Is there a published list of best practices from SUN for setting alarm thresholds? In particular I'm interested in hardware monitoring. For example:
1. memory DIMMs will be replaced by Sun if the count of non-intermittent ECC is over 24 in 24 hours.
2. Hard Disk Error count >1 should raise an alarm.
Thanks.
Hi Mike,
> Is there a published list of best practices from SUN
> for setting alarm thresholds? In particular I'm
> interested in hardware monitoring. For example:
Hardware alarms (i.e. coming from the "Config Reader" line of modules) have their hardware thresholds already set by Sun. i.e. you don't have to specify good ranges of temperatures, fan RPM, voltages or conditions of memory/devices since it's already been done for you. In fact, many of the hardware alarms can't be changed from the SunMC Console at all.
If you want to know some of things that are covered, click on the "Module Manager" tab for an Agent, click/highlight the Config Reader [X] module in the "Modules with Load Status:" table, then click the "Rules.." button. You may see some of the thresholds, but more likely see they are hidden in named rules (i.e. "rcr4u219"). You can dig those rules up by searching docs.sun.com.
As for other modules, some also ship with out-of-the-box thresholds (i.e. filesystem % full thresholds in Kernel Reader) or with rules (i.e. Healh Monitor module is full of them). But for generic kernel/log/process-type modules you have to tune the thresholds yourself. And those thresholds are going to be different for every system SunMC runs on.
Regards,
Mike.Kirk@HalcyonInc.com
