Introduction
Oracle Data Guard Broker simplifies switchover and failover operations by automating many of the underlying database and Clusterware tasks. However, during a planned switchover, administrators may encounter the following error:
DGMGRL> SWITCHOVER TO STANDBYDB;
Performing switchover NOW, please wait...
Error: ORA-16535: Oracle Restart or Oracle Clusterware prevented completion of broker operation
At first glance, the error suggests a problem with Oracle Restart or Oracle Clusterware. However, in one of our recent implementations, the root cause was related to the database instance name length rather than an actual Clusterware failure.
This article explains the issue and how it was resolved.
The Problem
While performing a Data Guard switchover through DGMGRL, the operation failed with the following message:
Performing switchover NOW, please wait...
Error: ORA-16535: Oracle Restart or Oracle Clusterware prevented completion of broker operation
The databases were healthy, Data Guard synchronization was normal, and no obvious Clusterware issues were reported. Despite this, the broker was unable to complete the role transition.
Investigation
The first step was to verify the broker configuration:
DGMGRL> SHOW CONFIGURATION;
The configuration appeared valid. Next, Clusterware resources and broker logs were reviewed for additional clues.
After further investigation, it was discovered that the database instance name was unusually long. Oracle Data Guard Broker interacts with Oracle Clusterware to manage database resources during switchover operations, and long instance names can sometimes create limitations when Clusterware attempts to update or manage these resources.
Example of a problematic instance name:
INSTANCE_NAME=PRODUCTIONDBPRIMARY01
Although the database itself operated normally, Data Guard Broker failed when performing Clusterware-related actions required during the switchover.
Solution
The issue was resolved by shortening the database instance name.
Check Current Instance Name
SHOW PARAMETER instance_name;
Update the Instance Name
Modify the parameter to a shorter value:
ALTER SYSTEM SET instance_name='PROD1' SCOPE=SPFILE;
Restart the Database
SHUTDOWN IMMEDIATE;
STARTUP;
Retry the Switchover
DGMGRL> SWITCHOVER TO STANDBYDB;
After the change, the switchover completed successfully without any broker or Clusterware errors.
Why This Happens
During Data Guard Broker operations, Oracle communicates with Oracle Restart or Oracle Clusterware to manage database resources, services, and role transitions.
If resource names become excessively long due to lengthy database or instance names, internal naming limitations may be reached, causing broker operations to fail with ORA-16535 even when the underlying database is healthy.
Because the error message points to Clusterware, administrators may spend significant time investigating CRS resources while overlooking naming-related issues.
Best Practices
- Keep database and instance names concise.
- Follow Oracle naming standards when designing Data Guard environments.
- Validate Data Guard Broker operations immediately after configuration.
- Review broker logs and Clusterware logs whenever ORA-16535 occurs.
- Consider instance naming conventions when deploying RAC, Oracle Restart, or Data Guard configurations.
Conclusion
ORA-16535 is commonly associated with Oracle Restart or Oracle Clusterware issues, but the root cause may not always be a Clusterware failure.
In this case, the error occurred during a Data Guard switchover because the database instance name exceeded practical limits for broker-managed Clusterware operations. Shortening the instance name resolved the issue immediately and allowed the switchover to complete successfully.
When troubleshooting ORA-16535 during Data Guard Broker operations, don’t forget to verify your database and instance naming conventions in addition to checking Clusterware health.
