[TriLUG] SCP acting funny

[Aaron S. Joyner]

Unfortunately, I think you may have gone a step too far in abstracting 
the data in your post.  Three questions:
1) what are the filenames you're attempting this with?  Have you tried 
it with something like "test.txt" as opposed to perhaps something like 
"A very korn-fusing f!le,name"?  It would help to be able to rule out 
shell expansion from odd characters in the filename as the culprit.
2) What files exist, if any, in your .ssh directory?  Have you attempted 
to setup public-key authentication between these hosts?  Do you have a 
authorized_keys or authorized_keys2 file?
3) Is there anything particularly strange in your login scripts?  The 
fact that you can ssh to the remote host successfully and don't mention 
any other wierd login-related problems leads me to believe this isn't an 
issue, but it doesn't hurt to ask.

Tip: create some test files with a few lines of bogus data each, call 
them "test1.txt" and "test2.txt".  Attempt to scp them from machine to 
machine.  Post the precise output, obfuscating only the usernames and 
hostnames if you must.  Do an "ls -la" before and after each attempted 
scp.  Another thing to try would be remote single-command execution - 
i.e. to do an ls of the remote host, do it in your session as "ssh 
hismachine ls -la <filename>".  This has the added benefit of showing 
you if your running into issues w/ shell expansion or space confusion 
from funny file names.

Aaron S. Joyner

Brian A. Henning wrote:

>Hello all,
>  I've recently encountered this problem in a few places, and was wondering
>if anyone had some insight or could tell me where to look for the solution.
>
>scp doesn't like me.  Here are a couple example scenarios involving my
>personal RH7.3 box and another host.  For simplicity, we'll call mine "mine"
>and the other "his."
>
>Example 1:  his runs RH8.0, or possibly FC1 (I helped him upgrade, but I
>forget whether this scenario occurred before or after the upgrade).  I am
>physically at his.
>
>his% scp me at mine:~/some_file_that_exists .
>Password: ********
>/etc/bashrc processed   # <-- that is an echo in my /etc/bashrc, put in to
>help troubleshoot
>                              some odd environment issues
>his% _
>(nothing happens--the file is not transferred...so I try:)
>his% ssh me at mine
>me at mine's password: ********
>/etc/profile processed  # <-- also have this for the same reason as
>mentioned above
>/etc/bashrc processed   # <-- as mentioned above
>A fortune (in my ~/.bashrc)
>the uptime (likewise)
>me at mine% scp some_file_that_exists me at his:~
>me at his's password: ********
>me at mine% _
>(still nothing happens; the file is not transferred..  so I give up on scp
>and put the file in my public_html directory and get it that way)
>
>Example 2: his runs FC1.  I am physically at mine, ssh'd into his.
>
>me at his% scp me at mine:~/some_file_that_exists .
>me at mine's password: ********
>/etc/bashrc processed
>me at his% _
>(nothing--no transfer occurs, so I try on mine:)
>mine% scp some_file_that_exists me at his:~
>me at his's password: ********
>some_file_that_exists  100% |*****************************| 38980
>00:00
>mine% _
>(and it works, but only when initiated from mine, as shown)
>
>What causes this?  I tried an scp -v, and there's a ton of output, very
>little of which indicates the problem that is actually occurring.  In fact,
>comparing that to the same -v output of a successful transfer (the second
>half of Example 2) only confuses me further, as it seems that the debug
>output is the same..
>
>Could it be the output from /etc/bashrc is somehow confusing the process?
>
>Thanks in advance.
>
>~Brian
>
>  
>