Hi everyone.<br>
<br>
I have a question to session replication performance and like to clarify some things.<br>
<br>
As to your question, Hans, about the DAS' role in session replication: It does not participate<br>
in replication as the DAS has the SPECTATOR role in the shoal process. If you shut<br>
down the DAS the cluster instances keep running and communicating the replication<br>
data.<br>
<br>
For documentation of the replication process look for the [i]shoal[/i] project.<br>
<br>
The node agent is responsible for starting the server-instance on the particular machine.<br>
Moreover it is used for communication with the DAS. The DAS for instance uses the<br>
node agent to emerge the instance start/ stop commands and controls the<br>
syncronisation of the domain data (applications and libraries and so on) that reside in<br>
the domain directory of the DAS.<br>
<br>
As for my question:<br>
I experience the same behaviour with respect to that [i]federated query[/i] thing.<br>
I use the mod_jk module on a apache server as load balancer to glassfish V2 beta3 (b50g).<br>
Everything works fine so far. The time the load balancer decides to switch to another cluster<br>
instance host i see the following log entries (switched form instance 1 to instance 2):<br>
<br>
[b]instance1[/b]:<br>
[pre]
[#|2007-08-08T08:51:31.499+0000|INFO|sun-appserver9.1|javax.enterprise.system.stream.out|_ThreadID=28;_ThreadName=pool-5-thread-1;|
beginning processBroadcastfindsession: cache dump|#]
[/pre]
<br>
[b]Instance2[/b]:
[pre]
[#|2007-08-08T08:51:31.494+0000|INFO|sun-appserver9.1|javax.enterprise.system.stream.out|_ThreadID=39;_ThreadName=TP-Processor6;|
nextSessionId = 4fc567cefe8d4017a4c6889ebebb|#]
[#|2007-08-08T08:51:31.494+0000|INFO|sun-appserver9.1|javax.enterprise.system.stream.out|_ThreadID=39;_ThreadName=TP-Processor6;|
expected respondants = 2|#]
[#|2007-08-08T08:51:34.498+0000|INFO|sun-appserver9.1|javax.enterprise.system.stream.out|_ThreadID=39;_ThreadName=TP-Processor6;|
federated query took 3003 millis|#]
[#|2007-08-08T08:51:34.500+0000|INFO|sun-appserver9.1|javax.enterprise.system.stream.out|_ThreadID=39;_ThreadName=TP-Processor6;|
SessionID: 4fc567cefe8d4017a4c6889ebebb|#]
[/pre]
<br>
Please note the time gap of 3003 milliseconds that instance2 needs to 'query' the session<br>
data.This is way too much for a server response that does nothing but to display the session id!<br>What am I doing wrong? Please help!<br>
<br>
Best regards,<br><br>
Alexander.
[Message sent by forum member 'sahlix' (sahlix)]
http://forums.java.net/jive/thread.jspa?messageID=229978