Parenting vs Dependencies

Marc Powell marc at ena.com
Mon Feb 11 18:40:04 CET 2008



> -----Original Message-----
> From: nagios-users-bounces at lists.sourceforge.net [mailto:nagios-users-
> bounces at lists.sourceforge.net] On Behalf Of mark.potter at academy.com
> Sent: Monday, February 11, 2008 11:14 AM
> To: nagios-users at lists.sourceforge.net
> Subject: [Nagios-users] Parenting vs Dependencies
> 


> I have a medium sized network (100-200 hosts).
> Hosts are in cabinets on site (no off site monitoring yet)
> Cabinets are not setup by subnet (I think this is fairly normal)
> 1 switch per cabinet
> Multiple subnets across the farm
> 
> The boss and I agree that the switch in each cabinet needs to be part
of
> our monitoring setup and be part of some sort of parenting/dependency.
> This is where we cease seeing eye to eye. For the purpose of making
this
> easy lets say we are dealing with three subnets and ten cabinets only.
> Here is what the boss wants (at least as far as I can understand):
Setup
> three hosts per cabinet to represent the three subnets, that actual IP
> would be the IP of the switch. Parent the hosts in the cabinet to
these
> and parent these to the gateways for the subnet. This is, with the
> parameters above, thirty extra hosts, and IMO a management nightmare.
> 
> Since the switch is not part of the route my thought is that we parent
in
> the normal manner as I understand it, host parents to router, and so
on
> and at that point we would make the hosts have a dependency on the
switch
> through which their connection passes. The current setup doesn't take
into
> account the switches at all but rather is setup to where each host is
> parented to it's gateway IP and that has been working rather well but
if a
> switch dies we need to be able to see that via nagios and not have 20
or
> servers and their services going off all at once.. I cannot see why
adding
> tens of special hosts for the boss's parenting solution will help
anything
> in the long run.
> 

Both solutions sound overly complex to me. It seems to me that you only
need to add the 10 switches, with their own IP's, as parents. I expect
that you shouldn't really care what subnet a server is on through that
switch, you should only be thinking about physical connections. If
Switch1 goes down, what servers are connected to it and unreachable as a
result? Do you think the switch could fail in such a way that one of the
subnets would work and the others won't? It's been my experience that a
switch works or it doesn't, outside of human error. 

The parents directive is used to determine network outages (down v.s.
unreachable) and would represent the problem correctly if your switch is
down. In your situation I would end up with something like --

Host (subnet1) has parent Switch1 has parent Router1
Host (subnet2) has parent Switch1 has parent Router1

Host (subnet1) has parent Switch2 has parent Router1
Host (subnet3) has parent Switch2 has parent Router2

And so forth.

Is there some additional complexity that I'm not grasping? Using
dependencies _would_ add the tens of extra definitions that you don't
want because you'd need a host dependency for host1 -> switch1, then
another for switch1 -> router1, etc, etc.

--
Marc

-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2008.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
_______________________________________________
Nagios-users mailing list
Nagios-users at lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/nagios-users
::: Please include Nagios version, plugin version (-v) and OS when reporting any issue. 
::: Messages without supporting info will risk being sent to /dev/null





More information about the Users mailing list