[jboss-jira] [JBoss JIRA] (WFLY-4697) EJBCLient access to SFSBs does not play well with clean shutdown

[Richard Achmatowicz (JIRA)]

    [ https://issues.jboss.org/browse/WFLY-4697?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13083026#comment-13083026 ] 

Richard Achmatowicz commented on WFLY-4697:
-------------------------------------------

In the case of a managed transaction context. there are two problems when an invocation arrives at a server and comes back to the client with a NoSuchEJBException:
(i) the invocation will be automatically retried by the EJBClient retry mechanism, which should not happen as there is no replicated transaction state on the other nodes in the cluster
(ii) the retry will not take into account the fact that the TransactionInterceptor on the client side has already set up an XAResource on the client side to represent the server receiving the invocation, and that XAResource will ultimately fail when it tries to commit or rollback as it points to the wrong node. In fact, because the NoSuchEJBException was raised, no work was done on the server side, as we did not reach the EJB interceptors.  

We need to disable retry of invocations in a managed context, because of point (i).

Due to the way the EJBClient invocation mechanism works, as well as the way transactions are associated to invocation threads, this is tricky. It is possible to assign a RETRY_DISABLED_KEY attachment to the invocation context when we know that the invocation is in an EJBCLientmanagedTransactionContext, so that when we are processing a response in the NoSuchEJBExceptionResponseHandler, we can avoid doing a retry at the message send/receive level. Additionally, the TransactionInterceptor processing of the result is only called when we call EJBClientInvocationContext.getResult(). It would be possible at this point to somehow clean up the XAResource, perhaps without triggering a rollback?

    


> EJBCLient access to SFSBs does not play well with clean shutdown
> ----------------------------------------------------------------
>
>                 Key: WFLY-4697
>                 URL: https://issues.jboss.org/browse/WFLY-4697
>             Project: WildFly
>          Issue Type: Bug
>          Components: Clustering, EJB, Remoting
>    Affects Versions: 10.0.0.Alpha1
>            Reporter: Richard Achmatowicz
>            Assignee: Richard Achmatowicz
>
> The clean shutdown mechanism allows EJB and web applications to make use of shutdown interceptors to allow the application server to refuse requests when a server is in the process of shutting down. These interceptors are tied to the processing of EJB invocations and web requests.
> In the case of EJBCLient invocations, they arrive at the remoting connector, and undergo some preliminary processing before being sent to the EJB interface in question. When a server is shutting down, EJBCLient invocations can arrive at the RemoteConnector and start processing, even when the EJB interface has been locked down, so to speak.
> I am seeing various types of exceptions arising from this preliminary processing (e.g. NPE on DeploymentRepository lookups) and these get returned to the client as exceptions on the SFSB invocation, before even reaching the SFSB interceptors.
> If the EJBCLient is running in a managed transaction context, these returned exceptions will case the SFSB to be discarded (as the SFSB invocation is considered failed with the exception returned) and the transaction will attempt to rollback. If the rollback processing fails (because the original node in the transaction is down the bean gets removed anyway. The SFSB session state is lost, even if there is another node in the cluster which can support the invocation.
> In short, it looks as though the clean shutdown mechanism needs to be used to also lock down the processing of EJBCLient invocations in some way.
>  



--
This message was sent by Atlassian JIRA
(v6.3.15#6346)