Solaris Cluster HA-Oracle (SUNW.oracle_server) Resource Fails to Start Database due to Error "ORA-27102: out of memory"
(Doc ID 1007002.1)
Last updated on FEBRUARY 07, 2018
Applies to:Solaris Cluster - Version 3.0 to 4.3 [Release 3.0 to 4.3]
Oracle Solaris on SPARC (64-bit)
Oracle Solaris on x86-64 (64-bit)
Oracle Solaris on SPARC (32-bit)
With Solaris 10 and 11, the settings related to shared memory and semaphores are configured by means of projects.
* Each user can be associated to one or more project but a default project exist for each user.
* Each process is associated to a task which is a group of processes running under a project.
* Each project establish limits for a number of system resources jointly consumed by the processes running in any given task.
As an Oracle database (DB) can make use of large amount of shared memory, it is possible that the default settings for the shared memory are just not enough to permit allocation and explicit settings must be setup for a project associated to the user launching the Oracle database.
Failure to configure the project correctly can lead to DB startup failure as described in the following
Solaris Cluster adds some extra steps as you not only have to set up the projects correctly but also have to make sure that the individual database resources are assigned the correct projects.
We assume that the DB have been installed and configured and it is able to startup fine using DBA sqlplus commands. If the DB cannot be started by the DBA using sqlplus, the root cause for this failure must be investigated upfront before going into troubleshooting the Cluster behavior.
When the Database is started by the HA-Oracle, it fails to start with an "out of memory" errors , complaining that shared memory cannot be allocated.
The following error shall appear in the Oracle DB Alert log and/or in /var/opt/SUNWscor/oracle_server/message_log.<oracle_server-rsname>
There are slightly different scenarios which could trigger the startup failure associated to ORA-27102:
- The DB is just installed and configured. A project have been setup for the oracle user, and the DB started using DBA sqlplus command but HA-Oracle cluster resource fails to start it.
- The DB was already installed and configured to run under cluster control without problems, but memory usage as been increased within the Oracle configuration. Even if the DB can still be started using DBA sqlplus commands, HA-Oracle is no longer able to start it.
- The DB is able to start fine with DBA sqlplus commands. HA-Oracle sometimes is able to start the DB just on some nodes and fails on another nodes where other databases are running already.
- Another DB have been added to the configuration or another DB have been reconfigured to start with more memory, the other DB started fine but upon an intentional failover, one DB running under cluster control fails to start.
To view full details, sign in with your My Oracle Support account.
Don't have a My Oracle Support account? Click to get started!