To start shmtimer

The UNIX time() function is a system call often called in database servers. On multiprocessor systems, use of the time() function can be quite heavy. Even though a process can run successively on different processors, this must be invisible to the process. The process requires that each processor return the same time, using time(). Therefore, the processors must synchronize their internal clocks each and every time the time() function is called.

The shmtimer program prevents this overhead. shmtimer is a daemon process that writes the current time in shared memory. The time is updated every second, which reduces the number of calls of the time() function to a maximum of one per second.

When you initialize shared memory (shmmanager –i), the shmtimer program starts. If you remove shared memory, shmmanager –k, the shmtimer stops first. In shmtimer these options are available:

• –i: Initialize shmtimer

  A shmtimer program is started, and runs in the background.

• –k: Kill shmtimer

  The running shmtimer program is killed. The allocated shared memory is cleared, but not removed.

• –s: Show status as stored in shared memory:
  • The current time
  • The process ID of the running timer
• –u: Show information about shmtimer
• –v: Show version and porting data of shmtimer

If shmtimer stops, the shared memory is cleared, as described for the –k option. From then on, the UNIX time() function is called, rather than reading the time from shared memory.

If shmtimer is killed, shmtimer cannot clear its shared memory. Because shmtimer is killed, the time in shared memory is never updated, but the time is still read from shared memory. In that case, shmtimer –s returns a warning. You can start a new shmtimer with shmtimer –i.