Performance Metrics

Bingsan exposes performance metrics via Prometheus at the /metrics endpoint.

Pool Metrics

Object pool utilization metrics help monitor memory efficiency.

bingsan_pool_gets_total

Type: Counter Labels: pool

Total number of Get() operations on the pool.

# Get rate per pool
rate(bingsan_pool_gets_total[5m])

# Total gets by pool type
sum by (pool) (bingsan_pool_gets_total)

bingsan_pool_returns_total

Type: Counter Labels: pool

Total number of successful Put() operations returning items to the pool.

# Pool efficiency (returns/gets)
rate(bingsan_pool_returns_total{pool="buffer"}[5m])
/ rate(bingsan_pool_gets_total{pool="buffer"}[5m])

bingsan_pool_discards_total

Type: Counter Labels: pool

Total number of discarded items (oversized or invalid).

# Discard percentage
rate(bingsan_pool_discards_total{pool="buffer"}[5m])
/ rate(bingsan_pool_gets_total{pool="buffer"}[5m]) * 100

bingsan_pool_misses_total

Type: Counter Labels: pool

Total number of pool misses requiring new allocations.

# Hit rate (estimated)
1 - (rate(bingsan_pool_misses_total[5m]) / rate(bingsan_pool_gets_total[5m]))

Pool Labels

Label	Values	Description
`pool`	`buffer`, `bytes`	Pool type identifier

Alerting Rules

Low Pool Hit Rate

- alert: LowPoolHitRate
  expr: |
    (rate(bingsan_pool_returns_total{pool="buffer"}[5m])
     / rate(bingsan_pool_gets_total{pool="buffer"}[5m])) < 0.8
  for: 5m
  labels:
    severity: warning
  annotations:
    summary: "Pool hit rate below 80%"

High Discard Rate

- alert: HighPoolDiscardRate
  expr: rate(bingsan_pool_discards_total{pool="buffer"}[5m]) > 10
  for: 5m
  labels:
    severity: warning
  annotations:
    summary: "High pool discard rate"

Recording Rules

Pre-compute expensive queries:

groups:
  - name: bingsan_pool_recording
    interval: 30s
    rules:
      - record: bingsan:pool_hit_rate:5m
        expr: |
          rate(bingsan_pool_returns_total[5m])
          / rate(bingsan_pool_gets_total[5m])

      - record: bingsan:pool_discard_rate:5m
        expr: rate(bingsan_pool_discards_total[5m])

Interpreting Metrics

Healthy Pool

gets_total:    1,000,000
returns_total: 1,000,000
discards_total:        50
misses_total:         100

Utilization: 100% (returns = gets)
Discard rate: 0.005%
Miss rate: 0.01%

Pool with Leaks

gets_total:    1,000,000
returns_total:   800,000  ← 200,000 missing!
discards_total:       100
misses_total:     200,100  ← High misses

Utilization: 80%
Miss rate: 20%

Action: Check for missing defer pool.Put() calls.

Pool with Large Responses

gets_total:    1,000,000
returns_total:   700,000
discards_total:  300,000  ← High discards!
misses_total:         50

Utilization: 70%
Discard rate: 30%

Action: Consider increasing MaxBufferSize if schemas are large.

Health Check Endpoints

/health

Basic health check (returns 200 if healthy):

curl http://localhost:8181/health

/metrics

Prometheus metrics endpoint:

curl http://localhost:8181/metrics | grep bingsan_pool

Metrics

On this page