Uploaded image for project: 'Core Server'
  1. Core Server
  2. SERVER-46968

Test locally that we eliminate retries of short-running operations due to planned mongos shutdown (without streamable isMaster)

XMLWordPrintableJSON

    • Type: Icon: Task Task
    • Resolution: Done
    • Priority: Icon: Major - P3 Major - P3
    • None
    • Affects Version/s: None
    • Component/s: Sharding
    • None
    • Repl 2020-06-01

      Write a workload of short-running reads and writes (<1 second). This should be run against a
      ShardingTest with 2 mongoses, where the two mongoses are alternately shut down. Run this workload with the 4.2 version of pymongo to test without streamable isMaster (consider using a smaller client heartbeat frequency or a longer quiesce mode, in order to be robust to slow machines). The test should disable retries in pymongo and assert that all commands succeed. This demonstrates that the need for retries is eliminated. Consider only running this test on variants that are not typically slow.

        1. mongos_quiesce_mode_non_streamable_ismaster.txt
          1.85 MB
        2. mongos_quiesce_mode_non_streamable_ismaster_with_hanging_read.txt
          1.34 MB

            Assignee:
            jason.chan@mongodb.com Jason Chan
            Reporter:
            tess.avitabile@mongodb.com Tess Avitabile (Inactive)
            Votes:
            0 Vote for this issue
            Watchers:
            2 Start watching this issue

              Created:
              Updated:
              Resolved: