Forum
Notifications
Clear all
Topic starter
21/07/2025 5:48 am
Hello,
I have a working skudonet cluster with node 1 (eth0) using 192.168.1.71, node 2 (eth0) using 192.168.1.72 and the VIP node 1 / node 2 (eth0.1) using 192.168.1.90. When node 1 (.71) goes down, node 2 (.72) takes over and the VIP (.90) and all services on it move to node 2. So as far as I can tell my setup is correct.
My issue is that my UniFi Network Application (a separate product I use to manage some of my network) periodically reports a duplicate IP address for 192.168.1.90. Interestingly, it does not only happen when one of the skudonet nodes go down, it also happens at least once a day by whenever the skudonet cluster is operational. My questions are the following:
1. Does the cluster test itself once a day or do something else that would cause the VIP to switch nodes even if there is no reason for it to move?
2. Is it possible to have the virtual interface MAC addresses for node 1 and node 2 to be the same so it won't trigger duplicate IP
3. Is this not expected and perhaps the result of something in my cluster configuration that is good enough to make the cluster work, but not quite perfect?
Thank you in advance for any assistance you can provide.
21/07/2025 8:24 am
Regarding the issue you describe with UniFi Network Application, I see a normal behaviour when UniFi announces you duplicated IP, as this product saves IP and MAC, when the cluster switches the MAC address for the VIP changed an no one indicated a UniFi that it is normal, you should try to connect to UniFi with some API or similar and suggest that the change is correct.
If you need to include some additional tasks when the cluster changes role, please refer to the config file:
/usr/local/skudonet/app/upcarp/etc/skudonet-cluster.conf
$up_script and $down_script
Regarding question 1) There is not anything once a day in the cluster, it sounds to me that the cluster changes of role or at least it tries to take the control, please have a look to the logs and search "ucarp" events, if you see some issues trying to switch daily I would recommend you to check NTP synchronization or increase the dead ratio (by default 2). See /usr/local/skudonet/app/upcarp/etc/skudonet-cluster.conf
Regarding question 2) MAC addresses can't be duplicated in a network.
Regarding question 3), follow my instructions. Check the logs in/var/log/syslog, verify the dead_ratio, and increase it to 5 or 10. Also, check the NTP sync.
Regards
