Stopping ZooKeeper nodes for data center maintenance

When bringing a data center down for maintenance, you must stop the ZooKeeper nodes in the data center because other components are also down and coordination services are not required. Also, you might be bringing the data center down to apply some fix or updates to the ZooKeeper nodes.

1. Ensure the ZooKeeper watchdog process is enabled on all ZooKeeper nodes. For information about ZooKeeper watchdog, see ZooKeeper Watchdog and watchdog.properties files.
2. In the data center that you are bringing down, stop the ZooKeeper watchdog process on each ZooKeeper node. Go to the <install_Dir>/MailboxUtilities/bin directory and run the stopGMCoordinateWatchdog script.

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

3. In the data center that remains online, start the ZooKeeper watchdog process on each ZooKeeper node. Go to the <install_Dir>/MailboxUtilities/bin directory and run the startGMCoordinateWatchdog script.
4. In the data center that you are bringing down, stop each ZooKeeper node. Go to the <install_Dir>/MailboxUtilities/bin and run the stopGMCoordinate script.
   Important: It might take a few minutes for the other data centers to respond to the loss of ZooKeeper nodes. During this time, some actions fail to complete because they cannot obtain locks.