Restarting Cassandra nodes after data center maintenance

You must restart Cassandra nodes after completing data center maintenance or when there is an unplanned data center outage.

About this task

You must restart the seed nodes first, followed by the non-seed nodes.

Procedure

To restart Cassandra nodes, complete the following steps:

  1. On each Cassandra node in the data center you are bringing back online, go to the <installDir>/MailboxUtilities/bin/ directory, and run the startGMData script.

    <InstallDIR> is the directory where Cassandra and ZooKeeper nodes are installed.

  2. Optional: Wait for the Cassandra Reaper to complete a repair before proceeding. On a well-maintained Cassandra cluster, a repair operation may take several hours to complete successfully. For details, refer to Checking repair status.
    Important: Skipping the step allows you to bring your data center online much faster. However, by allowing the repair to complete before proceeding, you can ensure that trading partners have immediate access to data created or modified during the maintenance period. Because the repair operation is resource-intensive, operations executed on any machines running Cassandra might be slower during the repair interval. Completing the step mitigates this concern as well.