The issue (http://trac.tools.ietf.org/wg/ipsecme/trac/ticket/196) was
fixed in -03.
This situation (multiple failovers) is mentioned in the third bullet of
Sec. 5.1, with the newly-active member required to synchronize its state
into other cluster members. It is noted that some race situations are
still possible in such cases, resulting in a replay-like situation and
the IKE SA being torn down.
Randomization of counter increment values is a possible remedy here. We
believe this situation is rare - and implementation dependent - so this
possibility is not mentioned in the text.
_______________________________________________
IPsec mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/ipsec
