JBoss 7 - EJB Remote Transaction Timeout

[sbourguiba]

Hi,

I have a remote call that allow to run a process that takes more than 5 minutes, after passing this time the transaction was canceled and the global transaction was rollbacked at commit.

Is there a way to change the default timeout (300 seconds) ?

Note: After consulting the method createOrResumeXidTransaction of the EJBRemoteTransactionPropagatingInterceptor class, I found that the timeout value is always 300 seconds (it seems that this value is hardcoded)

Thanks in advance,

[wdfink]

Looks like the is a TODO within the code, not sure whether it's related.

Is it possible to set the Tx timeout fo both sides with an annotation?

[sbourguiba]

I don't know, this is my question, i want to change the default timeout value.

[wdfink]

I recommenc to not set such high default timeout.

You might run into issues, i.e. if a action which is normally short takes longer all resources are blocked and you run out of DB connections or GC issues.

To change the timeout for a few methods you can use the ejb-jar.xml.

[sbourguiba]

thanks for the reply

But the problem is that every remote ejb call, a transaction is created with 300 seconds timeout and if i define a timeout for some methods in the file jboss-ejb3.xml the problem persist because the ejb remote transaction will be canceled after 300 seconds, so the global transaction will be rollbacked at the commit phase (the two-phase commit strategy)

[jaikiran]

That TODO in code needs to be fixed/implemented. Please file a WFLY JIRA with the details and please point to that code in the description of that JIRA

[sbourguiba]

Ok, thanks

[mp911de]

You know about the config (standalone.xml/domain.xml)?

<subsystem xmlns="urn:jboss:domain:transactions:1.3">

            <core-environment>

                <process-id>

                    <uuid/>

                </process-id>

            </core-environment>

            <recovery-environment socket-binding="txn-recovery-environment" status-socket-binding="txn-status-manager"/>

            <coordinator-environment default-timeout="300"/>

        </subsystem>

[wdfink]

Marc, it is more that a server-to-server invocation will not respect the timeout, looks like there is a hardcoded 300ms default in this case.