Feature request: forced copy on pscp like native linux cp about parallel-ssh HOT 3 CLOSED

pscp just calls scp behind the scenes, and it looks like scp will autoclobber files by default (on my CentOS 6 machine, anyways, there is no -f option to scp). If your machines have an alternate implementation of scp, let me know, and I'll reopen.

from parallel-ssh.

Thanks for looking at this.
Let me give you a realworld example where cp -f works but apparent pscp / scp does not.

It is possible to replace an nginx binary on a live running server and then just send a USR2 signal for it to gracefully reload itself without connections lost, ie.

cp -f objs/nginx /usr/local/sbin/nginx && kill -USR2 cat /var/run/nginx.pid

It's not possible to do this via pscp
Instead I get [FAILURE] Exited with error code 1 unless I stop nginx first.

Since scp doesn't have -f natively, perhaps it is possible to emulate it by forcefully removing the destination first before copying it.

I guess I can do that by first running a pssh command to forcefully remove the binary via rm -rf and then do the copy but it would be nice if the -f existed by emulation so it could be done almost in place as quickly as possible.

from man cp
-f, --force if an existing destination file cannot be opened, remove it and try again

from parallel-ssh.

Does a straight scp do the right thing for you? If not, that's what you need to troubleshoot.

If that does work, then I'm going to need the exact pscp command you're running (feel free to sanitize hostnames/IPs) and some information about what Linux distro you're using and which version of OpenSSH.

from parallel-ssh.