This is similar in nature to bsc#1017666 and bsc#1017668 in that when you build a package with osc build or rpmbuild, you dont want to have leftover processes after that filling up your build host system.
Wait a minute... But both issues assume that you don't use isolated environment for building, right? That means that epmd -kill may actually kill epmd daemon which has already been running in parallel when the build was started. And this way this will crash existing applications (like RabbitMQ) running on the host.
but the epmd man page says
* With relaxed command checking, the epmd daemon can be killed from
the localhost with i.e. epmd -kill even if there are active nodes
registered. Normally only daemons with an empty node database can
be killed with the epmd -kill command.
NicoK
Could you please be more verbose? Why do you need this?
This is similar in nature to bsc#1017666 and bsc#1017668 in that when you build a package with osc build or rpmbuild, you dont want to have leftover processes after that filling up your build host system.
Wait a minute... But both issues assume that you don't use isolated environment for building, right? That means that epmd -kill may actually kill epmd daemon which has already been running in parallel when the build was started. And this way this will crash existing applications (like RabbitMQ) running on the host.
but the epmd man page says * With relaxed command checking, the epmd daemon can be killed from the localhost with i.e. epmd -kill even if there are active nodes registered. Normally only daemons with an empty node database can be killed with the epmd -kill command.
That makes sense.
I tested this on a SUSE Cloud node with rabbitmq:
epmd -kill
Killing not allowed - living nodes in database.
Then probably return should be ignored like the following;
epmd -kill || /bin/true
?
I was thinking of that, too but found that it still returned 0 (aka success) in that case.