OpenTelemetry 远端采样与导出策略（Tail Sampling、Batch 与验证）

概述尾部采样在收集完整链路后按策略决定保留与丢弃，适合高流量与成本控制。结合批处理与重试队列，既减少导出压力又保留关键错误与热点服务的链路。关键实践与参数采样决策: `tail_sampling` `decision_wait=5–10s`策略组合: 错误优先、热点服务优先、默认速率限制导出与重试: `batch` `memory_limiter` `retry_on_failure`协议与端点: `otlp`/`otlphttp` 指向后端(Tempo/Jaeger)示例/配置/实现receivers:

otlp:

protocols:

http:

grpc:

processors:

batch:

timeout: 5s

send_batch_size: 1024

memory_limiter:

limit_mib: 512

spike_limit_mib: 128

tail_sampling:

decision_wait: 5s

policies:

- name: error-traces

type: status_code

status_code:

status_codes: [ERROR]

- name: important-service

type: string_attribute

string_attribute:

key: service.name

values: ["checkout", "payment"]

- name: default-rate

type: rate_limiting

rate_limiting:

spans_per_second: 200

exporters:

otlphttp:

endpoint: http://tempo:4318

service:

pipelines:

traces:

receivers: [otlp]

processors: [memory_limiter, tail_sampling, batch]

exporters: [otlphttp]

验证错误保留: 制造 `ERROR` 链路，确认采样后仍导出热点优先: 对 `service.name=checkout` 的高流量链路保留比例更高速率限制: 在峰值下导出速率受控，队列无积压延迟与资源: 记录导出延迟与内存使用，确保在阈值内注意事项决策等待时间影响端到端延迟与保留效果，需权衡策略顺序与组合会影响结果，先匹配高优先策略与日志与指标管线协同治理，统一后端容量规划关注后端存储成本与查询性能