In message ,
Jason Martin writes:
>One of the problems I've seen with the distributed monitoring
>model is that the configuration of the central server and
>distributed servers must be kept strictly in sync. Might I
>suggest that some sort of interface be provided such that
>distributed servers can add services to the central server
>without having to restart it? I'd imagine some sort of scenario
>where on startup a distributed server sends its service and host
>configuration through in some sort of external command. If the
>central server restarts then a mechanism could be set up so that
>it can notify the distributed servers to resend their
>configuration.
This is also useful for "dynamic services". E.G. consider a switch
with 100 interfaces of which 40 of them are monitored. I would like to
create a service called:
switch1/interface:if1
that will only be shown if it is a problem. I will have an associated
switch1/interface host/service that I can automatically update
(e.g. with a simple rule in sec, or a cluster mechanism) to indicate
that all interfaces are ok. This helps prevent the mess of
host/service entries that are displayed if you really want to see the
status of your network in the following senario:
passive alert for switch1/interface when port 1 goes down
passive alert for switch1/interface when port 2 goes down
passive alert for switch1/interface when port 2 comes up
Without individual services you don't see that port 1 is still down.
However if you specify every single port as a service, then you have
to manage a ton of interfaces/services in the gui's.
Currently I use volatile service, but that is not quite doing the
trick since the status of failing services isn't displayed.
-- rouilj
John Rouillard
===========================================================================
My employers don't acknowledge my existence much less my opinions.
This post was automatically imported from historical nagios-devel mailing list archives
Original poster: [email protected]