Recommendations for Process and Service Monitoring

When you are configuring process and service monitoring, consider the following:

Monitor the following:
- Status of the processes:
  - Use the processState (operating system dependent) or processStateStr (operating system independent) OIDs to return the process state. For more information, see empire.asn1.
  - For processes that run through an interpreter such as Perl, set the 0x00000800 flag to match the process name and arguments (UNIX only) when creating process monitoring entries. For more information, see Process Monitor Table Flags.
- CPU time over interval (if this value is not incrementing, the process might be in the zombie state.)
- Total CPU time (a high value can indicate problems)
- Leaking memory (RSS over time; use alarm thresholds to notify you of problems)
Configure SystemEDGE to automatically restart failed processes.
Set your UNIX application shutdown files to disable process and service monitoring to prevent race conditions.
Use the Process Group Monitor table to monitor multiple processes with the same name. For more information, see the chapter “Process Group Monitoring.”