C.3.33. Replicator fails to connect after updating password

Last Updated: 2013-11-01

Condition or Error

Tungsten Replicator fails to connect after changing the tungsten user password.

Causes

  • The most likely cause is that the configuration within ~/.my.cnf was forcing a connection to the cluster as tungsten user, and user change may have only been made on one host and not replicated to the other MySQL servers.

Rectifications

  • First, update the credentials in ~/.my.cnf and ensure you can connect to all the Replicas with the updated credentials.

    Also check that tpm has been configured with the right password and that all servers have the right information. Errors such as:

    ERROR>>host1>>Unable to connect to the MySQL server using » 
     tungsten@host1:3306 (WITH PASSWORD) (MySQLLoginCheck)

    Indicate that the password may not have been replicated properly. Check the following:

    1. Check the user configuration information within each MySQL server and compare the values:

      mysql> select * from mysql.user where user='tungsten';
    2. For any node that is not up to date, update the password manually:

      shell> mysql -u root -ppassword -P 3306 -h host1 
      mysql> UPDATE `mysql`.`user` SET Password=PASSWORD('secret') WHERE User='tungsten';
      mysql> flush privileges;
    3. Update the tpm and Tungsten Cluster configuration:

      shell> ./tools/tpm update alpha --datasource-password=secret
    4. Restart the replicators:

      shell> replicator restart

      Then put the replicators offline/online to refresh the configuration:

      [LOGICAL] /alpha > datasource host1 offline
      DataSource 'host1@alpha' is now OFFLINE 
      [LOGICAL] /alpha > datasource host1 online 
      Setting server for data source 'host1' to READ-ONLY