Did you set the oracle user's limits according to the docs ? You can verify with ulimit -a Cheers stefan On 3/28/07, William Wagman <wjwagman@xxxxxxxxxxx> wrote:
Greetings, I just upgraded one of our Linux boxes. It is now running 32-bit RHEL 2.6.9-42.Elsmp. The box has 16GB of memory. I installed 32-bit Oracle 10.2.0.3.0. I am encountering ORA-27102, out of memory errors when starting the database and I don't understand why. The kernel parameters are set correctly, I think... [root@willie ~]# /sbin/sysctl -a | grep sem kernel.sem = 250 32000 100 128 [root@willie ~]# /sbin/sysctl -a | grep shm vm.hugetlb_shm_group = 0 kernel.shmmni = 4096 kernel.shmall = 2097152 kernel.shmmax = 4294967295 [root@willie ~]# /sbin/sysctl -a | grep file-max fs.file-max = 65536 [root@willie ~]# /sbin/sysctl -a | grep ip_local_port_range net.ipv4.ip_local_port_range = 1024 65000 [root@willie ~]# /sbin/sysctl -a | grep rmem_default net.core.rmem_default = 262144 [root@willie ~]# /sbin/sysctl -a | grep rmem_max net.core.rmem_max = 262144 [root@willie ~]# /sbin/sysctl -a | grep wmem_default net.core.wmem_default = 262144 [root@willie ~]# /sbin/sysctl -a | grep wmem_max net.core.wmem_max = 262144 With these init parameters (which work by the way on a 64-bit server with 64-bit oracle installed, same versions and only 10GB of memory) I encountered the ORA-27102 on startup... edrs.__db_cache_size=788529152 edrs.__java_pool_size=33554432 edrs.__large_pool_size=16777216 edrs.__shared_pool_size=1275068416 edrs.__streams_pool_size=16777216 *.db_block_size=8192 *.db_file_multiblock_read_count=16 *.job_queue_processes=10 *.open_cursors=300 *.pga_aggregate_target=1560281088 *.processes=300 *.sga_max_size=3221225472 *.sga_target=2147483648 In order to start the database I had to change the parameters to this... edrs.__db_cache_size=50000000 edrs.__java_pool_size=33554432 edrs.__large_pool_size=16777216 edrs.__shared_pool_size=250000000 edrs.__streams_pool_size=16777216 *.cursor_sharing='EXACT' *.db_block_size=8192 *.db_file_multiblock_read_count=16 *.job_queue_processes=10 *.open_cursors=300 *.pga_aggregate_target=25000000 *.processes=300 *.sga_max_size=250000000 *.sga_target=250000000 The one which seems to make the most difference is the sga_target parameter. This database was created by restoring an RMAN backup from the production server. So I am completely at a loss to explain why, on this box with more memory than the other I have to reduce the memory allocation in order to start the database. Any insight would be greatly appreciated. Thanks. Bill Wagman Univ. of California at Davis IET Campus Data Center wjwagman@xxxxxxxxxxx (530) 754-6208 -- //www.freelists.org/webpage/oracle-l