running script in cron vs cmd line

[westwood01]

When I run the below script from the cmd line it works as expected. The script will accurately detect whether an ssh connection is available. But, when I put this same script in cron it fails every time, even when an ssh connection is available. Any ideas? Or possibly a better method to monitor ssh connectivity? Thanks guys.

#!/bin/ksh
#set -x
ssh somehost 2>/dev/null &
sleep 5
kill $!
if [ $? -eq 0 ];then
echo $? > /export/home/oper/host.exit
echo successful >> /export/home/oper/host.exit
else
echo $? > /export/home/oper/host.exit
echo failed >> /export/home/oper/host.exit
cat /export/home/oper/host.exit|mailx -r me@mywork.com -s "somehost failed ssh connection" me@mywork.com
fi

 

[cspilman]

depends on where ssh is.

If ssh is located, for example, under /usr/local/bin, then you will need to either source root's profile in the script or use absolute path names in the the script for the command.

Regards,
Chuck

 

[columb]

The environment used by cron is different from the environment used when you're running a command line. In particular .profile isn't run. This can be fixed in a number of ways but the simplest is to make the command

su - root - c /path/to/myscript

Ceci n'est pas une signature
Columb Healy

 

[KarveR]

I find this much easier.

if nmap -p 22 wo9|grep "22/tcp open"
then echo "ok"
else echo "failed"
fi



______________________________________________________________________
There's no present like the time, they say. - Henry's Cat.

 

[westwood01]

I tried using the first two suggestions:
specifying full ssh path
adding su - root -c /path/to/myscript to cron

Script still fails.

I'll have to install nmap to try the nmap option, since I dont have it by default w/ solaris10.

 

[columb]

Try

1. Reinstate the 'set -x' or make the first line '#!/bin/ksh -x'
2. Amend the cron entry to

su - root -c "/path/to/myscript > /tmp/mylog 2>&1"

And post the output.

Ceci n'est pas une signature
Columb Healy

 

[westwood01]

# more mylog
kill: 16760: no such process

I'm now thinking that the script is returning a failed run code, since the kill $! is failing to find a process. Is this more a problem with the writing of the script as opposed to running it out of cron?

 

[columb]

Hmm, I'm not sure that ssh can work in this way - which is, after all, your original point.

Howabout

#!/bin/ksh

ssh somehost ls >/dev/null 2>&1 && RES=success || RES=failure
echo ssh to somehost returned $RES | mail -s "ssh on somehost" me@myhost

Of course this doesn't e-mail the result code but I don't think that's an issue.


Ceci n'est pas une signature
Columb Healy

 

[westwood01]

thanks guys, I ended up going a different route. Instead of trying to initiate an ssh or telnet session in the script I just did an rsh. This ultimately gave me what I was looking for, since the box I was testing was oddly rebooting into single user mode, so if my rsh (telnet or ssh) failed I got what I was looking for. thanks again.