[LU-5462] trigger OI

• Status: Resolved
• Project: Lustre
• Affects Version/s: Lustre 2.4.3
• Type: Question/Request
• Priority: Minor
• Reporter: Mahmoud Hanafi
• Assignee: nasf (Inactive)
• Resolution: Not a Bug

Description

After restart of MDT we got this error. What does mean?
LustreError: 0-0: nbp7-MDT0000: trigger OI scrub by RPC for [0x200088c7c:0x14:0x0], rc = 0 [1]

Comments

[Peter Jones]

Fan Yong

Could you please advise with this one?

Thanks

Peter

[nasf (Inactive)]

Hi Mahmoud,

What is the output for "cat /proc/fs/lustre/osd-ldiskfs/nbp7-MDT0000/oi_scrub" ?

[Mahmoud Hanafi]

nbp7-mds1 ~ # cat /proc/fs/lustre/osd-ldiskfs/nbp7-MDT0000/oi_scrub
name: OI_scrub
magic: 0x4c5fd252
oi_files: 1
status: completed
flags:
param:
time_since_last_completed: 48833 seconds
time_since_latest_start: 51587 seconds
time_since_last_checkpoint: 48833 seconds
latest_start_position: 12
last_checkpoint_position: 536870913
first_failure_position: N/A
checked: 66205633
updated: 2
failed: 0
prior_updated: 0
noscrub: 94
igif: 519
success_count: 3
run_time: 2753 seconds
average_speed: 24048 objects/sec
real-time_speed: N/A
current_position: N/A

[nasf (Inactive)]

The OI scrub is used for rebuilding crashed "FID=>ino/gen" mapping. These message means that during your system running, it found bad "FID=>ino/gen" mapping, then triggered OI scrub to repair the inconsistency automatically. It may be because of data crash, it also may be because of system upgrade, but of course, we cannot say there are no any non-known bugs in the code as to trigger OI scrub by wrong.

From the output "cat /proc/fs/lustre/osd-ldiskfs/nbp7-MDT0000/oi_scrub", your system has ever run OI scrub for 3 times, and two items (mappings) have been repaired in the latest run. Since the OI scrub has completely already, we can generally say that your system comes back to consistency again. So you can keep it there without anything else, the system should can work well. But if you want to know more about the system status, you can disable OI scrub auto trigger mechanism by "echo 0 > /proc/fs/lustre/osd-ldiskfs/nbp7-MDT0000/auto_scrub", then when found bad OI mapping next time, the system will not repair the inconsistency automatically, instead, it may cause related application failure, and then we can try to dump more logs.

[nasf (Inactive)]

Hi Mahmoud,
Do we still have anything to resolve under this ticket? Or can we close it? Thanks!

[Mahmoud Hanafi]

can be closed

[Peter Jones]

ok thanks Mahmoud