This server release improves observability of Tuist-managed runners so you can diagnose capacity issues and cold-boot risk more accurately.
- Dispatch telemetry is now more granular. The runners dashboard can distinguish between a genuine per-shape stall, an empty queue, and a polling warm pool, because distinct dispatch outcomes are tracked instead of being collapsed into one.
- Dispatch metrics are now tagged by fleet, so you can slice dispatch count and latency per runner shape.
- New allocation metrics expose the warm pool floor the allocator could not fund under capacity contention, which is the leading indicator for cold boots. You also see the full target replica count, the allocated replica count, and the configured warm floor per pool.
- The runners dashboard no longer mixes environment data in the headline and capacity tiles, so “queued vs running” numbers are trustworthy. A new “Allocation (cold-boot risk)” row charts warm deficit and target-vs-allocated by platform.