Uploaded image for project: 'WiredTiger'
  1. WiredTiger
  2. WT-2193

Handle read-committed checkpoints during snapshot transactions

    • Type: Icon: Bug Bug
    • Resolution: Done
    • Priority: Icon: Major - P3 Major - P3
    • WT2.7.0
    • Affects Version/s: None
    • Component/s: None
    • Labels:
      None

      A recent change to WiredTiger forced metadata checkpoints to be performed at read-committed isolation, rather than whatever isolation level happens to be set in the operation (such at WT_SESSION::create) that triggered the checkpoint.

      Unfortunately, this trips an assertion in the WT_WITH_TXN_ISOLATION macro for snapshot transactions, because the pinned snap_min has been released by the time the checkpoint completes:

      [cpp_unit_test:storage_wiredtiger_record_store_test] 2015-11-01T23:33:58.629-0500 [1446438838:629625][60849:0x7fff748b3180], file:WiredTiger.wt, WT_SESSION.create: src/third_party/wiredtiger/src/meta/meta_track.c, 280: txn_state->id == saved_state.id && (txn_state->snap_min == saved_state.snap_min || saved_state.snap_min == WT_TXN_NONE)
      

            Assignee:
            michael.cahill@mongodb.com Michael Cahill (Inactive)
            Reporter:
            michael.cahill@mongodb.com Michael Cahill (Inactive)
            Votes:
            0 Vote for this issue
            Watchers:
            3 Start watching this issue

              Created:
              Updated:
              Resolved: