10-14-2019 01:01 AM - edited 02-21-2020 09:35 AM
We have an ASA failover pair but they are in two locations. The failover link goes through two switches (one on each site). As the most likely reason for a failover would be a power outage on the site with the active ASA, I was wondering if the failover actually worked. If there was a power outage, the switch would fail too and the secondary ASA just sees that the failover link is down and (might) not become active.
As this is a production network I'm not just going to pull the plug.
Has anyone some experience with this kind of setup?
10-14-2019 05:01 AM
Several things can trigger failover. Among them is failure to receive the heartbeat from the peer unit. That would happen whether it is a single cable connecting the units or multiple switches across two sites.
I've implemented scenarios like the one you describe and it works just fine.
10-15-2019 04:57 AM
Thanks for the reply!
I'm refering to the statement in this document:
HA active standby config guide
Table 1-1 Failover Behavior states that there will be no failover if the failover link failes during operation.
Our setup pretty much looks like Figure 1-2 Connecting with a Double Switch—Not Recommended (minus the ISL).
Not recommended but as a complete power outage is the most likely scenario, a separate switch for the failover link would not help anyway.
10-15-2019 05:18 AM
As mentioned in my previous post, as long as the monitored interfaces still have connectivity between the active and standby ASAs, if the failover link fails there will be no failover. But this is why it is very important that all the interfaces on the ASAs are in a monitored state. If an interface that is not monitored fails there will be no failover and clients connected to that interface will lose connectivity.
10-14-2019 12:03 PM - edited 10-14-2019 12:03 PM
What happens when the failover link fails is that the secondary ASA starts sending hello packets out the monitored interfaces (the interfaces MUST be monitored for this to happen). If the secondary ASA receive a reply it know that the primary is still active and will not assume the active role. If no reply is received after 3 attempts the secondary ASA becomes active.
10-15-2019 05:15 AM
Hmm, maybe I should try in in GNS3...
The configuration guide says no failover with a failed failover link but maybe with monitored IFs...
10-15-2019 05:25 AM - edited 10-15-2019 05:26 AM
testing in GNS3 is a pain when testing failover on ASA. You cannot just do a shutdown on the switch port to initiate a failover as the interface will always remain up as the link is virtual. You could try to remove the link connecting the ASA to the switch, though I have not tried that, but I assume it will give the result you are looking for.
But just to repeat myself. If just the failover link fails, there will be NO FAILOVER if other interfaces are monitored and connectivity is established between the primary and secondary ASAs.
If both ASAs are up, but connectivity between the ASAs on failover link and monitored interfaces fails, then you will have a split brain situation.
If an interface fails (not failover link interface) then a failover will happen, depending on what you have configured as your failover criteria.
Discover and save your favorite ideas. Come back to expert answers, step-by-step guides, recent topics, and more.
New here? Get started with these tips. How to use Community New member guide