Cassandra is a replicated data store. Each Cassandra node contains a copy of all data.
When a node is lost, it is sufficient to restore the node and sync the data from cluster to this
node. No backups or snapshots are necessary.
To synchronize a corrupted Cassandra node, complete the following steps:
-
Uninstall the corrupted Cassandra node.
-
Delete the Global Mailbox Cassandra
installation and directories.
The data directories are defined in the cassandra.yaml file.
-
Reinstall Cassandra on that node with the same configuration.
-
Update the <install dir>/apache-cassandra/conf/cassandra-env.sh file
with the following information:
JVM_OPTS="$JVM_OPTS
-Dcassandra.replace_address=<ip_address_of_this_node>
-
Start the Cassandra server.
-
Remove the
JVM_OPTS="$JVM_OPTS
-Dcassandra.replace_address=<ip_address_of_this_node>
line from the
cassandra-env.sh file.
-
Make sure that the Cassandra Reaper tool is running. The Reaper will automatically repair the
node, for details refer to Checking repair status.