ICMP Redirect creacting invalid route
From: Doug (drheams_at_yahooo.com)
Date: 07/28/03
- Next message: Hiranya Samarasekera: "killing sendmail"
- Previous message: Nick Austin: "Re: No floppy disk"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Date: Sun, 27 Jul 2003 22:30:54 -0700
Hello,
I have a redhat 9 box running nagios to monitor my wide area network. Nagios
pings all of the remote routers every 5 minutes to ensure connectivity and
sends an email when a node is unreachable. The problem I have is due to a
poorly designed network but it is not currently an option to repair it.
Because of the network design, when a wan link goes down the monitoring
system receives an icmp redirect from its default gateway which points it
towards a new invalid gateway. This icmp redirect causes the linux kernel to
add a new route to its cached routing table, viewable through the route -Cn
command, and retry the ping through the new gateway. This obviously times
out and I receive a text page from nagios. When the wan link comes back up
nagios never knows it because it is using an invalid gateway for the ping.
To further add confusion, tcp traffic such as telnet uses the correct route
and works perfectly but a ping or traceroute clearly use the invalid route
and fail. I am able to clear the cach with the ip route flush cache command
but I am trying to avoid the manual intervention. Is there a way to decrease
the time that these routes are cached when an icmp redirect is received? Or
even to disable the caching altogether? I realize this will create extra
traffic on the lan but it is my best option at the moment.
- Next message: Hiranya Samarasekera: "killing sendmail"
- Previous message: Nick Austin: "Re: No floppy disk"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Relevant Pages
|
