How to Configure the Management Server on Both Nodes

1. Create the configuration.
   1. On both nodes, create the data directory for the management server.

      phys-schost-1:/ # mkdir /mgm-data
      phys-schost-2:/ # mkdir /mgm-data

   2. On both nodes, copy the config.ini file from /temp/cluconfig to the mgm-data directory.

      phys-schost-1:/ # cp /temp/cluconfig/config.ini /mgm-data
      phys-schost-2:/ # cp /temp/cluconfig/config.ini /mgm-data

   3. Modify the config.ini file from /temp/cluconfig.

      Alternatively, copy the content from config.ini File for Both Nodes to Store in /mgm-data and overwrite the copied file.

      The configuration in the config.ini file for this example is shown in the following table:

      
      

      Server ID Node Type Global-Cluster Node to Run On Private Net Alias 1 Management node phys-schost-1 2 Management node phys-schost-2 3 Date node phys-schost-1 phys-schost-1-p 4 Date node phys-schost-2 phys-schost-2-p 7 SQL node phys-schost-1 8 SQL node phys-schost-2

   4. Configure the data nodes to communicate over the private interconnect clprivnet addresses.

      Create aliases in the /etc/hosts table for the clprivnet addresses and use them in the config.inifile as the host names.

   5. Set Arbitration=WaitExternal and an appropriate value for ArbitrationTimeout in the config.inifile.
2. On one node, set the heartbeat timeouts for Oracle Solaris Cluster software.

   phys-schost-1:/ # cluster set -p heartbeat_quantum=500 -p heartbeat_timeout=5000

   ---

   Note - The heartbeat timeout must be half of the ArbitrationTimeout in the config.inifile.

   ---

3. Start the management server.

   phys-schost-1:/ # cd /mgm-data
   phys-schost-2:/ # cd /mgm-data
   
   phys-schost-1:/mgm-data # /usr/local/mysql/bin/ndb_mgmd \
   > --configdir=/mgm-data -f /mgm-data/config.ini \
   --ndb-nodeid=1
   phys-schost-2:/mgm-data # /usr/local/mysql/bin/ndb_mgmd \
   > --configdir=/mgm-data -f /mgm-data/config.ini \
   --ndb-nodeid=2

4. Verify that the management server is running.

   Run the ndb_mgm show command on both nodes until the data nodes are connected to the management server.

   phys-schost-1:/mgm-data # /usr/local/mysql/bin/ndb_mgm \
   > --ndb-connectstring=phys-schost-1-p,phys-schost-2-p -e show
   phys-schost-2:/mgm-data # /usr/local/mysql/bin/ndb_mgm \
   > --ndb-connectstring=phys-schost-2-p,phys-schost-1-p -e show