Setting-Up the Oracle Memory Speed File System

After configuring the system components, set up the Oracle Memory Speed (OMS) file system by creating and mounting the file system.

About the OMS Daemon

OMS is a user-space file system. The OMS daemon handles management and CLI operations for OMS.

You should run the OMS daemon at start up. At start up, the OMS daemon creates a shared memory segment for the exclusive use of OMS. There is one daemon per instance. Set the instance identifier (SID) before you start the OMS daemon.

Start the OMS daemon by running the $ORACLE_HOME/bin/oms_daemon.

$ whoami
oracle
$ export ORACLE_SID="test"
$ $ORACLE_HOME/bin/oms_daemon
Starting daemon as a detached background.
OMS binary located at /mnt/data/ssd0/obase/test/bin/oms_daemon
OMS daemon startup: oms_test successfully created
OMS daemon creating tracefile
/mnt/data/ssd0/obase/diag/oms/test_oms_20758.trc
$ ipcs
------ Message Queues --------
key msqid owner perms used-bytes messages
------ Shared Memory Segments --------
key shmid owner perms bytes nattch status
0x00000000 32047104 oracle 640 135266304 1
$ ls -l $ORACLE_HOME/dbs/*.ora
-rwx------ 1 oracle dba 93 Feb 23 20:10
/mnt/data/ssd0/oracle/test/dbs/oms_test.ora

In this example, the daemon creates a shared memory segment with shared memory id 32047104. It also creates a configuration file $ORACLE_HOME/dbs/oms_test.ora.

Oracle recommends that you should not delete the shared segment manually. Similarly, you should not edit or delete the configuration file. If you accidentally kill the daemon, you can restart it by running the command $ORACLE_HOME/bin/oms_daemon.

Daemon traces are available in the $ORACLE_BASE/diag/oms directory.

Creating and Mounting the Oracle Memory Speed File System

Learn how to create an Oracle Memory Speed (OMS) file system and mount it.

Create the mount directories and remove the write and execute permissions:

$ mkdir /oracle/omsfs_1
$ chmod -wx /oracle/omsfs_1
$ mkdir /oracle/omsfs_2
$ chmod -wx /oracle/omsfs_2
$ ls -l /oracle

dr--r--r-- 2 oracle dba 6 Feb 23 21:36 omsfs_1
dr--r--r-- 2 oracle dba 6 Feb 23 21:36 omsfs_2

Start the omsfscmds utility and create and mount the file systems using the mkfs and mount commands respectively.

$ $ORACLE_HOME/bin/omsfscmds
OMS> mkfs /mnt/pmem0/omsuberfile.test
OMS:mkfs:No blocksize specified, using 4K
OMS:mkfs: Device /mnt/pmem0/omsuberfile.test formatted with
blocksize 4096
OMS> mount /mnt/pmem0/omsuberfile.test /oracle/omsfs_1
OMS:mount: Mounted /mnt/pmem0/omsuberfile.test at
/oracle/omsfs_1
OMS> mkfs /mnt/pmem1/omsuberfile.test
OMS:mkfs:No blocksize specified, using 4K
OMS:mkfs: Device /mnt/pmem1/omsuberfile.test formatted with
blocksize 4096
OMS> mount /mnt/pmem1/omsuberfile.test /oracle/omsfs_2
OMS:mount: Mounted /mnt/pmem1/omsuberfile.test at
/oracle/omsfs_2
OMS> lsmount
fsindex : 0
Mountpt : /oracle/omsfs_1
Deviceid: /mnt/pmem0/omsuberfile.test
fsindex : 1
Mountpt : /oracle/omsfs_2
Deviceid: /mnt/pmem1/omsuberfile.test

Currently, the only supported block size for OMS file systems is 4 KB. In this example, the two OMS file systems are mounted and available for use. /dev/pmem0 resides on socket 0, hence files placed in /oracle/omsfs_1 reside in socket 0. Similarly, /dev/pmem1 resides in socket 1, hence files placed in /oracle/omsfs_2 are in socket 1.

You can mount a maximum of two file systems simultaneously. This enables OMS to use PMEM devices on two socket servers.

OMS mount points are automatically remounted if you reboot the node and restart the daemon.