[Nagiosplug-help] nsca_send and xinetd

[Earl C. Ruby III]

I am using nsca_send to send event messages to a central Nagios server running 
xinetd version 2.3.12-27. When I first started doing this it worked great, 
but as I've added more and more event handling to my server applications I'm 
now seeing event handling lock up more and more often on the Nagios server. 
(I currently have about 225 event checks implemented.) Usually restarting 
xinetd clears the problem, but I'm guessing I have something misconfigured 
and that is messing things up.

There's a setting in nsca.cfg that I think might be relevant:

---------------------------------------------------------------------------------------
# APPEND TO FILE OPTION
# This option determines whether or not the nsca daemon will
# will open the external command file for writing or appending.
# This option should almost *always* be set to 0!
# Values: 0 = open file for writing, 1 = open file for appending
append_to_file=0
---------------------------------------------------------------------------------------

If multiple xinetd sessions all try to write to the command file at the same 
time or if they were writing as the command processor was reading/truncating 
the file I'd assume that there would be contention and occasional lock-ups IF 
command-file locking was poorly-implemented. 

I just changed this setting to "append_to_file=1" to see what happens, but I'm 
concerned about the warning that "This option should almost *always* be set 
to 0!" Why should this option almost always be set to zero? Is command-file 
locking poorly implemented?

-- 
Earl C. Ruby III
Senior Systems Engineer / Developer
Switch Management