Re: SCTP failover

From: Armando L. Caro, Jr. <me@armandocaro.net>
Date: Fri Feb 11 2005 - 16:16:27 EST

('binary' encoding is not supported, stored as-is) ('binary' encoding is not supported, stored as-is) On Fri, 11 Feb 2005, Randall Stewart wrote:

> IMO considering Armando's work and all.. we really need:
>
> 1) 2 levels of failure.
> a) When you hit a Retranmist limit (set to say 1) you
> switch to the alternate
> b) When you hit a Max retransmit you mark the destination
> as down.
> c) If the HB sent to the place the t-o happened is sucessful
> then you clear your count and the primary stays where it
> was... Note you should send a HB right after a RTO on the
> place where the RTO happened IMO

Yes, I agree. Although you are "failing over" to the alternate on a single
timeout, the primary shouldn't be considered failed just yet. However, if
the primary is eventually marked as failed (eg, 5-6 consecutive timeouts),
then the sender would want to avoid using it as a possible dest to send
new data or rtxs until it becomes reachable again.

~armando

0-- --0
| Armando L. Caro, Jr. | Protocol Engineering Lab |
| www.armandocaro.net | University of Delaware |
0-- --0
Received on Fri Feb 11 16:28:28 2005