Issue Station
Sign In
Tools hereToolbox

openclaw - 💡(How to fix) Fix [Bug]: Gateway startup hangs indefinitely at pre-sidecar readiness path (never reaches 'ready') [2 comments, 3 participants]

Official PRs (…)
ON THIS PAGE

Recommended Tools

×6

Utilities matched from this issue’s tags and category — try them while you read without losing context.

GitHub issue graph ai analysis

Paste a GitHub issue URL. We fetch that issue, discover linked issues from bodies/comments/timeline, collect linked pull requests, and produce a structured English report.

The report is written in English Markdown for sharing and archival.

Helpful · Quick feedback

Loading…
GitHub stats
openclaw/openclaw#73865•Fetched 2026-04-29 06:14:00
View on GitHub
Comments
2
Participants
3
Timeline
2
Reactions
0
Author
lpendeavorslpendeavors
Participants
clawsweeper[bot]clawsweeper[bot]groff-labgroff-lablpendeavorslpendeavors
Timeline (top)
commented ×2

OpenClaw gateway container hangs during startup and never becomes healthy. Startup trace shows it reaches runtime.early but never progresses to sidecars.*, agent model, or ready log events. The container stays hung indefinitely.

Root Cause

OpenClaw gateway container hangs during startup and never becomes healthy. Startup trace shows it reaches runtime.early but never progresses to sidecars.*, agent model, or ready log events. The container stays hung indefinitely.

Fix Action

Workaround

None found. The gateway is stopped to avoid EADDRINUSE restart-loop churn.

RAW_BUFFERClick to expand / collapse

Description

OpenClaw gateway container hangs during startup and never becomes healthy. Startup trace shows it reaches runtime.early but never progresses to sidecars.*, agent model, or ready log events. The container stays hung indefinitely.

Environment

  • Image: ghcr.io/openclaw/openclaw:2026.4.24
  • Host: Ubuntu VPS, Docker 27.x, network_mode: host
  • Config: Qwen model via vLLM provider (OpenAI-compatible), Discord enabled
  • Port: 18793 (no conflict — confirmed free)
  • Data root: Fresh, no state carried over

Steps to Reproduce

  1. Deploy gateway with network_mode: host on a machine already running 3 other healthy OpenClaw gateways (ports 18789, 18790, 18791)
  2. Start container with docker compose up -d
  3. Container starts, logs show runtime.early, then hangs
  4. Health check never passes; wget localhost:18793/healthz never responds

What We Ruled Out

Every one of these was tested individually — the hang persists regardless:

  • Model API: /v1/chat/completions returns 200 from inside the container (curl works)
  • Discord: Disabled entirely (blank token) — still hangs
  • Provider config: Tried vllm and local-qwen provider IDs, with and without compat.requiresStringContent — still hangs
  • Identical config: Copied exact working config from Ninja gateway (same Qwen model, same image, same host) — still hangs
  • Bind address: Tried loopback instead of lan — still hangs
  • Health monitor: channelHealthCheckMinutes: 0 — still hangs
  • Plugin allowlist: Removed vllm from plugins.allow — still hangs
  • Fresh data root: Created entirely new data root + fresh Docker volumes — still hangs
  • Different ports: Tried multiple free ports — still hangs

Diagnostic Evidence

Startup trace

OPENCLAW_GATEWAY_STARTUP_TRACE=1 shows hang after runtime.early, before any sidecars.* event.

Node diagnostic report

Captured via NODE_OPTIONS="--report-on-signal --report-signal=SIGUSR2":

  • No JS stack trace (not stuck in JS)
  • libuv handles show idle event loop
  • Suggests the process is waiting on an internal native/uv condition that never resolves

EADDRINUSE pattern

  • Healthy gateways (Ronin, Shinobi) log one EADDRINUSE during startup but recover within ~120s
  • Katana logs the same EADDRINUSE but never recovers
  • This suggests the gateway's internal HTTP listener collides with itself during the pre-sidecar readiness path, and the retry/recovery logic fails to complete for this instance

Other Healthy Gateways on Same Host

All using identical shared template, same image, same host:

GatewayPortModelStatus
Ninja18789Qwen (vLLM)Healthy
Ronin18790MiMo v2.5-proHealthy
Shinobi18791Kimi K2Healthy
Katana18793Qwen (vLLM)Hung

Expected Behavior

Gateway should progress past pre-sidecar readiness and begin serving requests within ~60s, like the other 3 gateways on the same host.

Workaround

None found. The gateway is stopped to avoid EADDRINUSE restart-loop churn.

extent analysis

TL;DR

The OpenClaw gateway container hangs during startup due to a potential port collision or internal HTTP listener issue, and a workaround or configuration change is needed to resolve the issue.

Guidance

  • Investigate the EADDRINUSE pattern and the gateway's internal HTTP listener to determine if there's a collision or conflict with other gateways on the same host.
  • Try changing the port or configuring the gateway to use a different network mode to avoid potential conflicts.
  • Review the Node diagnostic report and startup trace to identify any other potential issues or clues that may indicate the root cause of the problem.
  • Consider temporarily stopping or reconfiguring other gateways on the same host to see if the issue persists or if it's related to a specific gateway or configuration.

Notes

The issue seems to be related to the specific gateway instance (Katana) and its configuration, as other gateways on the same host are working correctly. However, without further information or debugging, it's difficult to determine the exact root cause of the issue.

Recommendation

Apply a workaround by changing the port or network mode configuration to avoid potential conflicts, as the issue seems to be related to the gateway's internal HTTP listener or port collision.

Vote matrix · Quick signals

Works
Did the solution work? Tap to confirm.
Easy Fix
Was it a quick fix?
Time Saver
Did it save you time?
Blocking
Was it severely blocking?
Common Issue
Are others likely hitting this too?
Flaky / Intermittent
Is it intermittent?
Verified / Reproducible
Can you reproduce it reliably?
Loading…
#api#vector store#embedding generation#cache error#pipeline error

Still need to ship something?

×6

Another batch ranked right after the header list — different links, same matching logic.

Back to top recommendations

TRENDING

openclaw - 💡(How to fix) Fix [Bug]: Gateway startup hangs indefinitely at pre-sidecar readiness path (never reaches 'ready') [2 comments, 3 participants]