Problem with InitialContext Creation in Weblogic 5.1

Question

Hi all,
We are using Weblogic 5.1 on Windows and Solaris.
We are trying to get InitialContext by the following code snippet.
This piece of code is in a Thread which is kicked off by a startup class.
The purpose of getting the initialContext here is to get a reference to a JMS
queue and to start listening for any messages that are put into the queue.
Hashtable htEnv = new Hashtable();
htEnv.put(Context.INITIAL_CONTEXT_FACTORY, "weblogic.jndi.WLInitialContextFactory");
htEnv.put(Context.PROVIDER_URL, "t3://ip_address:port_number");
return new InitialContext(htEnv);
Sometimes this code takes almost 15 minutes to return an InitialContext.
Some other times it does not return at all, neither does it throw any exception.
In the application this code may get called multiple times simultaneously.
The problem is not consistent also.
There is one aspect that has been noticed. A number of InitialContext Objects
are initiated in the application and are not closed after use (As recommended
in the weblogic documentation).
The other observation is that the occurrence of the above explained problem is
very consistent when the executeThreadCount of the Weblogic server is low (<=
15). However, once the executeThread count is increased to a higher value (30
to 45) the occurrence of the problem is very rare.
Any information regarding this problem would be extremely helpful.
Thanks and Regards,
Uday

Answer 1

Thank you Bhagvan,
The creation of initial context in every call is a clear mistake. Caching has
to be implemented.
However , i need some more information about this ....
As you said the initial context is being created at every instant for a lookup
of any object in the jndi tree in the remote machine(I'll name this "machine A").
Simultaneously there is an attempt for creation of an InitialContext in the other
machine(call it "machine B") for the remote "machine A".
This is needed because "machine B" has to get a reference to a JMS queue on "machine
A" and start listening for any messages in it.
It is at this point that the InitialContext creation takes randomly long times
to return.
That is the call is made by populating a Hashtable with the properties of the
remote "machine A" like the IP address, Port number of "machine A" and using this
Hashtable as a parameter in "machine B" to get the InitialContext.
I was also wondering if there is a limit on the number of InitialContext objects
that can be opened at any instant in a server.
Could you please throw more light on these areas ........
Thanks and regards,
uday
"bhagvan" <[email protected]> wrote:
>
Is there any particular reason why you want to get
a new initial context in every call ?? My suggestion
is to cache it based on the application parameters let us
say the client user or client's machine or any other parameter
which is unique for a client.
hope this helps.
"Uday Reddy" <[email protected]> wrote:
Hi all,
We are using Weblogic 5.1 on Windows and Solaris.
We are trying to get InitialContext by the following code snippet.
This piece of code is in a Thread which is kicked off by a startup class.
The purpose of getting the initialContext here is to get a reference
to a JMS
queue and to start listening for any messages that are put into thequeue.
Hashtable htEnv = new Hashtable();
htEnv.put(Context.INITIAL_CONTEXT_FACTORY, "weblogic.jndi.WLInitialContextFactory");
htEnv.put(Context.PROVIDER_URL, "t3://ip_address:port_number");
return new InitialContext(htEnv);
Sometimes this code takes almost 15 minutes to return an InitialContext.
Some other times it does not return at all, neither does it throw any
exception.
In the application this code may get called multiple times simultaneously.
The problem is not consistent also.
There is one aspect that has been noticed. A number of InitialContext
Objects
are initiated in the application and are not closed after use (As recommended
in the weblogic documentation).
The other observation is that the occurrence of the above explainedproblem
is
very consistent when the executeThreadCount of the Weblogic server is
low (<=
15). However, once the executeThread count is increased to a highervalue
(30
to 45) the occurrence of the problem is very rare.
Any information regarding this problem would be extremely helpful.
Thanks and Regards,
Uday