Problems with big SGAs (>200G) on Linux

I recently had an issue where a database with 240GB SGA (1 huge shared memory segment) configured with hugepages on a system with 512G RAM was suddenly becoming instable and new logons were denied with these error message:


ORA-01034: ORACLE not available
ORA-27123: unable to attach to shared memory segment
Linux-x86_64 Error: 22: Invalid argument
Additional information: 2667
Additional information: 1736718
Additional information: 215016800256

This was strange because ipcs -a showed all shared memory segments and all processes existed. It turned out that there is a known issue with the Linux feature “prelink” and very big SGAs (>200g). This MOS Note gives some details:

Connected to an idle instance, while database instance is running (Process J000 died, see its trace file kkjcre1p: unable to spawn jobq slave process ) (Doc ID 1578491.1)

Obviously, exadata customers also suffered from this issue and therefore prelink is disabled beginning with exadata version 11.2.3.3. Redhat is reporting the same issue in combination with Informix databases in Knowledgebase article: https://access.redhat.com/solutions/1186953

Leave Comment