p.mayers at imperial
Jun 12, 2012, 12:56 AM
Post #2 of 3
On 06/11/2012 05:55 PM, Arne Larsen / Region Nordjylland wrote:
> Hi all
> Can someone help me out here.
> If we want to migrate from glbp to hsrp anycast in the future, can that be done with out any downtime.
Not zero. You can do it with very little downtime though. Bear in mind
that, when you configure HSRP, HSRP will transition between the
LISTEN/SPEAK states before going active.
When making changes like this (e.g. moving from HSRPv1 to HSRPv2 to gain
IPv6 support) we normally do the following:
1. Shutdown the interface on the secondary, to avoid confusion
2. Turn the HSRP timers down to very (unreasonably) aggressive
settings, e.g. "standby X timers msec 50 msec 100"
3. Config/reconfigure HSRP; fast timers mean it takes over rapidly
4. Re-set the timers to something more reasonable
5. Configure the standby, then enable its interface
For your case, I would imagine you'll need to prepare a blob of config
that disables GLBP, sets HSRP timers to ultra-aggressive and then
enables HSRP, and paste it all in at once.
> Will all servers update their arp table, if the hsrp function propagates gratious arp.
I've never seen this fail except on REALLY REALLY old things; Solaris
2.6 had a problem, IIRC.
cisco-nsp mailing list cisco-nsp [at] puck
archive at http://puck.nether.net/pipermail/cisco-nsp/