From: Thomas Alexander (talexander1@xxxxxxxxxxxxx)
Date: Mon Feb 05 2001 - 12:59:48 GMT-3
Brian,
I think your problem is the idle-timeout on the called router, not
the router that is calling (that has dialer watch configured).
Dialer watch, when is enable disabled idle-timeout on the calling router
until
the route is restored by the primary connection.
You could do 2 things:
1) Make all IP interesting on the called router (for example the Hub
router), so it never
reaches the idle-timeout.
2) Make the idle-timeout to a very large value (ex. 10,000) on the called
router.
This will keep the line up until the calling router disconnects. The
calling router will
still use its default idle-timeout to disconnect the call once the
router is restored thru primary links.
Regds/ Tom
----- Original Message -----
From: Brian Hescock <bhescock@cisco.com>
To: <ccielab@groupstudy.com>
Sent: Sunday, February 04, 2001 7:38 PM
Subject: watch-group observation
> This looks like normal behavior but, in my opinion, is broken:
>
> - dialer watch-group configured for network 30.1.1.0, which is learned via
> ethernet 0
> - shut down ethernet 0 interface, lose route for 30.1.1.0 and isdn comes
> up due to no route for 30.1.1.0
> - eigrp forms neighbors over isdn and we get route for 30.1.1.0
> - isdn drops after 120 seconds, the default idle-timeout
> - isdn redials immediately since it has no route for 30.1.1.0
>
> Personally, I think IOS should be smart enough to know it only knows the
> route through isdn so it should keep the line up if it doesn't know of a
> route for 30.1.1.0 via some other interface. It's a waste of money to
> increase the idle-timeout so this doesn't happen as often because then
> isdn will always stay up longer.
>
> Anyone know of a way around this problem other than increasing the
> idle-timeout? Looks like I may need to file a sys-wish bug to get IOS
> changed so it doesn't drop the isdn link if the route is only known via
> isdn. Perhaps it's just me but I think is silly that we drop the link,
> only to bring it up again.
>
> Brian
>
This archive was generated by hypermail 2.1.4 : Thu Jun 13 2002 - 10:28:37 GMT-3