ZFS OSD missing OI iterator for scrub/LFSCK
(LU-3489)
|
|
| Status: | Resolved |
| Project: | Lustre |
| Component/s: | None |
| Affects Version/s: | Lustre 2.6.0 |
| Fix Version/s: | Lustre 2.6.0 |
| Type: | Technical task | Priority: | Major |
| Reporter: | Maloo | Assignee: | Andreas Dilger |
| Resolution: | Fixed | Votes: | 0 |
| Labels: | prz, zfs | ||
| Issue Links: |
|
||||||||
| Rank (Obsolete): | 11421 | ||||||||
| Description |
|
This issue was created by maloo for Nathaniel Clark <nathaniel.l.clark@intel.com> This issue relates to the following test suite run: http://maloo.whamcloud.com/test_sets/b6717d88-4462-11e3-8472-52540035b04c. The sub-test test_8 failed with the following error:
Info required for matching: sanity-lfsck 8 |
| Comments |
| Comment by Andreas Dilger [ 05/Nov/13 ] |
|
Is this only failing tests when the ZFS scrub code is enabled via http://review.whamcloud.com/7149 so this isn't actually a failure on the current master branch We still want to land that patch, but only once this problem is fixed. Is it possible that the ZFS scrub is just slower than ldiskfs? |
| Comment by nasf (Inactive) [ 06/Nov/13 ] |
|
https://maloo.whamcloud.com/test_sets/b6717d88-4462-11e3-8472-52540035b04c sanity-lfsck test_9a failed also: The ldiskfs scrub speed in my local VM environment is not slower than 5'000 objs/sec, even if ZFS scrub speed is only 1/10 of the ldiskfs, it should finished in time. |
| Comment by Alex Zhuravlev [ 06/Nov/13 ] |
|
hmm, it passes on my local setup consistently.. |
| Comment by Alex Zhuravlev [ 06/Nov/13 ] |
|
I changed number of objects in test 8 to 20K and run the test again: average_speed_phase1: 5254 items/sec |
| Comment by Alex Zhuravlev [ 06/Nov/13 ] |
|
the changed patch is submitted for testing: it should dump stats upon a failure. |
| Comment by Jian Yu [ 26/Mar/14 ] |
|
One more instance on master branch with FSTYPE=zfs: |
| Comment by nasf (Inactive) [ 27/Mar/14 ] |
|
More failure instance: https://maloo.whamcloud.com/test_sessions/a7e15cfa-b525-11e3-8598-52540035b04c |
| Comment by Andreas Dilger [ 18/Apr/14 ] |
|
I'm working on a patch to make sanity-lfsck use wait_update_facet() for most of the checks, so that the test will be more tolerant for slowdowns in the tests. |
| Comment by nasf (Inactive) [ 18/Apr/14 ] |
|
Andreas, the replacement sleep with wait_update_facet() has been done in the patch: |
| Comment by Jodi Levi (Inactive) [ 08/May/14 ] |
|
Patch landed to Master. Please reopen ticket if more work is needed. |
| Comment by Jessica A. Popp (Inactive) [ 11/Jun/14 ] |
|
Is this a relevant fix for getting landed on b2_5? |
| Comment by nasf (Inactive) [ 12/Jun/14 ] |
|
The patch http://review.whamcloud.com/9704 is valuable for b2_5 also, but it cannot be directly applied on b2_5. |
| Comment by Jian Yu [ 20/Sep/14 ] |
|
The back-ported patches are tracked under |
[LU-4202]