Start a conversation

Recovering an MGR Cluster With No Master

You may encounter a situation where you have a multi-MGR configuration and both MGRs are stuck in the slave configuration, possibly after a failed failover and failback.

You may notice output like this when attempting to start either MGR as the master:

[root@MGR ~]# /usr/TextPass/bin/tp_master
Starting with clean XML cache
Peer is already running as master

You can verify this by running tp_role on each server, and observing that both indicate the slave role.

Step 1:  Verify the Problem (Read-Only Check)

On both MGR-01 and MGR-02, check:

cat /usr/TextPass/etc/role

If both show "master", that confirms the issue: both nodes think they are the master per the configuration.

Step 2: Stop the MGR on BOTH Nodes

On BOTH MGR-01 and MGR-02

/usr/TextPass/bin/tp_mgr_stop

Verify that the processes have stopped

systemctl list-units --all | grep mgr

mgr.service should show exited and tp_mgr_poll and tp_mgr_sync should not appear.

Step 3: Decide Which Should Be Master

Choose your designated master (e.g., MGR-01).

Step 4: Set the Role Files Correctly

On MGR-01 (designated master):

echo "master" > /usr/TextPass/etc/role

Verify:

cat /usr/TextPass/etc/role

On MGR-02 (designated slave):

echo "slave" > /usr/TextPass/etc/role

Verify:

cat /usr/TextPass/etc/role

Step 5: Start Master First

On MGR-01 only:

/usr/TextPass/bin/tp_master

Verify it started successfully:

systemctl list-units --all | grep mgr

mgr.service should show exited and tp_mgr_poll and tp_mgr_sync should appear as running.

Step 6: Wait and Verify Master is Stable

Wait 30-60 seconds and confirm MGR-01 is running normally as master.

Step 7: Start Slave

On MGR-02:

/usr/TextPass/bin/tp_slave

Verify it started successfully

/usr/TextPass/bin/tp_status

Step 8: Verify HA Pair Status

On both nodes, verify:

Check the role file:

cat /usr/TextPass/etc/role

Check status:

/usr/TextPass/bin/tp_status

If the above instructions do not resolve the issue, it may be possible that there is an issue with the peerfserver configuration in the /etc/hosts file on each MGR server.

Choose files or drag and drop files
Was this article helpful?
Yes
No
  1. Matthew Mrosko

  2. Posted

Comments