[LU-4811] Interop 2.5<->2.6 failure on test suite sanity test_24u: error on ioctl 0x4008669a, no data availiable

• Type: Bug
• Resolution: Fixed
• Priority: Blocker
• Fix Version/s: Lustre 2.5.2
• Affects Version/s: Lustre 2.5.1
• Environment: server: 2.5.1; client: lustre-master build # 1945
• Severity: 3
• Assignee: Nathaniel Clark
• Reporter: Maloo

Description

This issue was created by maloo for sarah <[e-mail]>

This issue relates to the following test suite run: http://maloo.whamcloud.com/test_sets/acaf508a-b240-11e3-a93f-52540035b04c.

The sub-test test_24u failed with the following error:

error() without useful message, please fix

== sanity test 24u: create stripe file == 12:37:21 (1395430641)
error on ioctl 0x4008669a for '/mnt/lustre/f24u.sanity' (3): No data available
write: Bad file descriptor

Issue Links

• duplicates: LU-5139 Interop 2.5.1<->2.6 failure on test suite sanity test_27wa: FAIL: setstripe -c 5 -i 4 failed
• is duplicated by: LU-5821 sanity test_24u 2.4-2.7 interop: error on ioctl: No data available
• is duplicated by: LU-4814 Interop 2.5.1<->2.6 failure on test suite lustre-rsync-test test_7: Failure in replication; differences found.
• is related to: LU-3338 IOC_MDC_GETFILESTRIPE can abuse vmalloc()

Comments

[Nathaniel Clark]

Patch 6339 on LU-3338 is generally unrelated to this issue, it just happens to include the fix for this almost as an afterthought. It was infact added in rev 13 of that patch to fix an autotest issue.

[James A Simmons]

Your patch is less intrusive but I have to say back porting http://review.whamcloud.com/6339 is more desirable for us. The full patch from LU-3338 will also address the contention issues with vmalloc on the clients we see as well.

[Nathaniel Clark]

This was fixed on master as an addon to patch http://review.whamcloud.com/6339
see the last line of comment:

Also remove client eadatasize check in mdt xattr packing because
as said above client can handle -EOVERFLOW.

I believe this is what is tripping in this case.

http://review.whamcloud.com/9812

[Peter Jones]

Nathaniel is looking into this one

[Sarah Liu]

https://maloo.whamcloud.com/test_sets/2b16c06a-b242-11e3-a93f-52540035b04c

== sanity-benchmark test fsx: fsx == 17:01:05 (1395446465)
debug=0
error on ioctl 0x4008669a for '/mnt/lustre/f0.fsxfile' (3): No data available
error: setstripe: create stripe file '/mnt/lustre/f0.fsxfile' failed
Using: fsx -c 50 -p 1000 -S 2223 -P /tmp -l 3844216         -N 100000  /mnt/lustre/f0.fsxfile
Chance of close/open is 1 in 50
Seed set to 2223
trunc_hack: fstat: Numerical result out of range
no extend on truncate! not posix!
 sanity-benchmark test_fsx: @@@@@@ FAIL: fsx failed 

[Sarah Liu]

Hit a lot similar error in the interop testing between 2.5.1 server and master client:

https://maloo.whamcloud.com/test_sets/acaf508a-b240-11e3-a93f-52540035b04c

== sanity test 27b: create and write to two stripe file == 12:49:00 (1395431340)
error on ioctl 0x4008669a for '/mnt/lustre/d27/f01' (3): No data available
error: setstripe: create stripe file '/mnt/lustre/d27/f01' failed
 sanity test_27b: @@@@@@ FAIL: setstripe failed