remsh oddity

[uniq1]

Hi,

We're having an odd issue using 'remsh host -l user -n "ls -l"` type commands (with or without the -n) yet rcp & 'remsh host -l user' (rlogin) work fine.

The remsh is initiated on a host running SunOS 5.8 to a HPUX 11iv1 server through a firewall where ports 512/513/514 are open both ways.

Everything would appear to be enabled and .rhosts file is setup correctly (hence the successful rcp).

Has anyone come up against anything similiar and resolved the problem ?
(Please don't sugest ssh - we are unable to change the originating servers code).

Thanks in anticipation

 

[hoinz]

Imho you should tell us more about why the result of an remsh command does not meet your expectations...
Crystal balls are not a common tool at Tek-tips.

What happens, or what doesn't happen?
What is odd ?
Any error messages?
... ?

 

[spamly]

This may not help, but the first think I check when playing with remote authentication is hostname resolution. Make sure that both systems know who the other is. Check hosts files, DNS, and resolv.conf files as necessary.

 

[elgrandeperro]

rcp works, so it can't be authentication.

You might want to "quiesce" some of the .login, .cshrc or startup commands, check if you are using rbac.

Here is how I usually figure this out:

The remshd/rshd is usually spun off inetd. I simply truss or strace with the follow (-f) option inetd. Hopefully the system will be a bit quiet. This will tell you what files are being consulted and where rshd is quitting.