Skip to content

Discord inbound messages not reaching agent after gateway reconnects #75346

@bennettroller

Description

@bennettroller

Summary

Discord inbound delivery appears unreliable on Ben's Mac Studio OpenClaw gateway. User reports Q is not receiving Discord messages. Local WebChat is working.

Environment

  • Host: Mac-Studio
  • OpenClaw: 2026.4.25 (gateway log reports update available 2026.4.29)
  • Runtime node: Darwin arm64 / Node v22.22.1
  • Discord config: enabled, allowlist guild/channel config, requireMention=true for configured channels

Evidence from local logs

Recent gateway log shows repeated Discord websocket closes/reconnects:

2026-04-30T15:44:58.722-07:00 [discord] gateway: Gateway websocket closed: 1006
2026-04-30T15:44:58.728-07:00 [discord] gateway: Gateway reconnect scheduled in 1107ms (close, resume=true)
2026-04-30T16:00:53.616-07:00 [discord] gateway: Gateway websocket closed: 1000
2026-04-30T16:00:53.616-07:00 [discord] gateway: Gateway reconnect scheduled in 916ms (close, resume=true)
2026-04-30T17:32:37.613-07:00 [discord] gateway: Gateway websocket closed: 1006
2026-04-30T17:32:37.614-07:00 [discord] gateway: Gateway reconnect scheduled in 1131ms (close, resume=true)
2026-04-30T18:30:30.937-07:00 [discord] gateway: Gateway websocket closed: 1000
2026-04-30T18:30:30.938-07:00 [discord] gateway: Gateway reconnect scheduled in 951ms (close, resume=true)

openclaw status also hangs/times out locally rather than returning quickly. Watchdog previously reported /opt/homebrew/bin/openclaw status timed out after 45s.

Process sample at report time showed openclaw-gateway consuming high CPU (~39%) and large RSS (~2.5GB).

Expected

Discord gateway remains connected or self-heals after close/reconnect events, and allowlisted/mentioned Discord messages are enqueued to the agent.

Actual

User reports no inbound Discord messages are reaching Q. Logs show reconnects but no obvious inbound message handling after the report.

Initial suspicion

Could be one of:

  • Discord gateway reconnect/resume state goes stale after close 1000/1006.
  • Main gateway event loop is wedged or overloaded enough that Discord receives are dropped/delayed.
  • CLI/status path shares a stuck internal dependency and health checks are masking a degraded gateway.
  • Existing 2026.4.29 Discord/gateway fixes may already address part of this; needs verification on Ben's config.

Acceptance / test queue

  1. Apply candidate fix or upgrade path.
  2. Restart gateway safely.
  3. Ben sends a test message in the allowlisted Discord channel with required mention.
  4. Q confirms receipt and replies from Discord.
  5. openclaw status returns within 10s.

Assign/owner: Q / Ben testing queue.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type
    No fields configured for issues without a type.

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions