|
cygwin scp file copy counters stop incrementing/decrementing
Hello,
I'm running cygwin 1.7.7 on a win2k3R2 Standard edition server. I have a RHEL 4.7 linux host I'm scp'ing files from to the windows system using a simple cygwin scp command from what appears to be OpenSSH_5.5p1 on the cygwin host.
When I run the simple scp command on the cygwin host, the counters initially display and increment/decrement, albeit what is at a much slower rate than is actually taking place.
After the percent complete reaches a little less than 10%, all the counters from the cygwin console window cease to increment/decrement.
However, when I cd to the directory on the cygwin host where the file is being scp'ed to, the file is continuing to be transferred and a much higher rate than the counters seem to show.
Finally, when the file transfer is complete by directly seeing the byte count as equal between the hosts, the scp counters will suddenly jump to 100% complete.
This is not the case with my linux to linux scp's. The counters increment/decrement as expected and transfers are about what I'd expect over the same network infrastructure.
This appears to be a cygwin ssh/scp implementation issue. I've posited this question to cygwin with no response as yet. The only reference I've been able to find on the 'net refers to buffering causing file transfers to appear to complete later than the transfer really does.
The fact that the transfer appears to actually complete is good. However, I have some timing dependent processes before and after the scp's take place. Rather than using an old fashioned stopwatch and timing the file transfers manually, I was really wanting to use the full scp functionality to do the timing for me.
So the question is, is this normal (dis)behavior for the cygwin port of scp/ssh? I'm not using any options for the scp command. The version of ssh appears to be OpenSSH_5.5p1 called by the cygwin system or whatever cygwin is.
Any thoughts? Anyone have the same issue with this pretty vanilla configuration?
Thanks!
Blaine
|