I have deployed apollo federation to our on-prem Docker Swarm cluster. I have taken notice that sometimes responses get dropped (return timeouts) when I have more than one replica of a single Apollo Graphql Subgraph.
For instance, my users node, just querying users, will time out the first query. Then immediately when I re-run the query, it works just fine.
Any possible reason this could be happening or possibly not supported within subgraphs to be distributed like this?