Oracle 10G RAC stability problems

We've just installed a two node 10G RAC environment under Linux AS 3.0 for EM64T (Intel's new chip) with shared storage on an EMC box and got the thing running. However, it is not stable. From the logs, we see that one node misses 60 check-ins which causes the remaining node to start an eviction. As the last part of the eviction the remaining node trys to reboot the "dead" node but the "dead" node just hangs. After a bit the remaining node also fails.

Has anyone else seen this behavior? We've doubled checked our network and IP assignments for public and private subnets and have the latest versions of CRS and OCFS. We are not running ASM.

Any ideas?

Thanks in advance.

Stephen Received on Thu Aug 11 2005 - 04:22:35 CEST