UNIX for Dummies Questions & Answers

I'm new to scripting, and this forum has been invaluable in helping me out. I'm hoping I can get some personal help now though.

I have a korn script that takes a list of servers and either telnets or sshs into it (only some are set up for ssh). What I'm doing now is trying to telnet first, and if it doesn't work then I ssh. The problem is I get a broken pipe error if ssh is needed. The script still works fine, but people will be freaking out over the broken pipe. Is there a way to prevent the broken pipe error from showing, or a different way to check if the server should be telnet or ssh?

My current code:

Code:

(sleep 1
print <username>
sleep 1
print <password>
sleep 1
print "<some command I need output from>"
sleep 1
print "logout"
) | telnet server1 > $temp_file 2>/dev/null
 
if [ $? -ne 0 ]; then
    <code to ssh into server>
fi

I tried to build a file and do telnet < login_file but I couldn't get that work. I probably did that wrong though. Does anyone know how I'd build the file to redirect to the telnet command?

TIA!

Though I am amazed that this code works:

Maybe try appending this redirect of STDERR to the command which produces the error:

Code:

2>/dev/null

I edited my code. I actually already tried redirecting the STDERR.

Is the code that bad? I was basing it off of current scripts we have, which is how I even knew how to do the telnet in the first place....

That kind of thing is usually inside a here doc: the EOF marks the "boundaries", last boundary goes in column 1 - leftmost column. The other are not sensitive to columns.

Code:

telnet remotenode << EOF
username
password
command goes here
exit
EOF

process pipes are not files, they are memory objects and so are process persistent. That means they go away when the "reader" process ends - and when it ends too soon you get broken pipe errors. The reader is on | <right side> of the pipe.
The error you got was caused the the reader, telnet, exiting before the child process that was feeding data was finished. -- the part inside the ( )

Thanks so much, I will try this tomorrow. Would I be able to get the output of the command executed during the telnet session into a file?

This will not work because typeahead is not allowed for the username and password in telnet.

The original script in post #1 does work when all parameters are correct, with the exception of logout which should be exit .
Surely Remote Shell would be easier to work with - and that too can be configured for automatic login.