Uploaded image for project: 'Couchbase Server'
  1. Couchbase Server
  2. MB-45462

CLI should allow failing over inactive nodes when allowUnsafe is true

    XMLWordPrintable

Details

    • Bug
    • Status: Closed
    • Critical
    • Resolution: Fixed
    • Cheshire-Cat
    • 7.0.0
    • tools
    • None
    • Untriaged
    • 1
    • Unknown

    Description

      Problem
      The fix for MB-45110 in ns_server allows nodes that already regularly failed over to be subsequently unsafely failed over. We should permit this also in the CLI. Essentially if the --force option is passed to the CLI when failover is invoked, we should permit it to be applied to nodes that are in the inactiveFailed state. This is needed so clusters don't get stuck as they can do as described in MB-45110.

      Steps to reproduce
      1. Create a 5 node cluster: .137, .138, .139, .140, .142
      2. Stop-server on .140 and when the node becomes unresponsive, fail it over, but don't rebalance it out yet.
      3. Now stop server on .138, .139 nodes.
      4. --force failover of node .138, .139 and .140

      Attachments

        Issue Links

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

          Activity

            Sumedh Basarkod I did not know that I will reopen and add a new patch to add inactiveAdded as well. How does one get a node to be in the inactiveAdded state?

            carlos.gonzalez Carlos Gonzalez Betancort (Inactive) added a comment - Sumedh Basarkod I did not know that I will reopen and add a new patch to add inactiveAdded as well. How does one get a node to be in the inactiveAdded state?
            dfinlay Dave Finlay added a comment -

            Thanks Sumedh - I actually didn't think about this either when I filed the ticket. I was just thinking about the inactiveFaied case. I do think it makes sense for the CLI to mirror the behavior of the REST API, and it's true that Abhijeeth Nuthan's patch allows for the unsafe failover of nodes in any state.

            Thanks Carlos Gonzalez Betancort for reopening.

            dfinlay Dave Finlay added a comment - Thanks Sumedh - I actually didn't think about this either when I filed the ticket. I was just thinking about the inactiveFaied case. I do think it makes sense for the CLI to mirror the behavior of the REST API, and it's true that Abhijeeth Nuthan 's patch allows for the unsafe failover of nodes in any state. Thanks Carlos Gonzalez Betancort for reopening.

            force should now also allow failover of inactiveAdded nodes

            carlos.gonzalez Carlos Gonzalez Betancort (Inactive) added a comment - force should now also allow failover of inactiveAdded nodes

            Build couchbase-server-7.0.0-4885 contains couchbase-cli commit 02b1dd1 with commit message:
            MB-45462 Allow failover of inactiveAdded nodes

            build-team Couchbase Build Team added a comment - Build couchbase-server-7.0.0-4885 contains couchbase-cli commit 02b1dd1 with commit message: MB-45462 Allow failover of inactiveAdded nodes

            Verified on 7.0.0-4890. Closing this

            sumedh.basarkod Sumedh Basarkod (Inactive) added a comment - Verified on 7.0.0-4890. Closing this

            People

              carlos.gonzalez Carlos Gonzalez Betancort (Inactive)
              dfinlay Dave Finlay
              Votes:
              0 Vote for this issue
              Watchers:
              7 Start watching this issue

              Dates

                Created:
                Updated:
                Resolved:

                Gerrit Reviews

                  There are no open Gerrit changes

                  PagerDuty