[jira] [Commented] (UIMA-3011) DUCC Service Manager - parameterize pingers

["Jim Challenger (JIRA)" <de...@uima.apache.org>]

    [ https://issues.apache.org/jira/browse/UIMA-3011?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13693201#comment-13693201 ] 

Jim Challenger commented on UIMA-3011:
--------------------------------------

After some thought and experimentation it seems best to generalize the problem so I changed the title of task.  The default UIMA-AS pinger is refactored into the CLI project so it can better serve as a base class for those who want to make a more comprehensive pinger.  A new CLI parameter to ducc_service_submit and ducc_services, "service_ping_arguments" is supported.  Its value is a single opaque string passed to the init() method of every pinger, for interpretation by the individual pinger implementations.  For the immediate future, the default UIMA-AS pinger will accept service_ping_arguments='q_thresh=nn,window=mm' where nn is the max queue depth before the pinger marks the service "not healthy", and window is the size of the window (consecutive pings) used to reason on goodness or badness.
                
> DUCC Service Manager - parameterize pingers
> -------------------------------------------
>
>                 Key: UIMA-3011
>                 URL: https://issues.apache.org/jira/browse/UIMA-3011
>             Project: UIMA
>          Issue Type: Improvement
>          Components: DUCC
>            Reporter: Jim Challenger
>            Assignee: Jim Challenger
>             Fix For: 1.0-Ducc
>
>
> Currently SM sets state of a service to "healthy" if any response is received from the ActiveMq MBean for a queue (as well as if get-meta to the service works), and "unhealthy" otherwise.  
> Better would be if some reasoning is done on the queue statistics from ActiveMQ.  This is somewhat subtle as many things need to be taken into consideration to determine if a service is "healthy".
> For the moment, we will use the single heuristic, "is the queue depth too big" to better draw attention to *potential* (but maybe not actual) problems and defer more complete analysis to a later time.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira