Allowed number of sessions (CIFS)?

[back0rifice]

Hello,

a colleague of mine encountered the following problem: After uploading a big number of files he tried to access Alfresco via CIFS. But Windows (XP) said that his client has exceeded the maximum number of allowed connections and so he wasn't able to connect to Alfresco for some time. Remapping the network drive didn't solve the problem.

Where does this limiation come from and why did he open that much connections? He was using a FTP client that transfers files in sequence and not parallel.

Thanks
Robert

[gary_spencer]

Hi,

What system is Alfresco running on ?, have you got any other drives mapped to that system ?. Can you browse to the Alfresco CIFS server using Start->Run and enter '\\<servername>' ?.

Cheers

Gary

[back0rifice]

Hi,

Sorry forgot to mention, it's CentOS 4.4 Linux that Alfresco is running on.

Can you browse to the Alfresco CIFS server using Start->Run and enter '\\<servername>' ?.

I'm afraid I can't reproduce the scenario atm but iirc this didn't work either.

Cheers
Robert

[gary_spencer]

Hi,

Is the CentOS system also running Samba ?.

If you enable the following in the log4j.properties can you see any connection attempts to the Alfresco CIFS server in the alfresco.log :-

log4j.logger.org.alfresco.smb.protocol=debug
log4j.logger.org.alfresco.smb.protocol.auth=debug

Cheers

Gary

[back0rifice]

Hi,

Is the CentOS system also running Samba ?.

No.

If you enable the following in the log4j.properties can you see any connection attempts to the Alfresco CIFS server in the alfresco.log :-

log4j.logger.org.alfresco.smb.protocol=debug
log4j.logger.org.alfresco.smb.protocol.auth=debug

Ok, thanks. Is it possible to make it logging to a seperate file?

Thanks
Robert

[sylvaintv]

Hello,

I'm getting the same problem with Alfresco 1.3 server on Linux RHEL4, it happens after a lot of browsing (usually moving quickly through 20-25 folders ) in the Virtual drive (mapped drive) or directly via \\ip\alfresco on several Windows XP Pro SP1/SP2 machines.  The domain is an Active Directory domain on Windows 2003 Server.

The exact error is  the windows error 71.
" No more connections can be made to this remote computer at this time because there are already as many connections as the computer can accept."
Maybe a clue here :
http://support.microsoft.com/kb/314882/en?FR=1&PA=1&SD=HSCH

Nothing is wrong in the log with SMB/CIFS on the alfresco server.

The problem is probably on the Windows client side because other people (on other machines) can still connect to Alfresco without any problem when the error happens. I thought maybe that the Afresco SMB/CIFS implementation somehow "forgot" to release the SMB connections, or is blocking them. Is this something possible ?

It's a quite annoying problem, since it happens really randomly but when it happens, the ONLY way to fix it is to disable/reenable the Ethernet Network connection, or to unplug/plug the network cable (hence the idea of the connection "leak").  This prevents us to deploy Alfresco to all our users.

Thanks.

[gary_spencer]

Hi,

This sounds a bit odd, a Windows client would only use a single socket connection to the server then run multiple virtual circuits over that single connection.

You could check what connections your client has open when the problem occurs using 'nbtstat -S' and 'netstat -a'. you can also check connections using 'net use', and try disconnecting connections to the Alfresco share using 'net use <path> /d'. See if that helps.

Cheers

Gary

[tyve]

Hi,

This sounds a bit odd, a Windows client would only use a single socket connection to the server then run multiple virtual circuits over that single connection.

You could check what connections your client has open when the problem occurs using 'nbtstat -S' and 'netstat -a'. you can also check connections using 'net use', and try disconnecting connections to the Alfresco share using 'net use <path> /d'. See if that helps.

Cheers

Gary

Isn't this the same issue as in http://forums.alfresco.com/viewtopic.php?t=3767 ?
When I sent you some network traffic you said it was fixed in 2.0.1

Regards,
Johannes

[gary_spencer]

Hi,

That fix was for a problem with search handles leaking and eventually running out of handles. This looks like it's complaining about socket connections.

Cheers

Gary