db-derby-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Kathey Marsden (JIRA)" <j...@apache.org>
Subject [jira] Commented: (DERBY-4326) hang on winvista with IBM 1.6 SR5 in NSSecurityMechanismTest
Date Tue, 28 Jul 2009 23:05:14 GMT

    [ https://issues.apache.org/jira/browse/DERBY-4326?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12736355#action_12736355
] 

Kathey Marsden commented on DERBY-4326:
---------------------------------------

The thread dump in the javacore reveals that this exception causes our ClientThread to go
away, but the server stays up blocked in blockingStart() but has no ClientThread to handle
new requests.  I think typically the error Connection reset by peer would happen if the client
closed the socket.  Even though the client does do a quick close:
  DriverManager.getConnection(dbUrl).close();
I don't think the getConnection() should return until we are well through this setKeepAlive
code.  Regardless of why it happened I don't think such an error should cause loss of the
ClientThread. Currently we have:

         clientSocket.setKeepAlive(parent.getKeepAlive());
                        
                        // Set time out: Stops DDMReader.fill() from
                        // waiting indefinitely when timeSlice is set.
                        if (timeSlice > 0)
                            clientSocket.setSoTimeout(timeSlice);
                        
                        //create a new Session for this socket
                        parent.addSession(clientSocket);
                        
                    } catch (PrivilegedActionException e) {
                        // Just throw the underlying exception
                        throw e.getException();
                    } // end inner try/catch block



> hang on winvista with IBM 1.6 SR5 in NSSecurityMechanismTest
> ------------------------------------------------------------
>
>                 Key: DERBY-4326
>                 URL: https://issues.apache.org/jira/browse/DERBY-4326
>             Project: Derby
>          Issue Type: Bug
>          Components: Network Server
>    Affects Versions: 10.5.2.0
>            Reporter: Myrna van Lunteren
>         Attachments: javacore.20090728.095131.5196.0001.txt, javacore.20090728.114939.5640.0001.txt
>
>
> This is stopping my platform testing on winvista with ibm 1.6 SR5.
> During the initial run of suites.All, I got:
> --------------------------------
> 		testNetworkServerSecurityMechanism java.net.SocketException: Connection reset by peer:
setsockopt
> 	at java.net.PlainSocketImpl.socketSetOption(Native Method)
> 	at java.net.PlainSocketImpl.setOption(PlainSocketImpl.java:283)
> 	at java.net.Socket.setKeepAlive(Socket.java:1171)
> 	at org.apache.derby.impl.drda.ClientThread.run(Unknown Source)
> --------------------------------
> Rerunning suites.All gave the same error.
> There is only 1 java process active & hanging.
> Running just the test itself also results in a hang but there was (so far - it's been
sitting unchanged for 4 hours) java.net.SocketException.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


Mime
View raw message