By: Kevin Fletcher named 17 Jun 2020 at 2:49 p.m. CDT

4 Responses
Kevin Fletcher gravatar
I would like to understand how Cluster Manager works with an external load balancer. We are seeing strange behavior. We have an external load balancer setup and have pointed Cluster Manager to it. What we are seeing is that if we shut down the node1 and only run with node2 of Gluu server, it still tries to hit node1 even though it is down. We have confirmed this by removing node1 from the load balancer completely and the Gluu server still tries to hit the first node1 server. Is the Gluu servers doing something in the background themselves to load balance? Our goal is to remove single points of failure. We want a 2 node Gluu cluster with an external load balancer and an external Redis server. If we just used the internal Nginx load balancer on the Cluster Manager then it would become a single point of failure which is what we are trying to avoid.

By David van Hoose named 17 Jun 2020 at 3:35 p.m. CDT

David van Hoose gravatar
Additionally, we configured the external load balancer with a single node when we built it. We noticed that the IP address on the home page in Gluu would randomly change to one of the two nodes. This is not what we expected.

By Mohib Zico staff 17 Jun 2020 at 3:40 p.m. CDT

Mohib Zico gravatar
Hello Kevin, Unfortunately Cluster Manager is not doing any kind of load balancing yet but it's in our ToDo list for future releases. For now, you have to commission / decommission nodes from nginx or external LB side.

By David van Hoose named 17 Jun 2020 at 3:43 p.m. CDT

David van Hoose gravatar
Mohib, we didn't explicitly install nginx. How do we properly decommission nodes?

By Mohib Zico staff 17 Jun 2020 at 3:46 p.m. CDT

Mohib Zico gravatar
>> Mohib, we didn't explicitly install nginx. How do we properly decommission nodes? Load balancer which you are using should have config on that ( i.e. which ip addresses / hostnames are _under_ this load balancer ). From there you can do that.