Release 实战：用 v2026.3.1 健康探针把 OpenClaw 接入 Docker/K8s 自愈
Problem/Scenario: gateway卡死时容器平台无法及时感知。Prerequisites: 已升级v2026.3.1并可改编排配置。Steps: 1) 确认`/healthz`与`/readyz`可访问；2) 配置liveness/readiness；3) 设置失败阈值与重启退避；4) 进行故障注入验证；5) 依据恢复时延调优。Key commands/config: `openclaw gateway status`，探针路径`/health` `/healthz` `/ready` `/readyz`。Verification: 故障时可自动重启并恢复。Risks/Boundaries: 探针过激可能误杀慢启动实例。Source: v2026.3.1 release notes + 官方X发布。
GITHUBDiscovered 2026-03-03Author openclaw
Prerequisites
- OpenClaw upgraded to v2026.3.1+.
- Can edit Docker/K8s probe settings.
Steps
- Verify probe endpoints are reachable after upgrade.
- Configure liveness and readiness with conservative initial delays.
- Define restart backoff and failure thresholds.
- Inject a controlled failure and observe restart behavior.
- Tune probe intervals based on SLO recovery target.
Commands
openclaw gateway statusopenclaw gateway restartVerify
Under simulated unhealthy state, orchestrator restarts service and it returns healthy within target SLO.
Caveats
- Too frequent probes can cause restart loops on low-resource hosts.
- Path conflicts with legacy handlers should be regression-tested（需验证）.
Source attribution
This tip is aggregated from community/public sources and preserved with attribution.
Open original source ↗