AnushAmit Malhotra

Hi All,

We are on Weblogic version 10.3.X & are using OJDBC14. Two quick questions

(1) Are there memory leaks in OJDBC14 w.r.t t4cpreparedstatement which are sorted out on OJDBC7? If yes, how can I do a pre & a post to validate the same.
We are using a prepared Statement Cache Size of 200

(2) When we have a piece of code as below

try{

for (loop runs for 100 interations){

ps = conn.prepareStatement(sql);
rs = ps.executeQuery();
}finally{
      ps.close();
      rs.close()
}

and do close ps & rs in a finally block, do the hanging references of ps & rs (ones which are not closed in the for loop) remain with the connection object? What happens when the connection object goes back to the Weblogic Connection Pool?
Does it carry this state or is it cleaned up?

Hi Joe,

We are on Weblogic 10.3.3 and are facing a similar situation now, this does not happen too often but every couple of months, few messages are lost without any trace. Ours is not a high volume application. It is configured with Default work manager with 16 concurrent consumers. Messages are consumed by an MDB which triggers an SLSB. This is for End of Day processing which is carried out in Banks.

Please let me know if you did get a solution to your problem.

Hi All,

Our end of day processing gets triggered by an Autosys job which triggers a POJO, this POJO puts message in a queue which is picked up by an MDB for processing. This is part of an end of day process in which once triggered continues with MDB --> SLSB combination until done for a given Host (Country).

During our End of Day Batch for one Host, there were 52 messages created by the POJO which our application logs show, however only 46 of these were consumed, we are not able to find out from Application or Server logs on what happened to the pending 6 messages. Logs are clean without any error messages.

Messages are persistent & we are using JMS File store, there is no poison queue configured & everything else is on default Weblogic settings.

Can any of you help me with trouble shooting of this issue. Since the messages are persistent, is there a way to retrieve them from File Store? Are there any tools which could help us do this? Google was not much helpful in pointing out anything.

Are there any setting we can do on the Weblogic console which will help us avoid such situations in the future ?

Hi All,

We are working on Weblogic 10.3.3 with EJB 2.1 and our application uses the default work manager with Fair Share Based Request Class . I have a couple of questions that I need help on

(1) Did read on the net that the maximum number of concurrent MDB's spawned is 16, how does Weblogic handle things internally when JMS requests are greater than 16, how are they taken up for processing?
In our application, each JMS request triggers an EOD process which is triggered for a country for EOD Account Interest Calculation.
We have some some strange behavior in our Production Box, on month end processing specifically when the load is high, requests get queued up and it does not follow a FIFO model, requests which come in late are processed earlier than the ones which are waiting to be executed - Why does this happen?

(2) If we configure a workmanager specifically to handle this, what is the thumb rule to get the min and max threads right?

Please let me know.

Thanks Martin.

The same code works correctly in our local environment and has undergone complete system testing. Column is defined as num(18,5)

Application gets data from a JMS message , this gets printed into application log files correctly i.e. 99.99 but at the table level, when we view this using SQL Developer, it shows off as 99,99.

When the application reads this data, it considers it as 9999.00 and the computing logic totally goes for a toss.

Any other pointers which I can check?

Hi All,

The Decimal separator (.) in Amount fields in my Application are getting stored as comma (,) in Oracle 11g. This is for a European client. There was some discussion today on the NLS parameter & the way it is setup for Oracle Database.
As a simple example 99.99 is stored as 99,99 and while reading the value,application reads it as 9999.00.

Is there some way to handle this at the Oracle Driver Level or any other configuration level change as we do not want to make code changes.

Please let me know

Hi All,

The Decimal separator (.) in Amount fields in my Application are getting stored as comma (,) in Oracle 11g. This is for a European client. There was some discussion today on the NLS parameter & the way it is setup for Oracle Database.
As a simple example 99.99 is stored as 99,99 and while reading the value,application reads it as 9999.00.

Is there some way to handle this at the Oracle Driver Level or any other configuration level change as we do not want to make code changes.

Please let me know