On Tue, 5 Oct 2004, Susan McConnell wrote:
We're copying a number of files, sometimes up to 3-400, using
scp called from a perl script. The command issued by the perl script is:
my $cmd = "scp $basedir/data/$basename remote.example.com:/tmp";
system ($cmd);
We're finding that some ssh processes, with a PPID of 1, remain after the
script has completed. "ps -ef" shows:
UID PID PPID C STIME TTY TIME CMD
root 12904 1 0 09:35 pts/0 00:00:00 /usr/bin/ssh -x ...
root 12990 1 0 09:36 pts/0 00:00:00 /usr/bin/ssh -x ...
root 12995 1 0 09:36 pts/0 00:00:00 /usr/bin/ssh -x ...
root 13000 1 0 09:36 pts/0 00:00:00 /usr/bin/ssh -x ...
root 13251 1 1 09:37 pts/0 00:00:00 /usr/bin/ssh -x ...
The full command shown in each case (removed from above to prevent line
wrapping) is:
/usr/bin/ssh -x -oFallBackToRsh no remote.example.com scp -t /tmp
I've tried "shopt -s huponexit" but that hasn't resolved this. Can anyone
shed some light on why this is happening and how we can fix it?
Thanks for any help or ideas.
Sue
You didn't say what kind of a system you saw this on. Was it Solaris?
I saw these exact same symptoms on a solaris-9 box running sun's
implementation of ssh (SUNWssh*). Since I had set MAXUPRC on that
system, the leftover processes were causing work to come to a halt.
But, since the system was copying files to non-sun systems, I figured
I wouldn't be able to get support from sun on this issue, and I wasn't
able to firmly determine the cause from sunsolve and google searches.
I found the problem went away when I removed the sun ssh implementation,
and installed openssh-3.9p1 to replace it.
