Skip to Main Content

Oracle Database Discussions

Announcement

For appeals, questions and feedback about Oracle Forums, please email oracle-forums-moderators_us@oracle.com. Technical questions should be asked in the appropriate category. Thank you!

ORA-3135 issue with XP to Solaris environment

Stephen KarniotisJun 16 2009 — edited Jun 17 2009
Hi everyone

OK. I need a bit of help with this one as I have not diagnosed SQL*Net issues for quite some time. Here's the full scenario.

1. Environment: Desktop: 11g, 11.1.0.7 client
Desktop: 11g, 11.1.0.7 database
Solaris: 11g, 11.1.0.7 database
Solaris: 11g, 11.1.0.7 database client (same tree)
2. Application:
PL/SQL application that transforms data from one table to another. Process runs cleanly when using desktop to desktop, meaning the PL/SQL procedure invocation from SQL*Plus from the Windows client to the Windows database works without issues. I'm not sure if when I use the client SQL*Plus from another code tree, the process may or may not work.

When I invoke the packaged procedure from the PC client to the Solaris server, I get a 3135 error approximately 45-60 minutes into the process, disconnecting my session. With some modifications to my code, I confirmed that data is inserted into the table and committed. So I tried something different. I executed the procedure from the Sun Solaris database server and 3.5 hours later it completed successfully! So, given that SQL*Net on the same Unix box uses Bequeath vs direct SQL*Net for TCP/IP, I'm assuming that explains why it completed successfully.

3. I increased the SDU_SIZE and the QUEUESIZE for the listener and saw some improved performance but did not solve my issue.

4. I set the SQLNET.EXPIRE_TIME to 0 so that connections would not expire. From the documentation, this seems to be correct.

5. For the Listener, INBOUND_CONNECT_TIMEOUT is set to 60 seconds so this does not appear to have any affect as the connection is made right away.

6. Using a trace set to ADMIN shows a disconnect but does not explain the problem.

I've read a few posts on OTN but nothing seems to point to the issue at hand. I'm wondering if setting EXPIRE_TIME to something like 5 hours would do the trick or am I missing something with the Listener configuration.

We are running (MTS - Oops Shared Servers) so I'm not sure if the problem could be attributed to this. I did see a trace file for one of the dispatcher processes.

Any guideance would be wonderful. Am I completely off the board or very close. I do need to solve this problem as this PL/SQL code is going to be used to migrate production data from one system to another. The size of the data stores in the As Is system are variable so I need something that will work without me handholding the process.

Thanks in advance

Stephen Karniotis
Compuware
Stephen.Karniotis@Compuware.com
(313) 227-4350
Comments
Locked Post
New comments cannot be posted to this locked post.
Post Details
Locked on Jul 15 2009
Added on Jun 16 2009
3 comments
1,087 views