DB2 HADR takeover does not work as expected when also using TWS policy
In a TSAMP environment it might come to conflicts when a TWS policy was setup
and db2haicu and/or a "db2 takeover hadr" command may start resources on the same node again
Symptom:
In a TSA MP environment when a TWS policy was setup and then DB2 policy using db2haicu it might be seen that a "db2 takeover hadr" command seems to be successful, but starts resources on the same node as before. The HADR primary/standby do not switch
Cause:
The TWS policy defines affinity relationship against the IBM.PeerNode resources.
This prevents TSA BINDER to find the new location on the other side
Solution:
Delete the affinity relationship:
(resource names depend on current environment, Relationship=3 means Affinity )
1) Lock affected resource groups
rgreq -o lock tws1-rg
rgreq -o lock tws2-rg
2) Remove all Affinity relationships
rmrel -s 'Relationship=3'
NOTE: If you have other Affinity relationships defined then this command cannot be used.
Instead of this use:
rmrel -s "Name='<Relationship name>' "
3) unlock the resource groups again rgreq -o unlock tws1-rg rgreq -o unlock tws2-rg
=> check that no more Affinity relationships exist against IBM.PeerNode resources:
lsrel -P Affinity
4) Restart Master IBM.RecoveryRM daemon
a). identify the node running the IBM.RecoveryRM master
lssrc -ls IBM.RecoveryRM | grep Master
b) on 'master node' kill the process
ps -ef | grep RecoveryRMd
kill -9 <pid-of-IBM.RecoveryRMd-process>