openclaw - 💡(How to fix) Fix Follow-up to #76828: still reproduces on 2026.5.18 + DM silent-drop [1 comments, 2 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#84215Fetched 2026-05-20 03:42:35
View on GitHub
Comments
1
Participants
2
Timeline
10
Reactions
1
Author
Timeline (top)
labeled ×7closed ×1commented ×1cross-referenced ×1

On 2026.5.18, the same silent-delivery pattern still reproduces.

New detail: we now have one DM drop in addition to topic-thread drops.

  • Prior reports: topic-thread focused
  • New evidence (same day/session):
    • 1 topic-thread silent drop
    • 1 DM silent drop (agent:main:telegram:default:direct:<dm_id>)

This suggests the bug is not only :topic:<id> session-key shape.

Error Message

Likely not only topic-key parsing. More consistent with intermittent natural-stream dispatch failure/race/flush miss where turn text persists but channel send is not emitted (and no hard error surfaces).

Root Cause

On 2026.5.18, the same silent-delivery pattern still reproduces.

New detail: we now have one DM drop in addition to topic-thread drops.

  • Prior reports: topic-thread focused
  • New evidence (same day/session):
    • 1 topic-thread silent drop
    • 1 DM silent drop (agent:main:telegram:default:direct:<dm_id>)

This suggests the bug is not only :topic:<id> session-key shape.

Fix Action

Fix / Workaround

Hypothesis update

Likely not only topic-key parsing. More consistent with intermittent natural-stream dispatch failure/race/flush miss where turn text persists but channel send is not emitted (and no hard error surfaces).

Code Example

08:25:35.816  [context-engine] deferred turn maintenance queued
              sessionKey=agent:main:telegram:group:<chat_id>:topic:6
              ... assistant text persisted
              ... NO [telegram] outbound send ok follows
...
08:30:54.580  inbound prod from user
...
08:32:34.164  [telegram] outbound send ok ... [ws] message.action

---

09:11:56.651  [context-engine] deferred turn maintenance queued
              sessionKey=agent:main:telegram:default:direct:<dm_id>
              ... assistant text persisted
              ... NO [telegram] outbound send ok follows
09:12:27.760  inbound prod from user ("I didn't get your message")
09:12:48.671  [telegram] outbound send ok ... [ws] message.action
RAW_BUFFERClick to expand / collapse

Follow-up to #76828: still reproduces on 2026.5.18, plus one DM silent-drop

I tried to add this as a comment on #76828, but that issue appears locked for new comments from my account. Posting here to preserve the data.

Original: #76828

Summary

On 2026.5.18, the same silent-delivery pattern still reproduces.

New detail: we now have one DM drop in addition to topic-thread drops.

  • Prior reports: topic-thread focused
  • New evidence (same day/session):
    • 1 topic-thread silent drop
    • 1 DM silent drop (agent:main:telegram:default:direct:<dm_id>)

This suggests the bug is not only :topic:<id> session-key shape.

Incident A (topic-thread)

08:25:35.816  [context-engine] deferred turn maintenance queued
              sessionKey=agent:main:telegram:group:<chat_id>:topic:6
              ... assistant text persisted
              ... NO [telegram] outbound send ok follows
...
08:30:54.580  inbound prod from user
...
08:32:34.164  [telegram] outbound send ok ... [ws] message.action

Incident B (DM, new surface)

09:11:56.651  [context-engine] deferred turn maintenance queued
              sessionKey=agent:main:telegram:default:direct:<dm_id>
              ... assistant text persisted
              ... NO [telegram] outbound send ok follows
09:12:27.760  inbound prod from user ("I didn't get your message")
09:12:48.671  [telegram] outbound send ok ... [ws] message.action

Same-session DM control window

In the same ~50 minute window, 11 DM turns had normal delivery signatures (outbound send ~2-9s before turn-maintenance). Exactly 1 had none.

Turn maintenance queuedNearest preceding outbound send okResult
08:23:21.46008:23:17.286ok
08:24:54.67008:24:50.279ok
08:34:43.53208:34:40.816ok
08:37:52.52108:37:45.930ok
08:38:36.16808:38:31.352ok
08:55:57.69208:55:53.365ok
08:59:33.74008:59:30.007ok
09:00:52.02009:00:48.271ok
09:04:27.73509:04:22.413ok
09:11:56.651none in pre-turn windowdrop
09:12:50.97409:12:48.671retry turn
09:14:29.82609:14:21.158ok

Hypothesis update

Likely not only topic-key parsing. More consistent with intermittent natural-stream dispatch failure/race/flush miss where turn text persists but channel send is not emitted (and no hard error surfaces).

If maintainers prefer, I can provide a redacted longer log window.

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…

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 Follow-up to #76828: still reproduces on 2026.5.18 + DM silent-drop [1 comments, 2 participants]