Remote local transfers time out.

[Shock Diamond]

I have recently replaced Slackware Linux 10.1 with 13.1 and I am experiencing a local remote networking issue over FTP. When I am transfering data over a local route, whether it is to or from the server, the connection stops - all data transfer drops to 0.0k/sec, and even SSH locks out remote access when this occurs. I can still use the local machine in all its functions, but I cannot reconnect via FTP or SSH until I reboot the server.

 

I have tried to find a reason why this is occuring and I cannot. Transfer speeds are relatively healthy - 50-65 megabytes per second writing to the server and 60-90 megabytes per second reading data from it. If I throttle the transfer speed, it still times out regardless.

 

My local backup dumps over FTP to the server without locking up. I can retrieve from the FTP through the backup software as well without an issue. Yet still, direct FTP transfer using FlashFXP to ProFTP drops and locks out remote connections.

 

The only error I can find is in the ProFTPD log:

Jul 09 18:10:32 raven proftpd[1911] localhost (192.168.x.x[192.168.x.x]): USER <hiddenuser>: Login successful.

Jul 09 18:29:00 raven proftpd[1911] localhost (192.168.x.x[192.168.x.x]): sendfile transmission aborted

Jul 09 18:29:31 raven proftpd[1911] localhost (192.168.x.x[192.168.x.x]): notice: user <hiddenuser>: aborting transfer: Unknown error 4294967295

Jul 09 18:31:29 raven proftpd[1911] localhost (192.168.x.x[192.168.x.x]): Client session idle timeout, disconnected

Jul 09 18:31:29 raven proftpd[1911] localhost (192.168.x.x[192.168.x.x]): FTP session closed.

 

This error appears in the logs prior to locking out remote access, but I have yet to find out what the flux is causing it!

 

Any help would be greatly appreciated.

 

-Shock Diamond

[Dungeon-Dave]

I'm not certain what you mean by "local remote transfers" - if you could draw a small diagram showing machine names and what is/isn't working between then, that could clarify your issues.

 

However, other things to try are:

- what happens if you try an SCP or SFTP? (transfer over SSH) - do you get the same issue?

- do you have any firewall rules that throttle traffic and they start to shape the traffic?

- check other logfiles (dmesg, /var/log/messages) - could be some TCP setting that's causing a module (driver) to lock up.

[Shock Diamond]

Only one of my computers is used in coordination with the Slackware machine. I do not use SCP and SFTP - or even have before. I checked /var/log/messages as well as /var/log/secure to ensure there were no other logs or iptables logs that indicate an issue.

 

"Local remote" transfer is confusing and not exactly accurate as it's pretty much an oxymoron, sorry. "Local" as in Local Area Connection/LAN, from another machine on the network over FTP. It's used the same as you would connect from a remote machine to a server, but over LAN because it's gigabit and so much faster.

 

Below, the red line denotes the local (LAN) path where the issue persists. On the other hand, if I hop over the WAN (blue line) and back in to the Slackware machine (green line), it never times out. There has not been an issue.

 



 

I should add, my iptables firewall rules are written to either block a port or open a port, with only one exception. identd is using 'recent' so it's closed unless a local client is using ident. Then it opens long enough for the identd handshaking and closes again. I am not using ident with my FTP on either end.