Failed to Log In to MRS Manager After Changing the Domain Name

Symptom

After changing the domain name, the user cannot log in to MRS Manager through the console, or fails to log in to MRS Manager.

Cause Analysis

After the domain name is changed, the keytab file of user executor is not updated. As a result, the executor process repeatedly performs authentication after the authentication fails, causing memory overflow of the ACS process.

Procedure

1. Restart the ACS process.
   1. Log in to the active management node (master node marked a solid star on the Nodes tab of the MRS cluster) as user root.
   2. Run the following commands to restart the acs process:

      su - omm

      ps -ef|grep =acs (Query the PID of the acs process.)

      kill -9 PID (Replace PID with the acs process ID to kill the acs process.)

   3. Wait for several minutes and run the ps -ef|grep =acs command to check whether the acs process is automatically started.

2. Replace the keytab file of user executor.
   1. Log in to MRS Manager and choose System > User. In the Operation column where user executor resides, click Download Authentication Credential. Decompress the package to obtain the keytab file.
   2. Log in to the active management node as user root and replace the /opt/executor/webapps/executor/WEB-INF/classes/user.keytab file with the file obtained in 2.a.

3. Replace the keytab and conf files of user knox.
   1. Log in to MRS Manager and choose System > User. In the Operation column where user knox resides, click Download Authentication Credential. Decompress the package to obtain the keytab and conf files.
   2. Log in to the active management node as user root and replace the /opt/knox/conf/user.keytab with the file obtained in 3.a.
   3. Change the principal value in the /opt/knox/conf/krb5JAASLogin.conf file to the new domain name.
   4. Replace the /opt/knox/conf/krb5.conf file with the krb5.conf file obtained in 3.a.

4. Back up the original client directory.

   mv {Client directory} /opt/client_init

5. Reinstall the client. For details, see Updating a Client.
6. Log in to the active and standby management nodes as user root and run the following commands to restart the knox process:

   su - omm

   ps -ef | grep gateway | grep -v grep (Search for the PID of the knox process.)

   kill -9 PID (Replace PID with the ID of the knox process to kill the knox process.)

   /opt/knox/bin/restart-knox.sh (Start the knox process.)

7. Log in to the active and standby management nodes as user root and run the following commands to restart the executor process:

   su - omm

   netstat -anp |grep 8181 |grep LISTEN (Search for the PID of the executor process.)

   kill -9 PID (Replace PID with the ID of the executor process to kill the executor process.)

   /opt/executor/bin/startup.sh (Start the executor process.)