Starting RMI immediately causes Oracle dump with RmiServer.log showing a "No more data to read from socket" error.

Products

• CA Automic Applications Manager (AM)

Issue/Introduction

Error Message :
AwE-5001 Database Query Error
java.sql.SQLRecoverableException: No more data to read from socket

There is a very rare problem where the RMI process fails to start and simultaneously disconnects from the database due to an Oracle core dump. The above error is seen in the RmiServer.log.

Oracle alert log shows an error similar to:

Dumping diagnostic data in directory=[cdmp_20161014223952], requested by (instance=16, osid=25695), summary=[incident=307629].

Oracle trace log shows an error similar to:

REQUEST:trace dump in directory=[cdmp_20161014223952], requested by (instance=16, osid=25695), summary=[incident=307629, abnormal process termination].
Performing trace dump

Investigation

This has been observed only with the following configuration:

• Applications Manager version 8.x or 9.x
• Oracle 11.2.0.4
• Use custom parameter " _optimizer_unnest_all_subqueries" for History Purge job as documented in Know Error Article "After upgrading to Oracle 11.2.0.4, HISTORY_PURGE Job aborts with ORA-03113 error"

Cause

Cause type:
Configuration
Root Cause: This issue is related to a change made in Oracle version 11.2.0.4.

Environment

OS Version: N/A

Resolution

Please contact Oracle for additional information regarding this parameter.

Fix Status: No Fix

Fix Version(s):
N/A

Additional Information

Workaround :
Add the parameter "alter session set "_optimizer_unnest_all_subqueries"=false;" to the AW_MASTER_STMTS table and restart RMI process.