Uploaded image for project: 'Couchbase Server'
  1. Couchbase Server
  2. MB-19207

[FTS] MCP: after stop rebalance, rebalancing again doesn't move pindexes due to MCP deadlock

    XMLWordPrintable

Details

    • Bug
    • Resolution: Fixed
    • Critical
    • 4.5.0
    • 4.5.0
    • fts
    • None
    • Untriaged
    • Unknown

    Description

      Saw this along the way while diagnosing MB-19040, but by that time had ended up too many button presses to remember.

      Roughly, in 3 node cluster-run environment on OSX...

      CBFT_ENV_OPTIONS=hideUI=false COUCHBASE_NUM_VBUCKETS=1024 ./cluster_run --nodes=3
      

      Run one of Aruna's scripts....

      ./testrunner -i MY_CLUSTER.ini -p skip-cleanup=True,get-cbcollect-info=False,get-logs=False,stop-on-failure=False -t fts.moving_topology_fts.MovingTopFTS.node_reboot_only_kv_during_index_building,items=10000,cluster=D,F,F,GROUP=P0
      

      After it's done, start playing around with rebalancing, removing nodes, stopping rebalance, and restarting rebalance. You might see that the pindex layout isn't balanced and attempting to Rebalance again (incorrectly) does not help make the pindex layout balanced again.

      Attachments

        No reviews matched the request. Check your Options in the drop-down menu of this sections header.

        Activity

          People

            steve Steve Yen
            steve Steve Yen
            Votes:
            0 Vote for this issue
            Watchers:
            2 Start watching this issue

            Dates

              Created:
              Updated:
              Resolved:

              Gerrit Reviews

                There are no open Gerrit changes

                PagerDuty