[op5-users] check_esx3.pl high CPU load.

[Mattias Bergsten]

On 2010-02-19 14:57, Otto Wincrantz wrote:
> The script takes about 5-7 seconds to execute in my environment, with 15
> virtual hosts, 60 VMs and a 5 minute check schedule I am looking at
> close to a constant 100% CPU load on my monitoring server.

Make sure you're using the latest version of the plugin, there have been 
many performance improvements since it was written. It can be found in 
git at op5.org.

Unfortunately the VMware Perl API is a little strange, and there's not 
always much choice in requesting data. Sometimes all you can do is say 
"give me all the data" and then filter it clientside, which isn't all 
that efficient.

Having said that, there are definitely optimizations to be done to the 
script, and it is something that is on the roadmap (Peter, do you have 
more info on this? We've talked about it but I can't remember what 
timeframe we said except "post 5.0 release"..) but it might take a while 
more until someone at op5 can look at it. If anyone is conversant with 
Perl optimization and would like to take a look and submit some patches, 
by all means feel free - it would be appreciated by many.

Until then, I suggest running as few checks as possible, i.e., if you 
have a VirtualCenter, run _one_ VMFS check against the VirtualCenter to 
get the data for all hosts and all filesystems. I realize it's not a 
perfect solution but at least it's a workable one. :)

(hälsa Steve!)

/mattias, who does not work at op5 but helped create check_esx3