Uploaded image for project: 'Flume'
  1. Flume
  2. FLUME-191

Nodes cannot reconnect to master when master is not shut down gracefully

    XMLWordPrintableJSON

Details

    • Bug
    • Status: Closed
    • Critical
    • Resolution: Cannot Reproduce
    • 0.9.0
    • 0.9.1, 0.9.1u1, 0.9.2
    • Node
    • None
    • Ubuntu 10.04

    Description

      Version: 0.9.0

      When I stop the master using /etc/init.d/flume-master stop, nodes become disconnected as expected. When I start the master back up using /etc/init.d/flume-master start, nodes reconnect as expected.

      When I stop the master using kill -9, nodes become disconnected as expected. When I start the master back up using /etc/init.d/flume-master start, nodes DO NOT reconnect as expected. When this happens, the spinning nodes do not show up in the first table of the master's web interface. They only show up in the Physical/logical Node Mapping section.

      It appears that nodes cannot reconnect to the master if it was not shut down gracefully. This problem is significant because the multiple master feature does not work. It is also significant because if you are using BESink with tail(), all data is lost until you can restart the flume-node process on every node.

      -Ben Standefer

      Attachments

        Activity

          People

            jmhsieh Jonathan Hsieh
            flume_aguynamedben Disabled imported user
            Votes:
            0 Vote for this issue
            Watchers:
            1 Start watching this issue

            Dates

              Created:
              Updated:
              Resolved: