REP-300 Error memory problem running reports

[cmcintos]

We are receiving the following errors trying to run any report. Looking at task manager on the report server we see a lot of java and javax processes. Does anyone know what these are and how to overcome the following error?

REP-300: unable to allocate 4096 bytes of shared memory ("java pool","unknown object","joxs heap","Intern")
ORA-06512: at "GLOGOWNER.UTC", line 65

thanks

[chrisplough]

Re: REP-300 Error memory problem running reports

Hello,

This error (unable to allocate 4096 bytes of shared memory) indicates that your report server needs more memory allocated in order to successfully run the specified report. This is common and can be solved by increasing the memory allocated to the Oracle Reports JVM instance.

--Chris

[mkenney]

Re: REP-300 Error memory problem running reports

How do we Increase the Memory for the Reports JVM Instance?

This issue is high priority for us. All reports failing with memory error, and upgrading tomorrow! Any help would be much appreciated.

[cmcintos]

Re: REP-300 Error memory problem running reports

Thanks Chris.

We've allocated more memory to the server but not sure on specifics on allocating memory the Oracle Reports JVM Instance. Opened a ticket with Oracle.

[nick.polites]

Re: REP-300 Error memory problem running reports

You don't need to allocate anything additional to the JVM. How much ram does this server have? Also what other processes are running and how much swap is in use out of how much available?

[mkenney]

Re: REP-300 Error memory problem running reports

The Reports Server has 4GB of RAM available.

In an attempt to correct this issue we have specified an initial and max heap as JVM Options for the Reports Engine in the conf file:

-Xmx128M -Xms=32M

After doing this, we still get the "unable to allocate 4096 bytes of shared memory" error. How could it fail to find 4KB (Killobytes!) of memory?

The metalink people are thinking maybe it requires the Database to be rebooted because the root cause could be the ORA-06512 error, so we're rebooting all of the servers right now.

We are trying to get clairification on Chris P.'s response: "This is common and can be solved by increasing the memory allocated to the Oracle Reports JVM instance." Is the conf file change we did what he is talking about, or is it something else?

In general the system seems to be OK for memory when looking under Task Manager Processes.

[mkenney]

Re: REP-300 Error memory problem running reports

The java_pool_size was set to 0 bytes in the pfile in the database resulting in the errors. The value was changed to 150M, and ALL reports are working now.

Thank you everyone for your help looking into this issue.

[chrisplough]

Re: REP-300 Error memory problem running reports

Good work - I hadn't seen an OTM database with the java_pool_size set to 0 before, so I would have overlooked this.

I should have been a little more clear with this. In my experience, I've seen a few more scenarios that lead to the same and similar errors:

1. Large, complex reports are being run and not enough memory is assigned to each reports thread and needs to be increased. This is done by editing the JVM Options for the Reports Engine.
2. Too many reports are being kicked off at once, causing many running Oracle Reports threads. In order to resolve this, the thread stack size needs to be reduced by adding the -Xss parameter (and associated thread stack size value) to the JVM Options for the Reports Engine.
3. The server doesn't have enough physical memory to support OAS and the memory allocated to each reports thread. Edit the JVM Options for the Reports Engine and decrease the java heap allocation for each thread.

--Chris