My Oracle Support Banner

Oracle VM: How To Remove A Node From A Live OVM Server Pool Without Rebooting (Doc ID 2169757.1)

Last updated on AUGUST 04, 2018

Applies to:

Private Cloud Appliance - Version 2.0.5 and later
Oracle VM - Version 3.2.1 and later
Linux x86-64
In many cases, customer is not allowed to reboot the whole server pool/ocfs2 cluster to eliminate the discrepancy due to critical produ

Goal

Demonstrate how to remove a node from an active cluster without any rebooting.  This technique applies to clusters using OCFS2 1.8 and after.  The cluster must use global, not local, heartbeats.

For clusters using OCFS2 prior to 1.8, please refer to Oracle KM Note 852002.1 852002.1

 The procedure may be needed if the on-disk state information for OCFS2, contained in the file /etc/ocfs2/cluster.conf, becomes out of sync with the in-memory OCFS2 state contained in the /sys/kernel/config/cluster/<cluster-name>/node/ virtual file system maintained by the OCFS2 file handler.  Normally these data sources are kept in-sync but should an unexpected system event such as a storage power loss occur, the two states may become inconsistent.  

Solution

To view full details, sign in with your My Oracle Support account.

Don't have a My Oracle Support account? Click to get started!


In this Document
Goal
Solution
 Indicators This Process May Be Needed
 How To Perform The Procedure
 Step 1: observe the cluster definition in /etc/ocfs2/cluster.con
 Step 2: observe the in-memory OCFS2 configuration
 Step 3: make sure the node you want to remove is not on-line in the server pool cluster
 Step 4: use the o2cb service to set ALL cluster nodes on-line
  Step 5: verify the in-memory OCFS2 configuration is correct
References

My Oracle Support provides customers with access to over a million knowledge articles and a vibrant support community of peers and Oracle experts.