Availability collection interval defaults

Jay Shaughnessy jshaughn at redhat.com
Thu Mar 8 19:10:57 UTC 2012



On 3/5/2012 3:41 PM, Jay Shaughnessy wrote:
>
>
> On 3/5/2012 3:16 PM, Heiko W.Rupp wrote:
>> Am 05.03.2012 um 17:58 schrieb Jay Shaughnessy:
>>> The more I think about it the more I like the idea of a 10 minute 
>>> Service default.  This gives us a blanket performance enhancement 
>>> out of box, without, I
>>> ed intervals less than 5 minutes.  I think perhaps the best scenario 
>>> is a small number of relevant resources with fairly short intervals, 
>>> and the rest with larger intervals.
>>
>> Importance has (almost) nothing to do with platform/server/service.  
>> A datasource may be much more critical than the process itself.
>> Platforms with 100days+ uptime kill a lot of stuff when they fail. 
>> But a badly written application can fill up the datasource pool
>> quickly or fill memory quickly (gc stats are services), while the 
>> base java process will just not die.
> I agree and that is why the plugin descriptors should set overrides as 
> needed.  If the component code can perform fast avail checks, and the 
> resource seems like a critical failure point, then I think the default 
> interval should be relatively small.    And certainly the users can 
> adjust as they see fit.  But the idea here is to allow flexibility 
> while also being conservative out of box,  for the sake of overall 
> agent performance.   The server/service category approach is blunt but 
> it does give us a way to group most resources (services), with a 
> longer interval OOB.
>>
I have updated the default service interval to 10 minutes for the 
reasons stated above.  This is not set in stone, it's still in a feature 
branch, but will be used until we prove or decide it should be 
otherwise. The check-in comment:

Set the default avail check interval for Services to 10 minutes (from 5).
This gives us an immediate perf boost out of box. It is up to the plugin
descriptors to set lower (or higher) intervals for types they feel
should be checked more (or less) frequently, or disabled completely.
past that, users can change an avail check interval at any time, for any
type/group/resource.

>>
>>
>
> _______________________________________________
> rhq-devel mailing list
> rhq-devel at lists.fedorahosted.org
> https://fedorahosted.org/mailman/listinfo/rhq-devel



More information about the rhq-devel mailing list