Hello,
Surprisingly, the VRRP failover to slave is not triggered by all WAN links being in health check failed ( 
).
My request is simple, to add it as a trigger element to swap between master/slave.
I have a ticket open and I’ll get a special firmware to support it but I was really surprised about that, I think others need to know. It is always advertised as “if WAN is Disconnected”, meaning that the ethernet port is “link down” but in many scenarios with switches, port will stay up !
I provided the answer below.
HA failover is based on the criteria below:
WAN health check failure shouldn’t trigger the HA failover based on the design. Do allow me to explain this. WAN health check failure is due to the WAN interface failed to contact the health check target (e.g. 8.8.8.8). Normally, this is caused by the WAN connection (WAN link). For example, when WAN1 of the master device failed to contact 8.8.8.8, the slave device will face the same problem even the HA failover is triggered (since the master and slave are connected to the same WAN link). So, from the design point of view, there is no point to trigger failover when the WAN health check failure.
Anyway, maybe you have a special used case of this. Can you help to elaborate more of your used case?
Thanks.
TK,
Switch port of the master can be faulty, configuration of the switch or firewall can have issues for the master, lines behind master and slave can be different ( to avoid lightning damaging both master and slave) … Master could be a fixed WAN only (balance) and slave a full wireless WAN (HD2 dome) …
I see many reasons to have different WANs behaviour and a failover trigger.
Hello,
In your post you say.
HA failover is based on the criteria below:
The last condition (all WANs down) doesn’t seem to work. It seems not working on release 8.1.x…
Regards,
Hedi
HA failover will be triggered for the situation above since the slave device failed to receive VRRP hello packet from the master device.
We wish to know more about this deployment. This used case may help for the feature development. Any reason the WAN link for master and slave devices is different? Both master and slave devices are not next to each other?
Thanks.
@HA13029, All WANs down is referring to the all WANs status show No Cable Detected. May I know this is your case?
Hello,
I can’t remember the state of the link in the GUI.
But, I’m sure that NO WAN Cable were plugged into to the device.
I plugged this device to the network (LAN interface ONLY, the device was configured with HA master priority), and even if NO WAN cables were plugged, the device takeover… In my point of view, the device should stay in Slave state because no WAN cabled were plugged…
Regards,
HA
This is what we experienced:
Configuration issue that could cause a complete isolation of the site. Only way to recover would have been to poweroff the master. Switch ports were up, no failover triggered.
@HA13029, this is my HA testing with 8.1.1. I tested with a WAN link only.
The device below is master and the device above is slave.
I disconnect WAN1 from master device. The slave device picked up the master role. The master role stays with the slave device.
Both devices are running 8.1.1 firmware.
The HA is working as expected. Please verify your setup. Please contact your point of purchase if you need further help.
Thanks.
@Venn, thanks for your use case. We plan to implement the requested feature in 8.2.0 after having a discussion with engineering team.
Other use case is when mobile routers have cards with specific quota. Once the quota of the master expires, it should switch to the slave…
