if i add a script to be executed, metalog will block
all proceses actualy on write to the syslog pipe.
so just do call a command that does /bin/sleep 60, and
try to do an 'su - ' and if your system adds an entry
to syslog for processing you did su, your su process
will stall for 60 seconds.
as i.e. ssh commands allso will behave like that, your
server will become unreachable and any program
utilizing the syslog engine will halt.
If such a program does like that, you're bound.
I think, metalg should cmpletely fork off the process,
and put the output to another syslog pipe....
Logged In: YES
user_id=114429
Originator: NO
no, the problem here has nothing to do with the way forks are handled
metalog uses non-safe functions in its signal handler (like localtime() and syslog()) and the hang here comes from libc mutexes deadlocked
so all of metalog's logging from signal handlers needs to be rewritten / rethought
Logged In: YES
user_id=114429
Originator: NO
0.8-rc2 is out with this fixed