Uploaded image for project: 'Lustre'
  1. Lustre
  2. LU-6662

mkfs should allow use of errors=panic

Details

    • Bug
    • Resolution: Fixed
    • Major
    • Lustre 2.8.0
    • Lustre 2.5.3
    • 3
    • 9223372036854775807

    Description

      mkfs.lustre defaults mountoptions to error=mount-ro and doesn't allow error=panic. Changing this using tunefs.lustre also gives warning about 'error=mount-ro' is missing.

      Is the use of error=panic discouraged? if so why?

      Attachments

        Issue Links

          Activity

            [LU-6662] mkfs should allow use of errors=panic

            Landed for 2.8

            jgmitter Joseph Gmitter (Inactive) added a comment - Landed for 2.8

            Oleg Drokin (oleg.drokin@intel.com) merged in patch http://review.whamcloud.com/15870/
            Subject: LU-6662 utils: allow overriding default mountopts
            Project: fs/lustre-release
            Branch: master
            Current Patch Set:
            Commit: 927effd2c2374508d904d2022c74574e508c07f1

            gerrit Gerrit Updater added a comment - Oleg Drokin (oleg.drokin@intel.com) merged in patch http://review.whamcloud.com/15870/ Subject: LU-6662 utils: allow overriding default mountopts Project: fs/lustre-release Branch: master Current Patch Set: Commit: 927effd2c2374508d904d2022c74574e508c07f1
            pjones Peter Jones added a comment -

            Thanks Mahmoud but, given that there is a patch that - as far as I understand - you are using we should probably hold this open until that has landed to master

            pjones Peter Jones added a comment - Thanks Mahmoud but, given that there is a patch that - as far as I understand - you are using we should probably hold this open until that has landed to master

            This LU can be closed

            mhanafi Mahmoud Hanafi added a comment - This LU can be closed

            Yes, each subtest should be self-contained.

            adilger Andreas Dilger added a comment - Yes, each subtest should be self-contained.

            The conf-sanity.sh in 2.5.3 contains only up to test_77. Is it OK to just pick up test_88 added in #15870?

            jaylan Jay Lan (Inactive) added a comment - The conf-sanity.sh in 2.5.3 contains only up to test_77. Is it OK to just pick up test_88 added in #15870?

            Hongchao Zhang (hongchao.zhang@intel.com) uploaded a new patch: http://review.whamcloud.com/15870
            Subject: LU-6662 utils: allow overriding default mountopts
            Project: fs/lustre-release
            Branch: master
            Current Patch Set: 1
            Commit: 27bdc3417985dbcbe9790492b60ce0d8141bdbba

            gerrit Gerrit Updater added a comment - Hongchao Zhang (hongchao.zhang@intel.com) uploaded a new patch: http://review.whamcloud.com/15870 Subject: LU-6662 utils: allow overriding default mountopts Project: fs/lustre-release Branch: master Current Patch Set: 1 Commit: 27bdc3417985dbcbe9790492b60ce0d8141bdbba

            I don't think errors=panic is discouraged, this is likely just an oversight from the time when the errors=remount-ro option was never being specified and the warning message was added to ensure that this wasn't missed.

            To fix this problem, I think the right approach is to change the code to check for any errors= option being specified. However, this is complicated by the way that the code is currently implemented, since for it sets errors=remount-ro into default_mountopts in ldiskfs_prepare_lustre(), and then later checks in check_mountfsoptions() whether those default options are present.

            There isn't really an easy way to change this currently, though I think it would be better to change the osd_prepare_lustre() method to be passed only actual mountopts string, ldd_mount_opts[] and its maximum size, instead of passing the default_mountopts and always_mountopts back to the caller. That would allow ldiskfs_prepare_lustre() to verify the mountopts string to handle cases like this, store the defaults into ldd_mount_opts[], and append any user-specified values to override the default values.

            adilger Andreas Dilger added a comment - I don't think errors=panic is discouraged, this is likely just an oversight from the time when the errors=remount-ro option was never being specified and the warning message was added to ensure that this wasn't missed. To fix this problem, I think the right approach is to change the code to check for any errors= option being specified. However, this is complicated by the way that the code is currently implemented, since for it sets errors=remount-ro into default_mountopts in ldiskfs_prepare_lustre() , and then later checks in check_mountfsoptions() whether those default options are present. There isn't really an easy way to change this currently, though I think it would be better to change the osd_prepare_lustre() method to be passed only actual mountopts string, ldd_mount_opts[] and its maximum size, instead of passing the default_mountopts and always_mountopts back to the caller. That would allow ldiskfs_prepare_lustre() to verify the mountopts string to handle cases like this, store the defaults into ldd_mount_opts[], and append any user-specified values to override the default values.
            pjones Peter Jones added a comment -

            Hongchao

            Could you please advise on this one?

            Thanks

            Peter

            pjones Peter Jones added a comment - Hongchao Could you please advise on this one? Thanks Peter

            People

              hongchao.zhang Hongchao Zhang
              mhanafi Mahmoud Hanafi
              Votes:
              0 Vote for this issue
              Watchers:
              7 Start watching this issue

              Dates

                Created:
                Updated:
                Resolved: