Uploaded image for project: 'Lustre'
  1. Lustre
  2. LU-5999

sanity test_44a: FAIL: ll_sparseness_verify 2097152

Details

    • Bug
    • Resolution: Fixed
    • Blocker
    • None
    • Lustre 2.7.0
    • FSTYPE=zfs
    • 3
    • 16729

    Description

      Patch http://review.whamcloud.com/9760 for LU-4820 introduced a regression failure in sanity test 44a under ZFS configuration:

      == sanity test 44a: test sparse pwrite ================================= 12:21:51 (1416486111)
      --------writing /mnt/lustre/d44a-2097152 at 2097152
      read(): Input/output error
       sanity test_44a: @@@@@@ FAIL: ll_sparseness_verify  2097152 
      

      Maloo reports:
      https://testing.hpdd.intel.com/test_sets/0f950f94-711f-11e4-95d2-5254006e85c2
      https://testing.hpdd.intel.com/test_sets/a78e1d32-7e89-11e4-afba-5254006e85c2
      https://testing.hpdd.intel.com/test_sets/4ebe303c-7e86-11e4-bf5a-5254006e85c2
      https://testing.hpdd.intel.com/test_sets/b9ecadae-7e7f-11e4-bf5a-5254006e85c2

      Info required for matching: sanity 44a

      Attachments

        Issue Links

          Activity

            [LU-5999] sanity test_44a: FAIL: ll_sparseness_verify 2097152

            Patch landed to Master.

            jlevi Jodi Levi (Inactive) added a comment - Patch landed to Master.

            Have pushed http://review.whamcloud.com/12990 to revert the change, since the patch only appears to be a performance optimization and nothing will re-break if it is reverted.

            If the original patch is actually good, and the test is just broken because it isn't relevant for ZFS, then someone should push a patch to fix or disable the patch for ZFS. I'd question disabling the test, however, because it passed previously for ZFS so if it is claimed that this is not relevant for ZFS there needs to be a good reason behind it.

            adilger Andreas Dilger added a comment - Have pushed http://review.whamcloud.com/12990 to revert the change, since the patch only appears to be a performance optimization and nothing will re-break if it is reverted. If the original patch is actually good, and the test is just broken because it isn't relevant for ZFS, then someone should push a patch to fix or disable the patch for ZFS. I'd question disabling the test, however, because it passed previously for ZFS so if it is claimed that this is not relevant for ZFS there needs to be a good reason behind it.

            People

              wc-triage WC Triage
              yujian Jian Yu
              Votes:
              0 Vote for this issue
              Watchers:
              6 Start watching this issue

              Dates

                Created:
                Updated:
                Resolved: