@Joshqun Ismayilov wrote:
Yes, / we can reduce ARP timeouts and use shorter DHCP lease times to improve responsiveness.(Default ARP cache timeout Windows: 15-45 minutes,)
Thanks !
Yup, and we usually don't consider 15 to 45 minutes as acceptable HSRP failover times.
(Laugh, or ouch, the first time I bumped into this, was doing sometime, I recall, on a FW, that changed its MAC, but not the IP, and FW didn't do a gratuitous ARP. I was very new to networking, but why the heck were so many hosts not able to get to the Internet, when I fired up a host to try Internet, it worked fine. Again, laugh or ouch, but it was a very impressable learning experience, especially when behind you is your manager asking how did you break the network, and when will you fix it.)
Hmm, unsure just decreasing DHCP least times would help, as DHCP often renews a lease. I.e. unless host IP changes, it wouldn't flush its ARP cache, right?
Where what you suggest would work very well, is on separate NICs, assuming they see the active gateway physically drops the connection.
I mean, there's a like issue between even between routers, on a shared medium, that aren't using a dynamic routing protocol that incorporates some form of keep alive. If the peer router just drops off, the remaining router, too, will not switch to an alternate floating static until the ARP cache times out. (The latter might be addressed by SLA.)
So, although your "correct" response is, in fact, correct (and a bit ingenious for ordinary hosts), it's probably not a practical replacement for most FHRP host applications. (I recall, there are some other [host] protocols to deal with finding gateways, and possibly dealing with changing gateways, as needed, but also possibly many not be available on most hosts, as FHRPs usually satisfy the host need.)
