Gateway 重启防丢消息：启用并验证 outbound 写前队列恢复

用于“gateway 重启后 outbound 丢消息”场景。前置：已升级到 2026.2.13。步骤：构造发送压测→触发重启→观察写前队列恢复重试→核对线程/reply 路由。关键点：重启后应继续 drain 而非丢失。验证：重启窗口内消息最终送达且无错路由。风险：重复重试策略不当会导致重复消息。

GITHUBDiscovered 2026-02-15Author steipete

Prerequisites

Gateway runs 2026.2.13+ and at least one channel adapter is configured.
You can trigger controlled restarts in a non-production test window.

Steps

Send a burst of outbound messages with explicit `replyTo`/`threadId` metadata.
Restart gateway during in-flight delivery (`openclaw gateway restart`).
Monitor logs and confirm write-ahead queue recovers and retries pending items after restart.
Compare sent/received counts and verify thread/reply targets remain correct.

Commands

openclaw gateway status

openclaw gateway restart

openclaw logs --local-time

Verify

No outbound loss across restart; queued items deliver after recovery with correct thread context.

Caveats

If wrappers also retry at app layer, deduplicate IDs to avoid duplicate sends.
Provider-side throttling may delay eventual delivery and requires separate SLA checks（需验证）.

Source attribution

This tip is aggregated from community/public sources and preserved with attribution.

Open original source ↗

Visit original post