[LU-8601] sanity test_230d: Timeout on ZFS backed MDSs

• Type: Bug
• Resolution: Duplicate
• Priority: Minor
• Affects Version/s: Lustre 2.10.0, Lustre 2.11.0
• Labels: zfs
• Severity: 3
• Assignee: Alex Zhuravlev
• Reporter: Maloo

Description

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

This issue relates to the following test suite run:
https://testing.hpdd.intel.com/test_sets/7a155540-7613-11e6-b08e-5254006e85c2
https://testing.hpdd.intel.com/test_sets/d94ac9c8-75ff-11e6-8a8c-5254006e85c2
https://testing.hpdd.intel.com/test_sets/f347090a-7630-11e6-8a8c-5254006e85c2

The sub-test test_230d failed with the following error:

test failed to respond and timed out

Please provide additional information about the failure here.

Info required for matching: sanity 230d

Issue Links

• is duplicated by: LU-11235 sanity: test_230d timeout on ZFS backends
• is related to: LU-9247 replay-ost-single test_5: test failed to respond and timed out

Comments

[Andreas Dilger]

Close as a duplicate of LU-11235.

[Bob Glossman (Inactive)]

another on b2_10:
https://testing.hpdd.intel.com/test_sets/0db2bb16-5f93-11e8-b303-52540065bddc

[Mikhail Pershin]

+1
https://testing.hpdd.intel.com/test_sets/8d897e16-0f33-11e8-a6ad-52540065bddc

[Patrick Farrell (Inactive)]

Another:
https://testing.hpdd.intel.com/test_sessions/004b644b-af97-429d-954c-65316b8f7a96

[Sarah Liu]

another one on DNE+ZFS 2.9.58

https://testing.hpdd.intel.com/test_sets/3b4e3b94-44b5-11e7-b3fe-5254006e85c2

[James Nunez (Inactive)]

sanity 230d has timed out three times this year only when running with ZFS and DNE. This testing is done for tagged builds of master.

This is only happening with DNE, do we still think that this is just ZFS slowness?

For 2017, here are the only cases where 230d times out:
2.9.52 - https://testing.hpdd.intel.com/test_sets/f4f9d750-efe8-11e6-8c0d-5254006e85c2
2.9.54 - https://testing.hpdd.intel.com/test_sets/6e01cc88-0c1e-11e7-8c9f-5254006e85c2
2.9.55 - https://testing.hpdd.intel.com/test_sets/c4a941a4-1e09-11e7-b742-5254006e85c2

[Joseph Gmitter (Inactive)]

Thanks Alex, that direction is very helpful. Do you think the same case is true for LU-9247?

[Alex Zhuravlev]

migration on ZFS should be very very small as it (migration) involves synchronous I/O which is extremely slow on ZFS w/o ZIL.
I'd suggest to disable this test with ZFS.

[Joseph Gmitter (Inactive)]

Hi Alex,

Can you look into this issue? It also may be the same as LU-9247.

Thanks.
Joe

[Nathaniel Clark]

Even when this test passes on ZFS it is remarkably slow ~1400sec (2 MDS, with 2 MDT each)

Single MDS with 4 MDT it times out.