Redhat 9 scp problem

[ianicr]

I've got a redhat 9 server. Every few days when I try to scp a 50 meg file accross the network it gets to 99% then stalls. It won't copy across until I execute
/etc/init.d/sshd restart

Anyone seen this before or know how to fix it?

 

[clifflang]

When you do the restart, does the transfer complete successfully, or is it terminated?

 

[thedaver]

How long is the stall? Your remote end mind be an EXT3 system with a slow drive that is busy calculating and writing the journal for the filesystem.

I have all sorts of examples of untaring an archive and other ops that pause or slow down because the receiving end is bogged down with EXT3 journalling overhead.

Surfinbox.com Business Internet Services - National Dialup, DSL, T-1 and more.

http://www.surfinbox.com/business/

 

[ianicr]

When I restart the file copies accross fine. Sometimes have to restart 2-3 times though... Yes it is an EXT3 system, but it still seems weird to me that the file copies accross after the sshd is restarted.

 

[thedaver]

I'm telling you that I think the restart is unrelated. A large file will take CPU overhead to SCP (encrypt) and CPU to to do EXT3.

I asked "how long is the stall"??? Try allowing it to go 10 minutes to see if it ever terminates normally.

Surfinbox.com Business Internet Services - National Dialup, DSL, T-1 and more.

http://www.surfinbox.com/business/

 

[miblo]

When you restart the sshd server you just restart the listening service. Each scp session spawns a new sshd process. The 'stalled' process is unaffected by the restart.

Do a 'ps ax' or 'pstree -hupla' to list the running processes. 'top' will also show if you have a cpu-hog, or I/O-wait issue. 'lsof -i' will show which processes are listening, and which have active network connections.