[LU-3119] System hang when running sanity test 24x on DNE with ZFS

• Type: Bug
• Resolution: Duplicate
• Priority: Minor
• Labels: dne zfs
• Environment: server and client: lustre-master build #1370
• Severity: 3
• Assignee: WC Triage
• Reporter: Sarah Liu

Description

System hang when running sanity test_24x with 2MDTs over ZFS

MDS console shows:

Lustre: DEBUG MARKER: == sanity test 24x: cross rename/link should be failed == 11:47:46 (1365187666)
LustreError: 14710:0:(mdt_reint.c:944:mdt_reint_link()) Target directory [0x380000bd0:0x1ae82:0x0] is on another MDT
LNet: Service thread pid 14747 completed after 0.00s. This indicates the system was overloaded (too many service threads, or there were not enough hardware resources).

client console shows:

Lustre: DEBUG MARKER: == sanity test 24x: cross rename/link should be failed == 11:47:46 (1365187666)
Lustre: 8053:0:(dir.c:463:ll_get_dir_page()) Page-wide hash collision: 1706989648068149248
LustreError: 8053:0:(dir.c:594:ll_dir_read()) error reading dir [0x380000bd0:0x27db:0x0] at 1706989648068149248: rc -5
Lustre: 8053:0:(dir.c:463:ll_get_dir_page()) Page-wide hash collision: 1706989648068149248
LustreError: 8053:0:(dir.c:594:ll_dir_read()) error reading dir [0x380000bd0:0x27db:0x0] at 1706989648068149248: rc -5

Issue Links

• duplicates: LU-2990 Failure on sanity test_24v: error in listing large dir