Uploaded image for project: 'Lustre'
  1. Lustre
  2. LU-1074

servers are umounted while the client is still doing job when running racer.sh by auster

Details

    • Bug
    • Resolution: Fixed
    • Minor
    • Lustre 2.2.0
    • Lustre 2.0.0
    • None
    • 3
    • 2195

    Description

      Running racer by auster, servers are umounted when the client is still working. For detailed information please refer to LU-613

      Attachments

        Issue Links

          Trackbacks

          [Confluence: Whamcloud Community Space] Changelog 2.2 version 2.2.0 Support for networks: o2iblnd OFED 1.5.4 Server support for kernels: 2.6.32220.4.2.el6 (RHEL6) Client support for unpatched kernels: 2.6.18274.18.1.el5 (RHEL5) 2.6.32220.4.2.el6 (RHEL6) 2.6.32.360....

          Activity

            [LU-1074] servers are umounted while the client is still doing job when running racer.sh by auster
            adilger Andreas Dilger added a comment -

            This was fixed recently via a change to zconf_umount_clients (LU-452 change http://review.whamcloud.com/2015, Git hash 26bb579a96efced8825b63b84465334368ecc382). This was committed to master for 2.1.58 (i.e. before the 2.2 split), so it should be fixed.

            The zconf_umount_clients() function will now wait indefinitely for busy processes to finish using the filesystem, rather than pretending the unmount was successful and blindly continuing with the server shutdown (which will definitely block the client from ever unmounting correctly).

            adilger Andreas Dilger added a comment - This was fixed recently via a change to zconf_umount_clients ( LU-452 change http://review.whamcloud.com/2015 , Git hash 26bb579a96efced8825b63b84465334368ecc382). This was committed to master for 2.1.58 (i.e. before the 2.2 split), so it should be fixed. The zconf_umount_clients() function will now wait indefinitely for busy processes to finish using the filesystem, rather than pretending the unmount was successful and blindly continuing with the server shutdown (which will definitely block the client from ever unmounting correctly).
            simmonsja James A Simmons added a comment -

            I noticed this too recently. After racer finishes some dd's are still left running on the client nodes. The test suite then stops because it complains about the mounts being busy so it can't unmount for the next test.

            simmonsja James A Simmons added a comment - I noticed this too recently. After racer finishes some dd's are still left running on the client nodes. The test suite then stops because it complains about the mounts being busy so it can't unmount for the next test.
            chris Chris Gearing (Inactive) added a comment -

            Sarah this sounds like a Lustre issue, this is presumably in the test scripts and those are part of Lustre.

            Please update this issue with Lustre version information and then assign it to a suitable Lustre engineer.

            chris Chris Gearing (Inactive) added a comment - Sarah this sounds like a Lustre issue, this is presumably in the test scripts and those are part of Lustre. Please update this issue with Lustre version information and then assign it to a suitable Lustre engineer.

            People

              wc-triage WC Triage
              sarah Sarah Liu
              Votes:
              0 Vote for this issue
              Watchers:
              0 Start watching this issue

              Dates

                Created:
                Updated:
                Resolved: