Uploaded image for project: 'Core Server'
  1. Core Server
  2. SERVER-50971

Invariant failure, WT_NOTFOUND: item not found

    XMLWordPrintable

    Details

    • Type: Bug
    • Status: Closed
    • Priority: Critical - P2
    • Resolution: Done
    • Affects Version/s: 4.4.0, 4.4.1
    • Fix Version/s: None
    • Component/s: Stability, WiredTiger
    • Labels:
      None
    • Operating System:
      ALL
    • Steps To Reproduce:
      Hide

      Unfortunately I cannot get it reproduced on will yet.

      Show
      Unfortunately I cannot get it reproduced on will yet.
    • Case:
    • Linked BF Score:
      119

      Description

      Hi

       

      We have been getting infrequent (1-2 times a day) aborts of this kind lately. This happened on 4.4.1 and 4.4.0 too. We have had years without issue (3-node replica set), upgrading as new versions come around. It isn't tied to specific times either.

      This is on an AWS I3 instance with nvme drive, formatted to xfs.

      Snippet of the logs here and in the attachement.

      https://pastebin.com/uqauh8H0

      We have tried removing the DB path and resyncing from a secondary, but this did not fix it.

      What could cause this? Could it be a specific query? Could the disk itself be corrupt? How can I help pinpointing the issue?

        Attachments

        1. image-2020-10-15-16-49-06-849.png
          image-2020-10-15-16-49-06-849.png
          69 kB
        2. MongoDB Abort.json
          22 kB
        3. mongo-log.json
          10 kB

          Issue Links

            Activity

              People

              Assignee:
              jonathan.streets Jonathan Streets
              Reporter:
              pieterwjordaanpc@gmail.com Pieter Jordaan
              Participants:
              Votes:
              1 Vote for this issue
              Watchers:
              32 Start watching this issue

                Dates

                Created:
                Updated:
                Resolved: