Hosts that report down but aren't

[Carroll, Jim P [Contractor]]

Rico Gloeckner wrote:
> I wouldnt let any other Check define if a Host is up or down, let
> alone Availability of ssh.
> 
> I usually (if check_host_alive is no option) simply use /bin/true for
> the Hostcheck - Nagios simply sees the Host as always up and i
> will find any Service checked.

Hhhmmm, very interesting.  I hadn't thought of this approach.  But there is
a caveat with doing so:

If the host in question  has a *lot* of services being checked (eg, 3 HTTP
ports, 1 FTP server, 1 SSH server, 1 SMTP server, and a suite of NRPE checks
(system load, memory, zombie check, total process check, number of users
check, 15 partitions being checked)), you stand to be swamped with pages
if/when the host goes down.  Granted, you can mitigate this by creating NRPE
dependencies (do a simple /bin/true or "echo 'NRPE is alive'" as being the
depended-on service), but you'll still get more than just one page.

On the plus side, if you're only checking for one or two services, the
/bin/true hostcheck approach could be quite appealing.

> One might write a check_raw Plugin which uses various tricks 
> to find out
> if a Host is up, i.e. (for UDP) sending a Packet and expecting
> ICMP_PORT_UNREACH within 5 seconds. That would most probably requiring
> the Binary setuid root. 
> 
> What do others think? 

I guess this would depend on how strict the firewall is.  If you take the
case where everything's been turned off, but SSH and HTTP have been
explicitly permitted (the "that which is not expressly permitted is
prohibited" school of thought), then you'll have to work with what you're
given.  Which brings us full circle back to SSH.  ;)

jc


-------------------------------------------------------
This sf.net email is sponsored by: See the NEW Palm 
Tungsten T handheld. Power & Color in a compact size!
http://ads.sourceforge.net/cgi-bin/redirect.pl?palm0001en