In A Graceful Shutdown, the SOA Server Is Not Stopped Until JTA Timeout is Met
(Doc ID 2513600.1)
Last updated on OCTOBER 07, 2021
Applies to:Oracle WebLogic Server - Version 10.3.6 and later
Information in this document applies to any platform.
In a graceful shutdown, a SOA managed server takes around 20 minutes to be stopped, same value as JTA timeout. Apparently, there would be some pending transactions.
Other managed servers in the domain are not affected by this issue.
The issue does not occur if the shutdown is forced.
The issue happens even after applying PSU 10.3.6.0.181016 or PSU 10.3.6.0.190115.
In order to determine the root cause, a set of thread dumps were collected before and during the shutdown process, showing that 10 minutes after the shutdown process was started the SOA tip adapter was still processing transactions:
The JTA/transaction service is subject to an order of subsystem shutdowns, and cannot block new transactions during a graceful shutdown because there are other subsystems that require the creation and completion of transactions during their shutdown process.
Via thread dumps, it was determined that the issue would be related with dequeue operations triggered when the server is stopped. These operations will continue till timing out after 20m, same value as JTA timeout.
If the JTA timeout value is decreased to 10m (for example), graceful shutdown will take 10m too, but the idea is not adjusting the timeout but finding out why these transactions are taking that long.
To view full details, sign in with your My Oracle Support account.
Don't have a My Oracle Support account? Click to get started!
In this Document