I can suggest how you can resolve the problem, because you usually have the same problem: you build some complex, interdepent, overly complicated thing and expect it to work. How about building one small thing, test it, building the next-bigger thing, test again, and so on, until you are finished. Each time you add something and it doesn't work any more you would know exactly where to look for, no?

So let us start:

Code:

[user2@server1 ~]$ ssh user1@server2

Does this work? Do you get a prompt on the remote server? Only if "yes", proceed - otherwise you have a problem with the connection. What it may be i don't know, but i would start with a ping to see if the server is reachable, etc..

If it works, what does this

Code:

[user2@server1 ~]$ sshpass -p mypassword ssh user1@server2

do? sshpass uses a tty to trick ssh into believing the password was entered at command line. Chances are that this tty is somehow not there, taken away, whatever. It also might be that user2 has a different environment and some PATH or alias or privilege to access a certain command (or some dozens of other similar differences) is the culprit. Do you still get the command prompt with this?

By now you should know how to continue: add one step at a time and test. Good luck.

Finally, to emphasize what Neo has said, here is a quote from the man page of sshpass:

You have been told this several times in earlier threads and, honestly, i don't understand why you insist on using the "least best" (or, in other words: worst) practice, even in the words of the people who have written the tool you are using.

bakunin