Re: [Resin-interest] Re sin 3.1.9, Scheduled Tasks Broken after upgrade

2009-11-02 Thread Alex
 I stepped through CronType in the debugger and 3.1.9 is setting the  
 next
 start time for the scheduled task a week out.  Here's the config we  
 have:

scheduled-task type=mm.services.DownloadService
 name=DownloadService
cron3,7,15,20,25,30,35,40,45,50,54,58 * * * */cron
/scheduled-task

 Running this today (Nov 1), the next start time is set to Nov 8.  I

Thanks, I've reported a bug: http://bugs.caucho.com/view.php?id=3736

--Alex

 set
 resin's debug level to finer, and all scheduled tasks now start a  
 week late.

 Although I have * in the day of week block, CronType is advancing  
 the next
 time to current day + 7.  There was one change made to CronType  
 between
 3.1.8 and 3.1.9, which is highlighted below.  3.1.8 does not push  
 the next
 start time out a week.

 if (_daysOfWeek != null) {
   int oldDayOfWeek = cal.getDayOfWeek() - 1;
   int dayOfWeek = nextInterval(_daysOfWeek, oldDayOfWeek);

 if (dayOfWeek  0) {
 day += (dayOfWeek - oldDayOfWeek);
   }
   else {
 dayOfWeek = nextInterval(_daysOfWeek, 0);
   
 day += (dayOfWeek - oldDayOfWeek + 7);
   }



 tweihs wrote:

 Things we've tried:
 * changing our tasks to implement Runnable vs Work (a swag, as
 com.caucho.resources.ScheduledTask operates on Runnables, which  
 work is
 * Put breakpoints in com.caucho.resources.ScheduledTask methods  
 including
 the init, start, and handleAlarm methods. The debugger breaks on  
 init and
 start, but handleAlarm is never called.

 Is something killing the alarm service?  Did something change in the
 schedule-task tags that require changing syntax in 3.1.9?

 The 3.1.8 mem leak issues are a problem for us in production, thus  
 the
 desire to move to 3.1.9.

 Thanks,

 -tyson


 tweihs wrote:

 Bumping this, still seeing the issue with 3.1.9.  We tried to move
 forward to 3.1.9 again today, but all ScheduledTasks would not run.
 Reverted to 3.1.8 and AOK.

 -tyson


 tweihs wrote:

 We upgraded to resin 3.1.9 yesterday in our production  
 environment.  The
 upgrade process for us was simple: unzip the 3.1.9 bundle, copy  
 over the
 resin.conf, license, other supporting files and restart the server.
 However, 3.1.9 will not run any of our scheduled tasks.  The app  
 comes
 up under 3.1.9, responds to requests, but won't run tasks (I  
 think this
 carries over to JMS as well, but need to confirm).  Bring the  
 server
 down and start up in 3.1.8 and schedule tasks resume executing.

 I recall when upgrading to 3.1.8 there may have been an issue  
 with some
 of the Resin generated wrappers/proxies on registered beans that
 prevented tasks from running, and clearing something (WEB-INF/work,
 IIRC) ended up unblocking whatever kept ScheduledTasks from  
 running.
 We've tried doing clean builds and deploying under clean installs  
 to
 3.1.9 and can't get those tasks to run.

 We experience the exact same behavior in development, stage, and
 production.

 Any ideas?

 -tyson






 -- 
 View this message in context: 
 http://old.nabble.com/Resin-3.1.9%2C-Scheduled-Tasks-Broken-after-upgrade-tp23584210p26152486.html
 Sent from the Resin mailing list archive at Nabble.com.



 ___
 resin-interest mailing list
 resin-interest@caucho.com
 http://maillist.caucho.com/mailman/listinfo/resin-interest



___
resin-interest mailing list
resin-interest@caucho.com
http://maillist.caucho.com/mailman/listinfo/resin-interest


Re: [Resin-interest] Adding node to a cluster on the fly

2009-11-02 Thread Alex

 I just created my first resin based ec2 server on amazon.  I want to  
 use amazons cloud computing to grow and shrink my cluster on the  
 fly.  Is there a way to achieve this in resin.

Nathan,

In 4.0 there is a new -dynamic-server capability that allows servers  
be added to the cluster dynamically. The argument parameter value  
takes cluster to join, address and port for the new server. (more info  
available with resin.sh -help.

You may find these two caucho blog entries of use: 
http://blog.caucho.com/?p=177 
  , http://blog.caucho.com/?p=220

Thanks,
Alex

 Nathan
 ___
 resin-interest mailing list
 resin-interest@caucho.com
 http://maillist.caucho.com/mailman/listinfo/resin-interest



___
resin-interest mailing list
resin-interest@caucho.com
http://maillist.caucho.com/mailman/listinfo/resin-interest


[Resin-interest] Controlling pool size for stateless EJBs

2009-11-02 Thread Jeff Schnitzer
Is it possible to control the pool size for stateless EJBs?

I think I'm looking for something equivalent to ejb-message-bean
message-consumer-max=2 but for ejb-session-bean.  The docs say
that session beans are pooled, but I don't see anything about how to
configure this pool.

The use case is that I have a rather expensive resource (held-open
connections to Apple's crappy iPhone Push Notification Service) that
requires pooled access.

Thanks,
Jeff


___
resin-interest mailing list
resin-interest@caucho.com
http://maillist.caucho.com/mailman/listinfo/resin-interest