How to monitor a pool of processes

[Roy Colica]

All,
I've the following trouble and I'm wondering if someone can help.
I have to monitor a pool of processes already started with a new process I've to implement in order to restart a process if dies.
If the death process was a son of my process I can manage it with a SIGCHL but as this is not the case I want understand if it's mandatory to loop on it to check if still running or I can do something in order to be waked by the system when it dead.
Do you think exists something like a "process group" I can setup and then have info by the system on the processes belonging to it?
Thanks for your help and suggestions.
Roy

[G. Vrijhoeven]

Hi Roy,

You could take a look at MC/SG this enables you to guard services. If the service is not available you can let MC/SG restart the proc if this happens you can do a restart n times. n+1 can also trigger a restart of the service on an other node. But this software is not free and requires a HA env. to be usefull ( no SPOF)

You can also create a script that checks if a proc is running, and if so if its parant is alive, and if do if a proc is running with PPID -eq 1 and if not restart the Parant.


HTH,

Gideon

[Roy Colica]

Thanks.