Change RTIF_56 checkSMS to restart instead of shutdown?

[Peter Heggie]

I think it was release 5.8 that brought in restart commands, in addition to shutdown and start, for various objects like the monitor, threads and processes.

For Cloverleaf users connecting to Invision through SNA, I’m wondering what the downside would be to changing the logic in checkSMS / rtif:shutdown to invoke a restart instead of a shutdown of the process.

The current code is:

Code:

##################################################################
   # shutdown – Received a dellocate or shutdown command
   #
   # Called:
   # shutdown
   #
   # Notes:
   # Issue an error message and issue shutdown for the
   # process.
   #
   # Note that process is shut as the thread shutdown
   # would hang as SNA blocks.
   #
   # Returns:
   # Nothing significant
   ##################################################################

   proc shutdown {} {
global HciConnName

set procname [file tail [pwd]]

set errmsg “Received SHUT or dealloc from RTIFnStopping”
   the $procname process ….nn”

debug $errmsg
exec hcienginestop -p $procname &
   
return “”
   }

We are thinking of changing the system command to:

Code:

exec hcienginerestart -d 30 -p $procname &

We can no longer rely on the Invision system shutting down at the same time every night, and our cron recycle scripts can’t always recycle the Cloverleaf processes within the Invision recycle. One or more processes can get shutdown if the other side of the SNA connection is stopped. The Cloverleaf process will stay shutdown until the scheduled Cloverleaf recycle. The operators get anxious when the threads are down for more than a minute, even at 1am.

Would appreciate any input on possible downsides of automatically restarting a Cloverleaf process when one of its threads’ SNA connection shuts down.

Peter Heggie
PeterHeggie@crouse.org

Topic