Hello Stian,
Thank you for the detailed response. I'm checking into them. One 
question. I do not know how to set/check if the job (workflow?) is 
specifying an UpdateInterval. I re-read the TV 1.7 manual and poked 
about the GUI but can find nothing of the sort. I looked into a workflow 
.xml file and search around but also find nothing obviously tied to this.

Sorry to ask such a simple question, but how do I check such a thing?
--jeff




Stian Soiland-Reyes wrote:
> On Mon, Mar 23, 2009 at 14:50, Jeffrey L. Tilson <[email protected]> wrote:
> 
>> One thing I've noticed was a seemingly high rate of access to the sql
>> database by the RES. For example, the sql log would
>> grow to 13 GB in size in just 1-2 days. We needed to turn off the
>> logging to prevent filling up the filesystem where logs are stored. Most
>> of the sql commands captured looked to be simply polling for state. The
>> rate seems to be ~1 per sec. (a very crude measurement).
> 
> 
> My first rhought was that it's the update interval specified in the
> job. If this is true, you should only see this while jobs are running.
> 
> 
> You can either change it in the job document initially sent, or change
> it in a separate POST later. (The idea is that the client can say
> "Quick updates please" - and after the old value has passed than the
> worker will report more often. The client can then reduce the rate
> again)
> 
> Perhaps you want to force a server-wide minimum level by modifying the
> code for 
> taverna-rest/src/main/java/net/sf/taverna/service/rest/resources/JobResource.java
> - look for getUpdateInterval()
> 
> 
> taverna-datastore/src/main/java/net/sf/taverna/service/datastore/bean/Job.java
> specifies a default update interval of 1 minute though:
> 
>         // 1 minutes in xsd:duration format.
>         public static final String DEFAULT_UPDATE_INTERVAL = "PT60S";
> 
> .. so unless your client is sending something else then it should be
> 60s, not 1s.. Not sure what the Taverna plugin does, but looking at
> taverna-rest-client/src/main/java/net/sf/taverna/service/rest/client/JobsREST.java
> there doesn't seem to be any update interval specified, meaning the
> default should be used.
> 
> 
> Perhaps what is causing this is rather the checking for new jobs. See
> taverna-engine/src/main/java/net/sf/taverna/service/queue/DefaultQueueMonitor.java
> 
>         private final long CHECK_PERIOD = 300; // checks every 0.3 seconds.
>         Thread.sleep(CHECK_PERIOD);
> 
> This could do quite a lot of SELECTs I guess- but 3 SELECTs a second
> would hardly kill a mySQL database..
> 
> 
> However, if you reduce this you would have to live with it taking
> longer for workflow execution to start.
> 
> 

-- 
******************************************
Jeffrey L. Tilson, Ph.D.
Senior Research Scientist
Renaissance Computing Institute (RENCI)
Campus Box 3455
University of North Carolina
Chapel Hill, NC 27599-3455

919.445.9657 (o)
919.962.1380 (f)
*****************************************

------------------------------------------------------------------------------
Apps built with the Adobe(R) Flex(R) framework and Flex Builder(TM) are
powering Web 2.0 with engaging, cross-platform capabilities. Quickly and
easily build your RIAs with Flex Builder, the Eclipse(TM)based development
software that enables intelligent coding and step-through debugging.
Download the free 60 day trial. http://p.sf.net/sfu/www-adobe-com
_______________________________________________
taverna-hackers mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/taverna-hackers
Developers Guide: http://www.mygrid.org.uk/usermanual1.7/dev_guide.html
FAQ: http://www.mygrid.org.uk/wiki/Mygrid/TavernaFaq

Reply via email to