Your requirements cannot be handled in a single tool with the possible exception of HP's Openview operations center. Out of your list, the online diagnostics will only handle 3 and 4, part of 5 and 7 when the issue is hardware.
Your list:
> 1. Multiple instances of a service running.
This is an undefined requirement. HP-UX will schedule and manage kernel services. Your applications require their own monitor since there is no way for the OS to know what is normal for your codce.
> 2. Service hanged/stopped/terminated abnormally.
OpenView's operation manager can handle these items but it requires extensive setup anjd testing. Oh, it is an extra cost product.
> 3. Failure of device/file system/volume, etc.
> 4. Memory corruption.
The online diagnostics can be configured to do both 3 and 4.
5. Disk corruption.
If the disk has a hardware failure, this is handled by the diagnostics. If the system is missing patches, or a bad program corrupts a filesystem, this requires a special program to test the integrity of the directory. There is no way to test that the contents of any particular file has not been changed incorrectly.
> 6. High CPU/memory utilization for an extended period of time.
OpenView can do this but you will probably need Glance/MWA (another extra cost product) to determine the reason(s) for the load.
> 7. Any other abnormal/critical events affecting system performance/operation
Not defined in a meaningful way.
All of the above requirements are the responsibility of a well-trained and experienced system administrator. By spending a LOT of money for monitoring products, you may in time be able to eliminate the need for a real system administrator, but just make sure that no changes can be made to the system (no new disks or other peripherals, no patches, no changes to the applications, etc).
Bill Hassell, sysadmin
