November 1, 2007

Sometimes our customers come to us for tailored solutions to meet a specific need. Recently I got a call from one of our customers who wanted to be alerted when their servers were rebooting, so they can track the times at which reboots are occurring. Although Longitude does detect servers that are down, there are cases in any interval-based monitoring situation where a server can reboot without being caught because it is not down long enough to warrant the trigger of a “server down” alert. However, in this case, the customer needed to know about every single reboot on their Windows and Unix servers, even reboots that cause only momentary interruptions.

First we helped them deploy a custom solution for Windows that reads a value in WMI and creates an event whenever a reboot occurs. At each interval it checks the reboot time and compares it to what it observed in the previous interval. If there is a difference in value, Longitude knows there was a reboot and sends the alert.

Unix doesn’t give us anything as easy as a WMI value, but it’s still simple. For their Unix servers we created a Unix_Script Transaction that runs the following command: cut –d” “ –f1 /proc/uptime