Hi:
I need create alerts to a lot of servers, like below:
Can you tell me what is the best practice to create these alerts ?
i would create a custom property against the nodes table for
CPU, Memory and up/down
your node down will contain a condition to match the monitored yes/no property
your CPU and RAM alerts will contain a complex trigger condition for
field cpu-utilisation is greater than field custom-property-for-cpu
same for RAM
for the volume utilisation, use the volumes table and add a custom property for disk space, then do the same thing fro the disk space alert
any probs, give me a shout
We implemented a strategy similar to this. Custom properties were set up for CPU, Memory, Disk and Network Interfaces. We then assign alert 'package' codes to groups of alerts that the we created and include the package code name in the trigger logic. It is then just a matter of adding these package codes to the nodes that we want to alert on. This allows us to not only manage different alerts/trigger levels but also what the alerts do, for example if for some servers we want to create a service desk ticket at a medium priority and others a high priority when CPU Load is too high we can just create 2 different alert 'packages' (CPUTICKETMED and CPUTICKETHIGH). We have found that this approach gives us a significant amount of flexibility in they way that we manage alerting.
I dont know best practice but I would suggest custom properties and then make alerts that use the custom properties.
I have a similar thing for disk volumes... they all have a custom property, if this property is empty it alarms when more than 80% of the disk space is used, otherwise it alarms once the custom threshold is surpassed. I have also written a custom report which shows volumes in "warning" state (e.g. 80% used, while 90% would be critical and alarmed upon).
You could do the same thing for CPU and memory...
Hi Valermirs--
I'm going to move this to the Alert Lab forum so it can get even more great feedback on it. Also, have you read Andy McBride's paper on advanced alerts? It provides alot of information.
HTH,
M
