[#LU-354] cascading_rw keep running without actually write a thing on RHEL6

[LU-354] cascading_rw keep running without actually write a thing on RHEL6 Created: 23/May/11 Updated: 28/Sep/12 Resolved: 28/Sep/12
Status:	Resolved
Project:	Lustre
Component/s:	None
Affects Version/s:	Lustre 2.1.0
Fix Version/s:	Lustre 2.3.0

Type:

Bug

Priority:

Minor

Reporter:

Sarah Liu

Assignee:

Oleg Drokin

Resolution:

Fixed

Votes:

Labels:

None

Environment:

RHEL6/x86_64

Attachments:

mds-trace.log

ost-trace.log

Severity:

Rank (Obsolete):

4665

Description

cascading_rw keeps running for an abnormal long time without actually writing, so does metabench

Comments

Comment by Peter Jones [ 25/May/11 ]

Oleg

Could you please look into this one? It cropped up multiple times during RHEL6 NFS testing

Thanks

Peter

Comment by Oleg Drokin [ 25/May/11 ]

Would be good to have some sort of a log or a failed test run pointer.

Comment by Sarah Liu [ 27/May/11 ]

syslog from mds:
--------------------
Lustre: 3093:0:(quota_master.c:793:close_quota_files()) quota[0] is off already
Lustre: 3093:0:(quota_master.c:793:close_quota_files()) quota[1] is off already
Lustre: DEBUG MARKER: ----~~============= acceptance-small: parallel-scale ============~~---- Fri May 27 15:52:46 PDT 2011
Lustre: DEBUG MARKER: only running test cascading_rw
Lustre: DEBUG MARKER: excepting tests: parallel_grouplock
Lustre: DEBUG MARKER: Using TIMEOUT=20
Lustre: 3093:0:(quota_master.c:793:close_quota_files()) quota[0] is off already
Lustre: 3794:0:(debug.c:323:libcfs_debug_str2mask()) You are trying to use a numerical value for the mask - this will be deprecated in a future release.
Lustre: DEBUG MARKER: == parallel-scale test cascading_rw: cascading_rw == 15:52:52 (1306536772)
Lustre: 3093:0:(ldlm_lib.c:871:target_handle_connect()) lustre-MDT0000: connection from 2861796d-8301-8621-b112-6fe55999d574@192.168.4.16@o2ib t0 exp (null) cur 1306536778 last 0
Lustre: 3093:0:(ldlm_lib.c:871:target_handle_connect()) Skipped 2 previous similar messages
Lustre: 3093:0:(sec.c:1474:sptlrpc_import_sec_adapt()) import lustre-MDT0000-> netid 50000: select flavor null
Lustre: 3093:0:(sec.c:1474:sptlrpc_import_sec_adapt()) Skipped 3 previous similar messages

syslog from ost:
------------------------
Lustre: DEBUG MARKER: ----~~============= acceptance-small: parallel-scale ============~~---- Fri May 27 15:52:46 PDT 2011
Lustre: DEBUG MARKER: only running test cascading_rw
Lustre: DEBUG MARKER: excepting tests: parallel_grouplock
Lustre: DEBUG MARKER: Using TIMEOUT=20
Lustre: 5902:0:(debug.c:323:libcfs_debug_str2mask()) You are trying to use a numerical value for the mask - this will be deprecated in a future release.
Lustre: 5902:0:(debug.c:323:libcfs_debug_str2mask()) Skipped 5 previous similar messages
Lustre: DEBUG MARKER: == parallel-scale test cascading_rw: cascading_rw == 15:52:52 (1306536772)
Lustre: 4085:0:(ldlm_lib.c:871:target_handle_connect()) lustre-OST0001: connection from 2861796d-8301-8621-b112-6fe55999d574@192.168.4.16@o2ib t0 exp (null) cur 1306507979 last 0
Lustre: 4093:0:(filter.c:2817:filter_connect()) lustre-OST0000: Received MDS connection (0x8fe0118409224bc6); group 0
Lustre: 4093:0:(filter.c:2817:filter_connect()) Skipped 1 previous similar message
Lustre: 4093:0:(sec.c:1474:sptlrpc_import_sec_adapt()) import lustre-OST0000->NET_0x50000c0a80410_UUID netid 50000: select flavor null
Lustre: 4093:0:(sec.c:1474:sptlrpc_import_sec_adapt()) Skipped 1 previous similar message
Lustre: 4085:0:(ldlm_lib.c:871:target_handle_connect()) Skipped 1 previous similar message

Comment by Sarah Liu [ 24/Jun/11 ]

It turns out that mpi causes this problem, just remove all the mpi options and the tests can pass. so close it.

Comment by Sarah Liu [ 24/Jun/11 ]

not lustre bug.