Uploaded image for project: 'Core Server'
  1. Core Server
  2. SERVER-67790

[4.2] Running with enableMajorityReadConcern=false can commit behind the oldest/stable timestamp

XMLWordPrintableJSON

    • Type: Icon: Bug Bug
    • Resolution: Won't Fix
    • Priority: Icon: Major - P3 Major - P3
    • None
    • Affects Version/s: 4.2.21
    • Component/s: None
    • Labels:
      None
    • ALL

      The bug manifests as seeing the following log line then crashing:

      WiredTiger error (22) [1655379983:484753][134088:0x7f405d0d9700], WT_SESSION.timestamp_transaction: __wt_txn_set_commit_timestamp, 676: commit timestamp (1655379983, 851) is less than the oldest timestamp (1655379983, 854): Invalid argument Raw: [1655379983:484753][134088:0x7f405d0d9700], WT_SESSION.timestamp_transaction: __wt_txn_set_commit_timestamp, 676: commit timestamp (1655379983, 851) is less than the oldest timestamp (1655379983, 854): Invalid argument

      This issue is specific to:

      • Running on 4.2
      • Running in replica set mode[1]
      • Running with enableMajorityReadConcern: false

      To resolve the issue, it's recommended to turn enableMajorityReadConcern: true (the default) or upgrade to 4.4.

      [1] Standalones can observe a similar symptom, but that is likely due to user error writing to the local.oplog.rs collection directly. See SERVER-55746.

            Assignee:
            daniel.gottlieb@mongodb.com Daniel Gottlieb (Inactive)
            Reporter:
            daniel.gottlieb@mongodb.com Daniel Gottlieb (Inactive)
            Votes:
            0 Vote for this issue
            Watchers:
            7 Start watching this issue

              Created:
              Updated:
              Resolved: