The Redis installation-token cache (src/selfhost/redis-token-cache.ts) that lets warm tokens survive restarts emits no metrics, so operators can't see cold-start re-mint rate (the whole point of #perf). Add a gittensory_redis_token_cache_total{result="hit"|"miss"} counter in get(): hit when a valid {token,expiresAtMs} is returned, miss when raw is null OR the parsed shape is invalid.
Deliverables
References
src/selfhost/redis-token-cache.ts:15
src/selfhost/redis-token-cache.ts:31
src/selfhost/metrics.ts:48
size: S · gittensor:feature
The Redis installation-token cache (src/selfhost/redis-token-cache.ts) that lets warm tokens survive restarts emits no metrics, so operators can't see cold-start re-mint rate (the whole point of #perf). Add a gittensory_redis_token_cache_total{result="hit"|"miss"} counter in get(): hit when a valid {token,expiresAtMs} is returned, miss when raw is null OR the parsed shape is invalid.
Deliverables
References
src/selfhost/redis-token-cache.ts:15src/selfhost/redis-token-cache.ts:31src/selfhost/metrics.ts:48size: S · gittensor:feature