[LU-2255] mds-survey falls over after ~30 consecutive runs

• Type: Task
• Resolution: Incomplete
• Priority: Major
• Affects Version/s: Lustre 2.4.0
• Assignee: Di Wang
• Reporter: Richard Henwood (Inactive)

Description

I'm running mds-survey using:

for i in {0..100}; do echo $i; thrlo=16 thrhi=16 file_count=16 dir_count=16 /usr/bin/mds-survey >> ./100_runs_latestbuild.txt; sleep 30; done

After 30 runs or so, the machine reboots. I'll try and capture more logs.

After reboot, I mount the drive and this is what I find:

# mount -t ldiskfs /dev/sdb /mnt
[root@fat-amd-4 ~]# ls /mnt
capa_keys          oi.16.0   oi.16.18  oi.16.27  oi.16.36  oi.16.45  oi.16.54  oi.16.63              seq_ctl
changelog_catalog  oi.16.1   oi.16.19  oi.16.28  oi.16.37  oi.16.46  oi.16.55  oi.16.7               seq_srv
changelog_users    oi.16.10  oi.16.2   oi.16.29  oi.16.38  oi.16.47  oi.16.56  oi.16.8
CONFIGS            oi.16.11  oi.16.20  oi.16.3   oi.16.39  oi.16.48  oi.16.57  oi.16.9
fld                oi.16.12  oi.16.21  oi.16.30  oi.16.4   oi.16.49  oi.16.58  OI_scrub
last_rcvd          oi.16.13  oi.16.22  oi.16.31  oi.16.40  oi.16.5   oi.16.59  PENDING
lfsck_bookmark     oi.16.14  oi.16.23  oi.16.32  oi.16.41  oi.16.50  oi.16.6   quota_master
lost+found         oi.16.15  oi.16.24  oi.16.33  oi.16.42  oi.16.51  oi.16.60  quota_slave
NIDTBL_VERSIONS    oi.16.16  oi.16.25  oi.16.34  oi.16.43  oi.16.52  oi.16.61  ROOT
O                  oi.16.17  oi.16.26  oi.16.35  oi.16.44  oi.16.53  oi.16.62  seq-200000003-lastid