Envoy gateway deployment in front of code indexing web servers
Problem to Solve
Queries against the new code indexing web servers must reach an instance that holds the relevant data in a warm cache and runs a compatible version. Plain Kubernetes service load balancing spreads requests at random, which defeats the cache and complicates pagination and rate limiting.
Proposed Solution
- Deploy an Envoy gateway in front of the web servers in the Orbit clusters.
- Route requests to instances by cache shard and version so that repeated queries hit a warm instance.
- Use the gateway as the place for pagination support and, later, rate limiting.
Agent context — extended analysis, alternatives considered, related-code dumps
Source: item 9 of the "Scalable Code Indexing" section in Orbit GA Planning. Estimated size: 5 days. Envoy Gateway is already evaluated for the Orbit gRPC passthrough plan, so the same controller can be reused. Routing control is tracked separately in the follow-up issue for Envoy configuration control.