How to forcibly detach a node from a TDV cluster.
Article ID: KB0081957
Updated On:
| Products | Versions |
|---|---|
| TIBCO Data Virtualization | 7.0.0 and later |
Description
Your cluster has gotten into an inconsistent unmanageable state so that it doesn't respond to the browser-manager's commands to remove a node or re-group itself.
You then decide to dismantle the cluster and recreate it but cannot dissociate some of the nodes apart.
You then decide to dismantle the cluster and recreate it but cannot dissociate some of the nodes apart.
Issue/Introduction
This article discusses cluster maintenance.
Resolution
This procedure should only serve as a last resort, not as a regular maintenance strategy.
=============================================================
Let N be the problematic server node.
1. Log directly into the Postgres repository db of node N. You can use Psql (see article# 000034150) or a Postgres pgAdmin tool.
2. Manually use SQL to delete all rows in the system_cluster_members table.
3. In node N's /conf/server/server_values.xml file, edit and remove the values from these config items:
/server/config/cluster/name
/server/config/cluster/serverId
/server/config/info/displayName
(the values will then be nulled, appearing like this: <common:value></common:value> )
4. Restart N.
N should now startup as a standalone node.
For the rest of the nodes in the group, try removing node N from their lists using the web manager.
If that does not work, try removing N manually by removing its entry from the system_cluster_members table of every remaining node in the cluster.
=============================================================
Let N be the problematic server node.
1. Log directly into the Postgres repository db of node N. You can use Psql (see article# 000034150) or a Postgres pgAdmin tool.
2. Manually use SQL to delete all rows in the system_cluster_members table.
3. In node N's /conf/server/server_values.xml file, edit and remove the values from these config items:
/server/config/cluster/name
/server/config/cluster/serverId
/server/config/info/displayName
(the values will then be nulled, appearing like this: <common:value></common:value> )
4. Restart N.
N should now startup as a standalone node.
For the rest of the nodes in the group, try removing node N from their lists using the web manager.
If that does not work, try removing N manually by removing its entry from the system_cluster_members table of every remaining node in the cluster.
Feedback
Yes
No
Powered by
