Uploaded image for project: 'Lustre'
  1. Lustre
  2. LU-1749

llog_lvfs_create()) error looking up logfile

    Details

    • Severity:
      3
    • Rank (Obsolete):
      4411

      Description

      We had an MDS (running 2.1.1-17chaos) lock up due to bug LU-1276. The admins rebooted the node to work around the issue, and after the reboot when the MDS started up we hit the following error:

      2012-08-14 13:48:55 LustreError: 3697:0:(llog_lvfs.c:616:llog_lvfs_create()) error looking up logfile 0x1a768b0a:0xa2f17dca: rc -116
      2012-08-14 13:48:55 LustreError: 3697:0:(llog_cat.c:174:llog_cat_id2handle()) error opening log id 0x1a768b0a:a2f17dca: rc -116
      2012-08-14 13:48:55 LustreError: 3697:0:(llog_obd.c:318:cat_cancel_cb()) Cannot find handle for log 0x1a768b0a
      2012-08-14 13:48:55 LustreError: 3472:0:(llog_obd.c:391:llog_obd_origin_setup()) llog_process() with cat_cancel_cb failed: -116
      2012-08-14 13:48:55 LustreError: 3472:0:(llog_obd.c:218:llog_setup_named()) obd lsa-OST00bd-osc ctxt 2 lop_setup=ffffffffa05b0a60 failed -116
      2012-08-14 13:48:55 LustreError: 3472:0:(osc_request.c:4186:__osc_llog_init()) failed LLOG_MDS_OST_ORIG_CTXT
      2012-08-14 13:48:55 LustreError: 3472:0:(osc_request.c:4203:__osc_llog_init()) osc 'lsa-OST00bd-osc' tgt 'mdd_obd-lsa-MDT0000' catid ffff8808312bd860 rc=-116
      2012-08-14 13:48:55 LustreError: 3472:0:(osc_request.c:4205:__osc_llog_init()) logid 0x1a7680f6:0x615e782e
      2012-08-14 13:48:55 LustreError: 3472:0:(osc_request.c:4233:osc_llog_init()) rc: -116
      2012-08-14 13:48:55 LustreError: 3472:0:(lov_log.c:248:lov_llog_init()) error osc_llog_init idx 189 osc 'lsa-OST00bd-osc' tgt 'mdd_obd-lsa-MDT0000' (rc=-116)
      2012-08-14 13:48:55 LustreError: 3698:0:(llog_lvfs.c:616:llog_lvfs_create()) error looking up logfile 0x1a768b0a:0xa2f17dca: rc -116
      2012-08-14 13:48:55 LustreError: 3698:0:(llog_cat.c:174:llog_cat_id2handle()) error opening log id 0x1a768b0a:a2f17dca: rc -116
      2012-08-14 13:48:55 LustreError: 3698:0:(llog_obd.c:318:cat_cancel_cb()) Cannot find handle for log 0x1a768b0a
      2012-08-14 13:48:55 LustreError: 3407:0:(llog_obd.c:391:llog_obd_origin_setup()) llog_process() with cat_cancel_cb failed: -116
      2012-08-14 13:48:55 LustreError: 3407:0:(llog_obd.c:218:llog_setup_named()) obd lsa-OST00bd-osc ctxt 2 lop_setup=ffffffffa05b0a60 failed -116
      2012-08-14 13:48:55 LustreError: 3407:0:(osc_request.c:4186:__osc_llog_init()) failed LLOG_MDS_OST_ORIG_CTXT
      2012-08-14 13:48:55 LustreError: 3407:0:(osc_request.c:4203:__osc_llog_init()) osc 'lsa-OST00bd-osc' tgt 'mdd_obd-lsa-MDT0000' catid ffff8804333af960 rc=-116
      2012-08-14 13:48:55 LustreError: 3407:0:(osc_request.c:4205:__osc_llog_init()) logid 0x1a7680f6:0x615e782e
      2012-08-14 13:48:56 LustreError: 3407:0:(osc_request.c:4233:osc_llog_init()) rc: -116
      2012-08-14 13:48:56 LustreError: 3407:0:(lov_log.c:248:lov_llog_init()) error osc_llog_init idx 189 osc 'lsa-OST00bd-osc' tgt 'mdd_obd-lsa-MDT0000' (rc=-116)
      

      I don't see a file in the OBJECTS directory that seems to match 0x1a768b0a:0xa2f17dca (if that is where we are looking). Although -116 is -ESTALE, so I'm not sure that we're even getting to the lower-level lookup. It may be that mds_lvfs_fid2dentry() is returning -ESTALE because the id is 0.

      This has left the OST connection "inactive" on the MDS, so any users with data on that OST are currently dead in the water.

        Attachments

          Activity

            People

            • Assignee:
              laisiyao Lai Siyao
              Reporter:
              morrone Christopher Morrone
            • Votes:
              0 Vote for this issue
              Watchers:
              8 Start watching this issue

              Dates

              • Created:
                Updated:
                Resolved: