Removing Clustered Installations

Clustered or high-availability (HA) Manager installations are no longer supported for Manager software installations. Converting an HA installation to a standard installation requires a recent Manager backup to ensure safe transfer to a single Manager.

Once a backup is completed, record the following from the HA/Clustered Manager for reference:

• Manager backup (e.g. DTM.<file_name>.jar)
• HA Manager virtual hostname
• The UIDs of the transmgr, transusr and any other accounts used in Signiant jobs
• the DTM group GID

Note: Custom templates or jobs running directly on the Manager may use other accounts or group IDs. These accounts or groups must be included in the new deployment.

To convert an HA Manager:

1. Install a Manager on a machine that uses the cluster virtual hostname as its hostname so that Signiant services can function. The installation path must be the same as the current cluster deployment.

2. Ensure that UID for transmgr, transusr and any other account used by the Manager match. Also make sure that the GID for the DTM is the same.

   Note: If it is necessary to change the GID of the DTM group be sure to change the group for all directories and files in the installation directory (/usr/signiant/dds) that are owned by the DTM group. Ensure the transmgr and transusr are members of the DTM group. These changes can be made by editing the /etc/passwd and /etc/group files.

3. Create a new temporary directory in the /tmp on the new Manager and expand the clustered Managers backup file:

   $ /path/to/manager/3rdparty/jdk/CurrentVersion/bin/jar -xvf <filename>.jar

   Note: Keep a copy of the clustered Managers backup file safely outside this temporary directory.

4. Remove the settings for DTM_CLUSTER and DTM_CLUSTER_MEMBERS in the signiant.ini file.

5. Compress the files from the temporary directory into a new .jar file using the following command:

   $ /path/to/manager/3rdparty/jdk/CurrentVersion/bin/jar -cvf /tmp/newbackup.jar *

   Note: Make sure a copy of the original clustered Manager backup file doesn't get included in the new backup file.

6. Restore the Manager using the new backup file using the following command:

   $ restore_dtm -r /tmp/newbackup.jar

Once the restore_dtm command completes, thoroughly test your Manager to ensure normal operation.