[LU-7586] sanity-hsm test_406: "cannot complete migration after HSM remove"

• Type: Bug
• Resolution: Fixed
• Priority: Minor
• Fix Version/s: Lustre 2.8.0
• Severity: 3
• Assignee: John Hammond
• Reporter: Maloo

Description

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

This issue relates to the following test suite run: https://testing.hpdd.intel.com/test_sets/f5397500-a709-11e5-8a33-5254006e85c2.

The sub-test test_406 failed with the following error:

cannot complete migration after HSM remove

Please provide additional information about the failure here.

Info required for matching: sanity-hsm 406

/mnt/lustre/d406.sanity-hsm/f406.sanity-hsm migrate failed: Function not implemented (-38)
mv: cannot migrate '/mnt/lustre/d406.sanity-hsm/f406.sanity-hsm' to MDT0001: Function not implemented
 sanity-hsm test_406: @@@@@@ FAIL: cannot complete migration after HSM remove 

This test case was added by LU-6866 http://review.whamcloud.com/#/c/17511/

Issue Links

• is related to: LU-6866 MDT file migration is incompatible with HSM

Comments

[Joseph Gmitter (Inactive)]

Ticket cleanup: The patch had landed to master for the 2.8.0 release in January.

[Gerrit Updater]

Oleg Drokin ([e-mail]) merged in patch http://review.whamcloud.com/17828/
Subject: LU-7586 test: wait for remove in sanity-hsm test_406()
Project: fs/lustre-release
Branch: master
Current Patch Set:
Commit: 31e53250276a405b4a9c6aa689e1a9c3f09fc7b1

[James Nunez (Inactive)]

A recent sanity-hsm run on master fails with a different error message that the one reported in the description

Updated after 2s: wanted 'SUCCEED' got 'SUCCEED'
cb_migrate_mdt_init: Failed to open '/mnt/lustre/d406.sanity-hsm/f406.sanity-hsm': Not a directory (20)
mv: cannot migrate '/mnt/lustre/d406.sanity-hsm/f406.sanity-hsm' to MDT0001: Not a directory
 sanity-hsm test_406: @@@@@@ FAIL: cannot complete migration after HSM remove 

This is fixed with your proposed patch http://review.whamcloud.com/#/c/17828/.

Logs at
2016-01-11 13:42:44 - https://testing.hpdd.intel.com/test_sets/40779ea0-b88c-11e5-b10a-5254006e85c2

[Gerrit Updater]

John L. Hammond ([e-mail]) uploaded a new patch: http://review.whamcloud.com/17828
Subject: LU-7586 test: wait for remove in sanity-hsm test_406()
Project: fs/lustre-release
Branch: master
Current Patch Set: 1
Commit: 22e0876184f71a946197ece4a49324b869febab8

[Andreas Dilger]

The relevant part of the test looks like:

        # Should migrate $tdir but not $tfile.
        $LFS mv -M1 $DIR/$tdir &&
                error "migrating HSM an archived file should fail"

        $LFS hsm_restore $DIR/$tdir/$tfile
        wait_request_state "$fid" RESTORE SUCCEED

        $LFS hsm_remove $DIR/$tdir/$tfile ||
                error "cannot remove $DIR/$tdir/$tfile from archive"

        cat $DIR/$tdir/$tfile > /dev/null ||
                error "cannot read $DIR/$tdir/$tfile"

        $LFS mv -M1 $DIR/$tdir/$tfile ||
                error "cannot complete migration after HSM remove"

It seems there is some kind of race between removing the HSM attributes from the file and the second migration attempt? Alternately, is it incorrect to call "lfs mv -M1" on a file instead of a directory? I recall Di saying a few times that migration is for directories and not individual files.