[CXF-7881] HttpConduit.handleResponseOnWorkqueue will always handle response on current thread when allowCurrentThread is false and the work queue rejects the execution

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 3.2.6
• Fix Version/s: 3.2.7, 3.3.0
• Component/s: Transports
• Estimated Complexity: Unknown
• Assignee: Andrew J McMurry
• Reporter: Jan Hallonsten
• Created: 22/Oct/18 15:29
• Updated: 30/Oct/18 10:05
• Resolved: 23/Oct/18 11:00

Description

Creating this Jira according to the discussion here

In the method 

org.apache.cxf.transport.http.HTTPConduit.handleResponseOnWorkqueue(boolean allowCurrentThread, boolean forceWQ)

If the work queue is full so that RejectedExecutionException is thrown and allowCurrentThread is false like when called from AsyncHttpConduit the expression in the if statement below will always return false and the response will be handled on the current thread via the call to handleResponseInternal. When used from AsyncHttpConduit this will be the IO core thread which is not a good idea.

} catch (RejectedExecutionException rex) {
    if (allowCurrentThread
        && policy != null
        && policy.isSetAsyncExecuteTimeoutRejection()
        && policy.isAsyncExecuteTimeoutRejection()) {
        throw rex;
    }
    if (!hasLoggedAsyncWarning) {
        LOG.warning("EXECUTOR_FULL_WARNING");
        hasLoggedAsyncWarning = true;
    }
    LOG.fine("EXECUTOR_FULL");
    handleResponseInternal();
}

I think that the code above should be changed to

} catch (RejectedExecutionException rex) { 
     if (!allowCurrentThread 
         || (policy != null 
         && policy.isSetAsyncExecuteTimeoutRejection() 
         && policy.isAsyncExecuteTimeoutRejection())) { 
         throw rex; 
     } 
     if (!hasLoggedAsyncWarning) { 
         LOG.warning("EXECUTOR_FULL_WARNING"); 
         hasLoggedAsyncWarning = true; 
     } 
     LOG.fine("EXECUTOR_FULL"); 
     handleResponseInternal(); 
}

Issue Links

• links to: GitHub Pull Request #463