System Monitory and Notification

[Scott Frye_1]

HP-UX 11.0 and 11.i. Currently we use tools like top, bdf, glance ... to monitor our systems. This requires us to manually log into each box and run these commands. I'm looking for best practice (or personal experience) on ways to automate this checking. In addition, I would be very interested in knowing of ways to notify my via pager when levels are too high. I do not know Perl or C.

Thanks for you help.

[Steven E. Protter · Accepted solution]

I'm attaching a script that demonstrates the concept.

Our system monitoring is almost exclusively by cron. There are some manaul checks just in case networking goes off line.

SEP

[A. Clay Stephenson]

Best practices involve using daemons to continually monitor these metrics and then push messages to a central collection mechanism. The messages are then acknowledged or if ignored for a period are then escalated resulting in emails or pages.

You really need a product like OpenView VantagePoint/Operations. VP/O comes with tons of templates and monitors that alert you
and can even take automatic action. If this is a serious production environment, you need a tool like this.

You could certainly do the same thing on your own but to do this right is a daunting project. In fact, even with VP/O, it's still a daunting project to cover as many contingencies as possible. In my own case, for example, if a filesystem is filling up, VP/O can first do cleanup and if more is required it can even extend LVOL's and filesystems within limits. VP/O is now called something like "Operations for UNIX".