Cleaning Shared Memory Segment and semaphore, for the ASE startup

When we shutdown the Sybase dataserver abruptly, by using the kill -9 or kill -12(or due to any reason automatically).

During the restart, sometime it gives the problem unable to create shared memory region, as below:


os_create_region: shmget (0x%x): %s

os_create_region: Shared memory segment %d is in the way

os_create_region: uninitialized shared memory descriptor

os_create_region: shmat (%d): %

Killing Process may left behind the System V semaphore or shared memory left behind instead of being cleaned up automatically. To eliminate unneeded semaphores or shared memory segments, we need to release semaphore or shared memory segment manually.

For the Semaphores:

Run the command ipcs -sa, it will give all semaphores active in the UNIX box, look for the value "0" in the column NSEMS. 0 values indicate that unused semaphore, find all semaphore for the Sybase user and remove them using command ipcrm -s

testinghostname:(/sybase)=>ipcs -sa
IPC status from as of Thu Sep 17 08:50:11 EST 2007
T ID KEY MODE OWNER GROUP CREATOR CGROUP NSEMS OTIME CTIME
Semaphores:
s 201326671 0xc103d804 --ra-ra-ra- patrol dba patrol dba 2 8:45:15 0:00:08
s 201326668 0xc103d80e --ra-ra-ra- sybase dba sybase dba 0 8:50:08 0:00:00


testinghostname:(/sybase)=>ipcrm -s 201326668

For the Shared Memory:

To cleanup the Shared memory segment, which are assigned at OS level but not in use, run ipcs -ma command.

testinghostname:(/sybase)=>ipcs -ma
IPC status from as of Thu Sep 17 08:50:24 EST 2007
T ID KEY MODE OWNER GROUP CREATOR CGROUP NATTCH SEGSZ CPID LPID ATIME DTIME CTIME
Shared Memory:
m 654311446 0x4103d80e --rw-rw-rw- patrol dba patrol dba 0 2304 1766 1766 8:50:18 8:50:18 0:00:08
m 671088657 0x4103d80d --rw-rw-rw- sybase dba sybase dba 0 1792 1766 1762 8:45:15 8:45:15 0:00:08
m 671088650 0x4103d80b --rw-rw-rw- root root root root 0 768 1766 1766 8:50:18 8:50:18 0:00:08
m 671088648 0x4103d80a --rw-rw-rw- sybase dba sybase dba 1 1280 1766 1766 8:50:18 8:50:18 0:00:08
m 654311548 0x4103d809 --rw-rw-rw- sybase dba sybase dba 0 256 1766 1766 8:50:08 8:50:08 0:00:08


Now look in the NATTCH field for 0, if it 0 zero for Sybase user , it is unneeded, but still assigned at OS level. Remove it ipcrm -m by providing the id.

testinghostname:(/sybase)=>ipcrm -m 671088657 -m 654311548

Now Start the dataserver.. :)

Same case with backup server.