We're using a load balancer on AWS for nodes.lto.network. There've been issues where one of the nodes is still syncing, but already serving requests. Also the endpoint can simply be down.
Load balancing can be done in the client instead, choosing one of 3 nodes.
Pick a node at random. Use a cookie (1 day ttl) to prevent switching nodes unexpectedly
Check if the node is healthy via /index/health
If unhealthy, pick another node
If none of the nodes are healthy, display a message saying the service is currently unavailable.
We're using a load balancer on AWS for nodes.lto.network. There've been issues where one of the nodes is still syncing, but already serving requests. Also the endpoint can simply be down.
Load balancing can be done in the client instead, choosing one of 3 nodes.
/index/health