Xymon Mailing List Archive search

TS - 'route:' statement for DUAL homed devices

2 messages in this thread

list Tom Schmitt · Wed, 2 Feb 2011 12:24:58 -0700 · 
I am using the 'route:' statement for dependencies currently.

 
In my environment, I have dual everything to the wiring closet, or dual homed servers.

 
Is there a way to have a device watch for BOTH paths back?

In other words, if both connections fail then the SERVER will turn YELLOW and show the unreachable cause.

 
3 Examples:

 
I can watch for a failure of "wiring-closet-switch" but cannot go any farther back

CISCO-6500A à  1 gig fiber connectionà wiring-closet-switch à SERVER

CISCO-6500B  à 1 gig fiber connection ---^

 
CISCO-6500A à top-of-rack-switch à SERVER

CISCO-6500B ----^

 
CISCO-6500A à SERVER

CISCO-6500B ----^

 
Would this work?

                SERVER  # route:top-of-rack-switch,CISCO-6500A,CISCO-6500B  conn  ...

 
Thanks,

 
Tom Schmitt

Senior IT Staff - R&D

L-3 Communication Systems West

640 North 2200 West

P.O. Box 16850

Salt Lake City, UT  XXXXX

Phone (XXX) XXX-XXXX

Cell      (XXX) XXX-XXXX

eFax    (XXX) XXX-XXXX

user-9c1ae820b621@xymon.invalid

           \\\\||////

             \ ~  ~ /  

             | @  @ |   

--oOo---(_)---oOo--
list Buchan Milne · Wed, 2 Feb 2011 23:09:17 +0200 ·
▸ quoted from Tom Schmitt
On Wednesday, 2 February 2011 21:24:58 user-9c1ae820b621@xymon.invalid wrote:
I am using the 'route:' statement for dependencies currently.

In my environment, I have dual everything to the wiring closet, or dual
homed servers.
Is this for NIC + switch redundancy, or do you need separate IP addresses? On different VLANs?
 
Is there a way to have a device watch for BOTH paths back?
This may not help you, but what we do is:
1)Wire two NICs to different switches (say eth0 and eth1)
2)Bond the two NICs (as say bond0)
3)If we need more than one VLAN, trunk both ports and run VLANs on the host (e.g. bond0.100 and bond0.101).

Then we monitor the "closest" interface to the monitoring server (which is multi-homed), and on the route tag, we put the HSRP address of the gateway for the subnet (bet it switch or firewall).

Note: we only set one 'default' gateway, it's easier to know exactly what path the packets take, what their source addresses will be, when you have another team unfamiliar with the intricacies of your environment maintain the firewalls.

Regards,
Buchan