This happens on RedHat 7.3 – haven’t tried on something more recent.
Assume you have a machine where root can rsh to localhost (yeah, I know, but the machine isn’t on a network where there are any users, so it’s not as bad as it could be.)
rsh localhost "/etc/init.d/snmpd restart; echo 'DONE'"
will echo the “DONE” but never return unless you hit ^C twice.
rsh localhost "/etc/init.d/snmpd restart
works as expected.
Now take a script that does rsh'es to a bunch of machines and runs apt-get on them (as well as on the local machine) and does various configuration tasks on both the local and the other systems, including restarting services. See script run. Now, take the entire script, and
put a
{
} 2>&1 | tee -a /var/log/upgrade.log
around the whole thing, and suddenly it never finishes.
See Paul waste his whole day trying different variations, each time requiring 45 minutes to put all the machines back to the version 3.3 configuration, and at least 20 minutes for the script to run. Can you say “bored and frustrated”, ladies and gentlemen?
And to top it all off, there’s an AIRMET for icing all along the route to Ottawa, so I won’t be flying after all.