12C ORA-27106 Error Handling process

The first two days of customer Exadata 12c environment due to memory allocation error, the following is the processing process.

The node restarts with an error, but the other nodes start normally. Error content:

ORA-27106 System pages not available toallocate memory

Alert content:

Supported system pagesize (s):

Fri APR 10 18:07:43 2015

PAGESIZE available_pages expected_pages allocated_pages ERROR (s)

Fri APR 10 18:07:43 2015

2048K 3079 8194 3076 ORA-27125

The judgment is likely to be caused by a lack of distribution of hugepage.

Checked for modified content in the parameter file. Use_large_pages was found to be set. The value of the parameter is three delegates meaning the following:

True indicates that if the system is configured with Hugepage, it will be used, and if not, the SGA can also use the usual page-size memory, which means that the SGA can run in mixed mode.

False indicates that the instance does not use Hugepage.

Only indicates the use of huge.

The Use_large_pages value on the failed node is only. So there is not enough hugepage on the above error situation.

Knowing the root of the problem is easy to solve. Two types of solutions.

1, modify the operating system parameters Vm.nr_hugepages will increase the value.

2, modify the database parameter use_large_pages to TRUE.

After communication, the user side does not specifically require the value of the database parameter, so change it to true after the thing is over.

At this time there are classmates to ask, why the other nodes of the database normal open it?

In fact, there are two databases in this cluster. The second database was created later, and the value of Use_large_pages is the default value of True.

After the failure node restarts, the second instance of the database starts, using most of the hugepage, causing the remaining quantity to be insufficient for the first DB instance to start, and therefore an error.

12C ORA-27106 Error Handling process