Uploaded image for project: 'Lustre'
  1. Lustre
  2. LU-11749

sanity-sec test 23b fails with 'Should return gid=60010 or 60010 on client2'

Details

    • Bug
    • Resolution: Fixed
    • Minor
    • Lustre 2.13.0, Lustre 2.12.2
    • Lustre 2.12.0
    • None
    • 3
    • 9223372036854775807

    Description

      sanity-sec test_23b started failing on October 30, 2018 with 'Should return gid=60010 or 60010 on client2'

      Looking at the client test_log from a recent failure, https://testing.whamcloud.com/test_sets/1e06f64c-f9d7-11e8-b216-52540065bddc , we see an issue parsing output

      CMD: onyx-41vm12 /usr/sbin/lctl list_nids | grep tcp | cut -f 1 -d @
      waited 5 seconds for sync
      getfacl: Removing leading '/' from absolute path names
      CMD: onyx-41vm9.onyx.whamcloud.com getfacl /mnt/lustre/d23b.sanity-sec
      getfacl: Removing leading '/' from absolute path names
      CMD: onyx-41vm9.onyx.whamcloud.com getent passwd
      /usr/lib64/lustre/tests/sanity-sec.sh: line 1834: [: sanityusr: integer expression expected
       sanity-sec test_23b: @@@@@@ FAIL: Should return gid=60010 or 60010 on client2 
      

      This issue may be caused by the landing of the patch for LU-9795 https://review.whamcloud.com/28662.

      From sanity-sec test_23b, the code that fails is

      1799         local fs_id=$((IDBASE+10))
      …
      1828         # getfacl default acl on client2 (mapped gid=60010)
      1829         mapped_id=$(do_node ${clients_arr[1]} getfacl $testdir |
      1830                         grep -E "default:group:.*:rwx" | awk -F: '{print $3}')
      1831         fs_user=$(do_node ${clients_arr[1]} getent passwd |
      1832                         grep :$fs_id:$fs_id: | cut -d: -f1)
      1833         [ -z "$fs_user" ] && fs_user=$fs_id
      1834         [ $mapped_id -eq $fs_id -o "$mapped_id" = "$fs_user" ] ||
      1835                 error "Should return gid=$fs_id or $fs_user on client2"
      1836 
      

      sanity-sec test 23b requires two clients. The test can be kicked off/run on either client 1 or client 2. The issue may be that getfacl and getent needs to be run on the client that is not kicking off the test. In the failures, the test is being run on client 2, but looking at (a small number of) tests that pass, the test is kicked off on client 1.

      More failures are at:
      https://testing.whamcloud.com/test_sets/431b8374-dc8d-11e8-b46b-52540065bddc
      https://testing.whamcloud.com/test_sets/0059669a-dd10-11e8-975a-52540065bddc
      https://testing.whamcloud.com/test_sets/c813ae6e-dd76-11e8-b46b-52540065bddc

      Attachments

        Issue Links

          Activity

            [LU-11749] sanity-sec test 23b fails with 'Should return gid=60010 or 60010 on client2'

            Oleg Drokin (green@whamcloud.com) merged in patch https://review.whamcloud.com/34623/
            Subject: LU-11749 tests: sanity-sec 23b exec commands on right node
            Project: fs/lustre-release
            Branch: b2_12
            Current Patch Set:
            Commit: e13e0efe97c878206227044365391431fcf76644

            gerrit Gerrit Updater added a comment - Oleg Drokin (green@whamcloud.com) merged in patch https://review.whamcloud.com/34623/ Subject: LU-11749 tests: sanity-sec 23b exec commands on right node Project: fs/lustre-release Branch: b2_12 Current Patch Set: Commit: e13e0efe97c878206227044365391431fcf76644

            James Nunez (jnunez@whamcloud.com) uploaded a new patch: https://review.whamcloud.com/34623
            Subject: LU-11749 tests: sanity-sec 23b exec commands on right node
            Project: fs/lustre-release
            Branch: b2_12
            Current Patch Set: 1
            Commit: 33665776dcc25f9ba7a2c45a69f4ab41392b0e91

            gerrit Gerrit Updater added a comment - James Nunez (jnunez@whamcloud.com) uploaded a new patch: https://review.whamcloud.com/34623 Subject: LU-11749 tests: sanity-sec 23b exec commands on right node Project: fs/lustre-release Branch: b2_12 Current Patch Set: 1 Commit: 33665776dcc25f9ba7a2c45a69f4ab41392b0e91
            pjones Peter Jones added a comment -

            Landed for 2.13

            pjones Peter Jones added a comment - Landed for 2.13

            Oleg Drokin (green@whamcloud.com) merged in patch https://review.whamcloud.com/33846/
            Subject: LU-11749 tests: sanity-sec 23b exec commands on right node
            Project: fs/lustre-release
            Branch: master
            Current Patch Set:
            Commit: 3c64d3310b7b46689b69091f512663bcb5aecdaf

            gerrit Gerrit Updater added a comment - Oleg Drokin (green@whamcloud.com) merged in patch https://review.whamcloud.com/33846/ Subject: LU-11749 tests: sanity-sec 23b exec commands on right node Project: fs/lustre-release Branch: master Current Patch Set: Commit: 3c64d3310b7b46689b69091f512663bcb5aecdaf

            Sebastien Buisson (sbuisson@ddn.com) uploaded a new patch: https://review.whamcloud.com/33846
            Subject: LU-11749 tests: sanity-sec 23b exec commands on right node
            Project: fs/lustre-release
            Branch: master
            Current Patch Set: 1
            Commit: 7ac94583578a8ea633305e81b9ebd689e8835540

            gerrit Gerrit Updater added a comment - Sebastien Buisson (sbuisson@ddn.com) uploaded a new patch: https://review.whamcloud.com/33846 Subject: LU-11749 tests: sanity-sec 23b exec commands on right node Project: fs/lustre-release Branch: master Current Patch Set: 1 Commit: 7ac94583578a8ea633305e81b9ebd689e8835540

            Nice catch James! There is an issue with the test itself, not the feature being tested.

            The problem is indeed due to the fact that, for some reason that I was not aware of, it happens that the client on which the test is kicked off is not necessarily client 1.
            This is an issue with sanity-sec test 23b as it aims at exercising ACL mapping from different nodes.

            I will push a patch to fix this issue in the test.

            Thanks,
            Sebastien.

            sebastien Sebastien Buisson added a comment - Nice catch James! There is an issue with the test itself, not the feature being tested. The problem is indeed due to the fact that, for some reason that I was not aware of, it happens that the client on which the test is kicked off is not necessarily client 1. This is an issue with sanity-sec test 23b as it aims at exercising ACL mapping from different nodes. I will push a patch to fix this issue in the test. Thanks, Sebastien.
            pjones Peter Jones added a comment -

            Sebastien

            Could you please advise on this issue?

            Thanks

            Peter

            pjones Peter Jones added a comment - Sebastien Could you please advise on this issue? Thanks Peter

            People

              sebastien Sebastien Buisson
              jamesanunez James Nunez (Inactive)
              Votes:
              0 Vote for this issue
              Watchers:
              4 Start watching this issue

              Dates

                Created:
                Updated:
                Resolved: