For the moment we do rlogin to a Solaris 2.4 system. We used to have
a SunOS 4.1.X system. The problem is when the user from the PM drops
the line and then the SUN does not log out the user. The process
just takes a lot of CPU.
I have solved with with setting tcp_keepalive_interval to 10000 and
runing a perl script that kills them. (I do ha match between
/usr/ucb/ps -xau and the w command.)
Sun Support in Sweden tells me that the system that does the rlogin,
in this case the PortMaster, have to do a proper hangup. They say
that Sun system does this.
Is that true? If yes, why does not the PM do it right or?
--Bo Kullmar
ABC-Klubben, the home of pmwho, ftp://pub/pmwho/pmwho.c