Community
Participate
Working Groups
In I20030506, when synch'ing, I often get an error dialog with the message "Unable to synchronize: Cannot close connection". When I revert to I20030429 I do not get the error. However, it does not happen every time in I20030506. Not sure how to reproduce consistently.
I've seen other instances of this as well. Must be related to the fix for bug 37204.
The cause of this has been identified. I will be releasing a fix soon and will request it be included in the next rebuild (I assume there will be a rebuild given that there are test case failures).
there was a rebuild (I20030507) but I still see the same problem.
Can you tell what the side-effects of this error are? Do I have to switch back to I20030429?
Can you give a description of when this happens for you? Also, what conection method are you using? Since the error occurs when the connection is being closed, the only side effect is that anything that happens after tht point will not occur (no sync view or no error dialog if the operation failed).
Happened when using "Synchronize CVS Projects". I did not see any strange thing going on in the UI. Repeating it did not write to the log anymore. I use extssh to connect to dev.eclipse.org
I have updated the Connection.close() method to log any IOExceptions instead of throwing them. The rationale is that the close is issued either because the operation succeeded succesfully or was cut short by another exception. In the first case, the operation can continue and the close failure will be in the log to indicate that a problem did occur. Hopefully there will be enough in the traceback to narrow down the problem. In the second case, the original exception, which is now currently lost, will be shown to the user and the close exception will be logged. Leaving open to consider for 2.1.1
Not to be included in 2.1.1.
Not sure whether I understood: what does the close failure tell me if I see it in the log i.e. what would my (re)action be? Will there be no data loss?
Yes, if you see a close failure in the log and no other error indication, then you can assume that there was no data loss.
Daniel, does this happen often for you on I20030507?
It never happened until today using I20030507 and it only happened in the very beginning i.e. uponn my first synchronization. I supose I got one log entry per repository connection.
*** Bug 37620 has been marked as a duplicate of this bug. ***
*** Bug 37615 has been marked as a duplicate of this bug. ***
Using I20030520 I still get the .log below. This never happened with older 2.1 builds. I started my dev workspace, clicked "Synchronize CVS Projects" button, compared some of the files and then did a "Update from Repository". Do I have to live with these messages now and ignore them? Anything I can do to get rid of them or should this PR be reopened? !SESSION Mai 21, 2003 16:37:42.856 --------------------------------------------- java.version=1.4.1_02 java.vendor=Sun Microsystems Inc. BootLoader constants: OS=win32, ARCH=x86, WS=win32, NL=de_CH Command-line arguments: -update -showlocation -data c:\eclipse\workspaces\BuildMeister_2_2\plugins -install file:C:/eclipse/drops/I20030520/ !ENTRY org.eclipse.team.cvs.core 4 -6 Mai 21, 2003 16:37:42.856 !MESSAGE Cannot close connection !ENTRY org.eclipse.team.cvs.core 4 -6 Mai 21, 2003 16:37:45.871 !MESSAGE Cannot close connection !ENTRY org.eclipse.team.cvs.core 4 -6 Mai 21, 2003 16:37:54.543 !MESSAGE Cannot close connection !ENTRY org.eclipse.team.cvs.core 4 -6 Mai 21, 2003 16:38:06.410 !MESSAGE Cannot close connection !ENTRY org.eclipse.team.cvs.core 4 -6 Mai 21, 2003 16:38:29.103 !MESSAGE Cannot close connection !ENTRY org.eclipse.team.cvs.core 4 -6 Mai 21, 2003 16:39:03.212 !MESSAGE Cannot close connection
No, these messages should not be occuring when operations succeed.
I believe this problem is caused by how we are are closing the socket streams. In some cases, Java throws an exception because the socket is already closed. Since we are requesting a close, I don't think we care that an exception was thrown. Therefore, I have modified the code to ignore the exception unless debugging mode is on.