Disabled metrics are still collected

Jay Shaughnessy jshaughn at redhat.com
Tue Mar 11 13:23:08 UTC 2014


You're right,  plugins can query for metrics in any fashion that makes 
sense.  But, after some initial looking, this has to do with the 
requested metrics.  It seems they can get wonky after certain changes in 
the schedules and/or enablement.  I'm going to look further today.  I 
don't think it's a horrible issue because, at least for me, an agent 
restart brought the schedules back to a steady state.  So, it seems to 
only affect resources whose schedules get updated while the agent is 
running, and not every time (and I don't yet have it pinned down as to 
what the magic changes need to be to make this happen).


On 3/11/2014 4:36 AM, Michael Burman wrote:
> Are you sure this isn't because of how the agent-plugin has been 
> created? At least the SQL plugins do something like: SELECT metric1, 
> metric2, metric3, metric4 FROM metrics style of queries, which means 
> they'll collect all the data once the getValues is requested, but they 
> only report the values which are requested by the server (those which 
> are enabled). In this example it wouldn't make sense to collect each 
> value separately, since it reduces the load to gather them all at 
> once, but of course this would be dependant on the resource type.
>
>   -  Micke
>
>
>
> On Mon, Mar 10, 2014 at 9:01 PM, Jay Shaughnessy <jshaughn at redhat.com 
> <mailto:jshaughn at redhat.com>> wrote:
>
>
>     Yow - let me take a look at this right now...
>
>
>     On 3/10/2014 11:20 AM, Larry O'Leary wrote:
>
>         While looking into another issue dealing with metric collection
>         failures, I have stumbled across something strange. I can see
>         metrics
>         being collected by the agent for metrics which are not
>         enabled. Is this
>         expected?
>
>         In my case I am working with a non-Tx datasource resource from
>         the AS7
>         plug-in. Even though many of the metrics are disabled by
>         default, I am
>         seeing some disabled metrics included in the measurement
>         collection
>         request:
>
>         PreparedStatementCacheCurrentSize
>         PreparedStatementCacheMissCount
>         PreparedStatementCacheMissCount
>         PreparedStatementCacheAddCount
>         PreparedStatementCacheAccessCount
>         PreparedStatementCacheAccessCount
>         PreparedStatementCacheDeleteCount
>         PreparedStatementCacheDeleteCount
>         PreparedStatementCacheHitCount
>         PreparedStatementCacheHitCount
>         AverageBlockingTime
>         AverageCreationTime
>         MaxCreationTime
>         TimedOut
>         TotalCreationTime
>         TotalCreationTime
>         TotalBlockingTime
>         TotalBlockingTime
>
>
>         If this is happening with other resources and resource types,
>         I can
>         imagine that the load generated on and by the agent is much
>         higher then
>         what would be expected.
>
>         http://ur1.ca/gt6mv contains the log message showing these
>         disabled
>         metrics being included in the getValues call. Something to
>         note is that
>         even though the above metrics are disabled, the PC reports them as
>         enabled as seen in the log message.
>
>         Is this expected? A bug? Something we are already aware of?
>
>
>     _______________________________________________
>     rhq-devel mailing list
>     rhq-devel at lists.fedorahosted.org
>     <mailto:rhq-devel at lists.fedorahosted.org>
>     https://lists.fedorahosted.org/mailman/listinfo/rhq-devel
>
>
>
>
> _______________________________________________
> rhq-devel mailing list
> rhq-devel at lists.fedorahosted.org
> https://lists.fedorahosted.org/mailman/listinfo/rhq-devel

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.fedorahosted.org/pipermail/rhq-devel/attachments/20140311/cd347430/attachment.html>


More information about the rhq-devel mailing list