[Resin-interest] Clustering and load-on-startup

2011-11-08 Thread Mathias Lagerwall
Hi
I have a problem with clustering in Resin 4.0.23 that I hope someone
can help me with.

When I bring a cluster node back online after it has been shut down
the cluster recognizes it as up and functioning too soon (this can
be monitored on the caucho-status page). My problem is that I have a
servlet that is really heavy to start and it has load-on-startup
set. As it is set up now requests goes to this new node trying to
access the servlet prior too its startup and they are left waiting. I
would like to have the cluster node off line until all the
load-on-startup servlets has started. Is there an existing setting
that can make this work?

What I am looking for is something like load-balance-grace-time or
cluster-off-line-until-load-on-startup-done :)

Best regards
Mathias Lagerwall


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


[Resin-interest] pb with find() in resin 4.0.13

2011-11-08 Thread Riccardo Cohen
Hello
I'm using many times the find() call in EntityManager :

Status st=m_manager.find(Status.class,identifier);

But with no particular reason, in test or production, the find() call 
stop all processing, there is no exception and any System.out.println() 
juste after is simply not called.

While the server is in this undefined state, the rest of the 
application may work. I tried to add a createQuery call just before the 
find :

Query qr=m_manager.createQuery(select st from Status st);
ListStatus list=(ListStatus)qr.getResultList();

and the same problem happens at getRestulList(). Nothing happens, no 
exception, no further processing.

Is there any people that had this problem ? I could see that only in one 
of my applications, not all. Is this a problem with eclispelink ?

Thanks a lot.
-- 
Riccardo Cohen
Architecte du Logiciel
http://www.architectedulogiciel.fr
BP 77536
37075 Tours Cedex 2
+33 (0)6.09.83.64.49
Membre du réseau http://www.reflexe-conseil-centre.org


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


Re: [Resin-interest] pb with find() in resin 4.0.13

2011-11-08 Thread Alex Rojkov
 Hello
 I'm using many times the find() call in EntityManager :
 
 Status st=m_manager.find(Status.class,identifier);
 
 But with no particular reason, in test or production, the find() call 
 stop all processing, there is no exception and any System.out.println() 
Riccardo,

Can you produce a thread-dump of the server's JVM please and send that to us? 

Thanks,
Alex

 juste after is simply not called.

 
 While the server is in this undefined state, the rest of the 
 application may work. I tried to add a createQuery call just before the 
 find :
 
 Query qr=m_manager.createQuery(select st from Status st);
 ListStatus list=(ListStatus)qr.getResultList();
 
 and the same problem happens at getRestulList(). Nothing happens, no 
 exception, no further processing.
 
 Is there any people that had this problem ? I could see that only in one 
 of my applications, not all. Is this a problem with eclispelink ?
 
 Thanks a lot.
 -- 
 Riccardo Cohen
 Architecte du Logiciel
 http://www.architectedulogiciel.fr
 BP 77536
 37075 Tours Cedex 2
 +33 (0)6.09.83.64.49
 Membre du réseau http://www.reflexe-conseil-centre.org
 
 
 ___
 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] Clustering and load-on-startup

2011-11-08 Thread Alex Rojkov
 Hi
 I have a problem with clustering in Resin 4.0.23 that I hope someone
 can help me with.

Mathias,

Can you try adding bind-ports-after-starttrue/ to server-default?

server-default
  bind-ports-after-starttrue/bind-ports-after-start


Thanks,
Alex

 
 When I bring a cluster node back online after it has been shut down
 the cluster recognizes it as up and functioning too soon (this can
 be monitored on the caucho-status page). My problem is that I have a
 servlet that is really heavy to start and it has load-on-startup
 set. As it is set up now requests goes to this new node trying to
 access the servlet prior too its startup and they are left waiting. I
 would like to have the cluster node off line until all the
 load-on-startup servlets has started. Is there an existing setting
 that can make this work?
 
 What I am looking for is something like load-balance-grace-time or
 cluster-off-line-until-load-on-startup-done :)
 
 Best regards
 Mathias Lagerwall
 
 
 ___
 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