mirror of
https://github.com/fawney19/Aether.git
synced 2026-01-08 10:42:29 +08:00
b0d295c6c9
- model_alias_resolution_total: track resolution methods and cache hits - model_alias_resolution_duration_seconds: measure resolution performance - model_alias_conflict_total: monitor alias conflicts across GlobalModels
68 lines
2.0 KiB
Python
68 lines
2.0 KiB
Python
"""
|
|
Prometheus metrics for monitoring
|
|
"""
|
|
|
|
from prometheus_client import Counter, Gauge, Histogram
|
|
|
|
# 并发槽位占用时长分布
|
|
concurrency_slot_duration_seconds = Histogram(
|
|
"concurrency_slot_duration_seconds",
|
|
"Duration of concurrency slot occupation in seconds",
|
|
["key_id", "exception"],
|
|
buckets=[0.1, 0.5, 1, 5, 10, 30, 60, 120, 300, 600], # 0.1s 到 10 分钟
|
|
)
|
|
|
|
# 并发槽位释放计数
|
|
concurrency_slot_release_total = Counter(
|
|
"concurrency_slot_release_total",
|
|
"Total number of concurrency slot releases",
|
|
["key_id", "exception"],
|
|
)
|
|
|
|
# 当前并发槽位使用数
|
|
concurrency_slots_in_use = Gauge(
|
|
"concurrency_slots_in_use", "Current number of concurrency slots in use", ["key_id"]
|
|
)
|
|
|
|
# 流式请求时长分布
|
|
streaming_request_duration_seconds = Histogram(
|
|
"streaming_request_duration_seconds",
|
|
"Duration of streaming requests in seconds",
|
|
["key_id", "status"],
|
|
buckets=[1, 5, 10, 30, 60, 120, 300, 600, 1800], # 1s 到 30 分钟
|
|
)
|
|
|
|
# 请求总数（按类型）
|
|
request_total = Counter(
|
|
"request_total",
|
|
"Total number of requests",
|
|
["type", "status"], # type values: streaming/non-streaming, status: success/error
|
|
)
|
|
|
|
# 健康监控相关
|
|
health_open_circuits = Gauge(
|
|
"health_open_circuits",
|
|
"Number of provider keys currently in circuit breaker open state",
|
|
)
|
|
|
|
# 模型别名解析相关
|
|
model_alias_resolution_total = Counter(
|
|
"model_alias_resolution_total",
|
|
"Total number of model alias resolutions",
|
|
["method", "cache_hit"],
|
|
# method: direct_match, provider_model_name, alias, not_found
|
|
# cache_hit: true, false
|
|
)
|
|
|
|
model_alias_resolution_duration_seconds = Histogram(
|
|
"model_alias_resolution_duration_seconds",
|
|
"Duration of model alias resolution in seconds",
|
|
["method"],
|
|
buckets=[0.001, 0.005, 0.01, 0.025, 0.05, 0.1, 0.25, 0.5, 1.0], # 1ms 到 1s
|
|
)
|
|
|
|
model_alias_conflict_total = Counter(
|
|
"model_alias_conflict_total",
|
|
"Total number of alias conflicts detected (same alias maps to multiple GlobalModels)",
|
|
)
|