8 kernel: Kernel panic - not syncing: LBUG
soak-9.log:Dec  5 00:32:11 soak-9 kernel: LustreError: 2796:0:(lfsck_namespace.c:4582:lfsck_namespace_double_scan()) LBUG
soak-9.log:Dec  5 00:32:11 soak-9 kernel: Kernel panic - not syncing: LBUG

[Cliff White (Inactive)]

We just hit this assert on b2.10.2-RC1, crash dump is available on Spirit

[ 1256.526444] Lustre: Failing over soaked-MDT0000
[ 1256.536364] LustreError: 2796:0:(lfsck_namespace.c:4582:lfsck_namespace_double_scan()) ASSERTION( list_empty(&lad->lad_req_list) ) failed:
[ 1256.537017] LustreError: 2303:0:(ldlm_lockd.c:1415:ldlm_handle_enqueue0()) ### lock on destroyed export ffff88080e0d1800 ns: mdt-soaked-MDT0000_UUID lock: ffff880811089600/0x18106beaab642978 lrc: 3/0,0 mode: PR/PR res: [0x200000405:0x1:0x0].0x0 bits 0x13 rrc: 2 type: IBT flags: 0x50200000000000 nid: 192.168.1.117@o2ib remote: 0xd36fb2fda127ff7 expref: 3 pid: 2303 timeout: 0 lvb_type: 0
[ 1256.580979] Lustre: soaked-MDT0000: Not available for connect from 192.168.1.117@o2ib (stopping)
[ 1256.614544] LustreError: 2796:0:(lfsck_namespace.c:4582:lfsck_namespace_double_scan()) LBUG
[ 1256.625905] Pid: 2796, comm: lfsck
[ 1256.631630]
Call Trace:
[ 1256.639771] [<ffffffffc0dd57ae>] libcfs_call_trace+0x4e/0x60 [libcfs]
[ 1256.648850] [<ffffffffc0dd583c>] lbug_with_loc+0x4c/0xb0 [libcfs]
[ 1256.657599] [<ffffffffc14964c8>] lfsck_namespace_double_scan+0x108/0x140 [lfsck]
[ 1256.667697] [<ffffffffc148d6a9>] lfsck_double_scan+0x59/0x200 [lfsck]
[ 1256.676632] [<ffffffffc143344a>] ? osd_otable_it_fini+0xca/0x240 [osd_ldiskfs]
[ 1256.686386] [<ffffffff811dee33>] ? kfree+0x103/0x140
[ 1256.693515] [<ffffffffc1492204>] lfsck_master_engine+0x494/0x12f0 [lfsck]
[ 1256.702670] [<ffffffff810c4820>] ? default_wake_function+0x0/0x20
[ 1256.710967] [<ffffffffc1491d70>] ? lfsck_master_engine+0x0/0x12f0 [lfsck]
[ 1256.720038] [<ffffffff810b099f>] kthread+0xcf/0xe0
[ 1256.726787] [<ffffffff810bf114>] ? finish_task_switch+0x54/0x160
[ 1256.734873] [<ffffffff810b08d0>] ? kthread+0x0/0xe0
[ 1256.741642] [<ffffffff816b4fd8>] ret_from_fork+0x58/0x90
[ 1256.748880] [<ffffffff810b08d0>] ? kthread+0x0/0xe0
[ 1256.755609]
[ 1256.758443] Kernel panic - not syncing: LBUG
{code]

[Gerrit Updater]

Fan Yong ([e-mail]) uploaded a new patch: https://review.whamcloud.com/30165
Subject: LU-10134 lfsck: not add requests if engine out of work
Project: fs/lustre-release
Branch: master
Current Patch Set: 1
Commit: bd6411b44714543fae3e841ffd055e67f105d24b

[Joseph Gmitter (Inactive)]

Hi Fan Yong,

Can you please look at this issue? It has become a roadblock for soak testing on master.

Thanks.
Joe

[Cliff White (Inactive)]

After this incident, the system is completely non-recoverable. Any attempt to re-mount MDT0 causes this LBUG to hit again, and again. Multiple core dumps are available on spirit.