1. Cleanly shut-down the node (only 1 node should be brought down at a time)
a. Perform drainnodetool -hdrain
b. Disable Gossipnodetool -hdisablegossip c. Disable Thriftnodetool -hdisablethrift d. Disable binary (native protocol)nodetool -hdisablebinary
e. Stop Cassandra service
2. Set Cassandra service to disabled
3. Configure purge gossip state information:
Edit cassandra-env.sh file adding: JVM_OPTS="$JVM_OPTS -Dcassandra.load_ring_state=false"
4. Change yaml file setting start_rpc = false and start_native_transport = false (this is only for the node involved in the IP change)
5. Change hosts files on all nodes to the new IP address for the node involved in the change
6. Perform IP change
7. Reboot machine
8. Set Cassandra service to automatic
9. Start service
10. Verify node state, ring, load, activity, etc
11. Run repair on current column families that may have missed writes to re-IP'd node
12. Confirm that data is in sync
13. Edit cassandra-env.sh file removing the line to purge gossip state: JVM_OPTS="$JVM_OPTS -Dcassandra.load_ring_state=false"
14. Change yaml file for the node involved in the IP change setting start_rpc = true and start_native_transport = true
15. Enable binary
nodetool -h enablebinary
16. Enable thrift
nodetool -h enablethrift
No comments:
Post a Comment