[LU-7104] ASSERTION( osc == oap->oap_obj ) failed

• Status: Resolved
• Project: Lustre
• Affects Version/s: Lustre 2.7.0
• Type: Bug
• Priority: Minor
• Reporter: Matt Ezell
• Assignee: Zhenyu Xu
• Resolution: Fixed

Environment

2.7_fe Client

• Severity: 3

Description

During an OSS outage, we had a client crash with the following log:

Pid: 26171, comm: globus-gridftp-

Call Trace:
[<ffffffffa07db895>] libcfs_debug_dumpstack+0x55/0x80 [libcfs]
[<ffffffffa07dbe97>] lbug_with_loc+0x47/0xb0 [libcfs]
[<ffffffffa0f0a0eb>] osc_io_submit+0x48b/0x4f0 [osc]
[<ffffffffa097c30e>] cl_io_submit_rw+0x6e/0x160 [obdclass]
[<ffffffffa097c30e>] cl_io_submit_rw+0x6e/0x160 [obdclass]
[<ffffffffa097f4c0>] cl_io_read_page+0x180/0x190 [obdclass]
[<ffffffffa0dfdb9b>] ll_readpage+0xcb/0x280 [lustre]
[<ffffffff81128f7c>] generic_file_aio_read+0x1fc/0x700
[<ffffffffa0e2e953>] vvp_io_read_start+0x233/0x460 [lustre]
[<ffffffffa097c59a>] cl_io_start+0x6a/0x140 [obdclass]
[<ffffffffa0980694>] cl_io_loop+0xb4/0x1b0 [obdclass]
[<ffffffffa0dcbdfb>] ll_file_io_generic+0x5bb/0xad0 [lustre]
[<ffffffffa07f464b>] ? cfs_hash_add_unique+0x1b/0x40 [libcfs]
[<ffffffffa0dd8007>] ll_file_aio_read+0x1c7/0x6d0 [lustre]
[<ffffffffa0dd863e>] ll_file_read+0x12e/0x260 [lustre]
[<ffffffff81192215>] vfs_read+0xb5/0x1a0
[<ffffffff81192fc6>] ? fget_light_pos+0x16/0x50
[<ffffffff81192561>] sys_read+0x51/0xb0
[<ffffffff810e88ae>] ? __audit_syscall_exit+0x25e/0x290
[<ffffffff8100b0d2>] system_call_fastpath+0x16/0x1b

This looks like a duplicate of LU-6716, but that was marked as a duplicate of LU-3259. That ticket had two patches land to master, but only one landed to b2_7_fe. Assuming that's actually the fix for this issue, can we get patch 10859 backported to b2_7_fe?

Issue Links

• duplicates: LU-3259 cl_lock refactoring

Comments

[Matt Ezell]

Sorry, I started the copy/paste too late. Here's the actual relevant lines that printed earlier:

[<ffffffffa0d53108>] lov_io_submit+0x2c8/0x4a0 [lov]
LustreError: 26171:0:(osc_io.c:146:osc_io_submit()) ASSERTION( osc == oap->oap_obj ) failed:
LustreError: 26171:0:(osc_io.c:146:osc_io_submit()) LBUG

[James A Simmons]

http://review.whamcloud.com/#/c/16239

[Peter Jones]

Bobijam

Could you please confirm whether this analysis is correct and the identified patch is the one that is needed?

Thanks

Peter

[Zhenyu Xu]

#16239 is the patch, ready to land.

[Jian Yu]

Patch landed. Let's close this ticket.