openclaw - 💡(How to fix) Fix Gateway freezes after nested subagent activity, stops all Telegram polling [1 participants]

openclaw2026-03-24 14:46:00

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.

GitHub issue URL

Helpful · Quick feedback

GitHub stats

openclaw/openclaw#53745•Fetched 2026-04-08 01:24:02

View on GitHub

Comments

Participants

Timeline

Reactions

Author

rusvtiem

Participants

rusvtiem

After a coordinator agent calls subagents via sessions_send, the gateway process stays alive (HTTP 200 on healthcheck, PID running) but completely stops processing Telegram updates. Log output stops entirely — no incoming messages, no health-monitor, no sendMessage, nothing. All Telegram bots go silent simultaneously.

Error Message

Last entries before freeze:

[agent:nested] session=agent:skyler:main ... ANNOUNCE_SKIP

After this — complete silence in both gateway.log and gateway.err.log.

Error log context (shortly before freeze):

[agent/embedded] ... isError=true error=The AI service is temporarily overloaded
[agent/embedded] embedded run failover decision: ... decision=rotate_profile reason=overloaded

Previous occurrence (March 22): On the older version (2026.3.2), the same freeze pattern occurred, additionally accompanied by:

[gateway] full process restart failed (spawnSync launchctl ETIMEDOUT); falling back to in-process restart

Root Cause

Fix Action

Workaround

Kill and restart the gateway process:

launchctl kickstart -k gui/$(id -u)/ai.openclaw.gateway

Code Example

[agent:nested] session=agent:skyler:main ... ANNOUNCE_SKIP

---

[agent/embedded] ... isError=true error=The AI service is temporarily overloaded
[agent/embedded] embedded run failover decision: ... decision=rotate_profile reason=overloaded

---

[gateway] full process restart failed (spawnSync launchctl ETIMEDOUT); falling back to in-process restart

---

launchctl kickstart -k gui/$(id -u)/ai.openclaw.gateway

RAW_BUFFERClick to expand / collapse

Description

Environment

OpenClaw version: 2026.3.13 (also reproduced on 2026.3.2)
OS: macOS 15 (Darwin 24.3.0), Apple Silicon
Node: v25.8.0
Setup: 10 Telegram bot accounts (1 default + 9 subagents), each with separate bot token

Steps to Reproduce

Configure multiple agents (default + subagents with separate Telegram bots)
Send message to coordinator agent (e.g. "Saul") via Telegram
Coordinator calls subagent (e.g. "Skyler") via sessions_send
Subagent responds successfully
After subagent response, gateway log goes completely silent
All Telegram bots stop responding (including default/main agent)

Expected Behavior

Gateway should continue processing Telegram updates after subagent interaction completes.

Actual Behavior

Gateway process remains alive (responds HTTP 200 on port), but:

Zero log output after the last agent:nested entry
No Telegram polling — no incoming messages processed for any bot
No health-monitor checks running
No sendMessage attempts
Process must be killed and restarted to recover

Logs

Last entries before freeze:

[agent:nested] session=agent:skyler:main ... ANNOUNCE_SKIP

After this — complete silence in both gateway.log and gateway.err.log.

Error log context (shortly before freeze):

[agent/embedded] ... isError=true error=The AI service is temporarily overloaded
[agent/embedded] embedded run failover decision: ... decision=rotate_profile reason=overloaded

Previous occurrence (March 22): On the older version (2026.3.2), the same freeze pattern occurred, additionally accompanied by:

[gateway] full process restart failed (spawnSync launchctl ETIMEDOUT); falling back to in-process restart

Frequency

Reproduced 3 times in 2 days (March 22-24, 2026). Each time required manual launchctl kickstart -k to recover.

Workaround

Kill and restart the gateway process:

launchctl kickstart -k gui/$(id -u)/ai.openclaw.gateway

extent analysis

Fix Plan

To address the issue of the gateway process freezing after a subagent interaction, we will implement the following steps:

Increase logging verbosity: Modify the logging configuration to include more detailed logs, which can help identify the root cause of the issue.
Implement retry mechanism: Add a retry mechanism for the sessions_send method to handle temporary overloads and failures.
Handle AI service overload: Modify the AI service to handle overload situations more robustly, such as by implementing a queue or load balancing.

Code Changes

Here are some example code snippets to illustrate the proposed changes:

Increase Logging Verbosity

// Increase logging verbosity
const logger = require('logger');
logger.setLevel('DEBUG');

// Log detailed information about sessions_send method
logger.debug('Sending session to subagent:', subagentName);

Implement Retry Mechanism

// Implement retry mechanism for sessions_send method
const maxRetries = 3;
const retryDelay = 500; // 500ms

async function sendSession(subagentName, data) {
  let retries = 0;
  while (retries < maxRetries) {
    try {
      // Call sessions_send method
      const response = await sessionsSend(subagentName, data);
      return response;
    } catch (error) {
      logger.error('Error sending session:', error);
      retries++;
      await new Promise(resolve => setTimeout(resolve, retryDelay));
    }
  }
  // Handle max retries exceeded
  logger.error('Max retries exceeded for sending session to subagent:', subagentName);
  throw new Error('Max retries exceeded');
}

Handle AI Service Overload

// Handle AI service overload
const aiServiceQueue = [];

async function aiServiceRequest(data) {
  // Check if AI service is overloaded
  if (aiServiceQueue.length > 10) {
    // Handle overload situation, e.g., by returning an error or delaying the request
    logger.error('AI service is overloaded. Delaying request.');
    await new Promise(resolve => setTimeout(resolve, 1000)); // Delay for 1 second
  }
  // Process AI service request
  aiServiceQueue.push(data);
  // ...
}

Verification

To verify that the fix worked, you can:

Test the sessions_send method with a subagent and verify that the gateway process continues to process Telegram updates after the interaction.
Monitor the logs for any errors or issues related to the AI service overload or retry mechanism.
Verify that the retry mechanism is working correctly by simulating a temporary overload or failure of the AI service.

Extra Tips

Regularly review and monitor the logs to identify any potential issues or errors.
Consider implementing additional logging and monitoring tools to provide more detailed insights into the system's behavior.
Test the system thoroughly after implementing

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?

#API rate limit #retriever error #indexing error #inference speed #output truncation

Still need to ship something?

×6

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

Data

Security

Network

Code

UI/UX

Text

System

Multimedia

Protocol

API

Engineering

openclaw - 💡(How to fix) Fix Gateway freezes after nested subagent activity, stops all Telegram polling [1 participants]

Recommended Tools

GitHub issue graph ai analysis

Error Message

Root Cause

Fix Action

Workaround

Code Example

Description

Environment

Steps to Reproduce

Expected Behavior

Actual Behavior

Logs

Frequency

Workaround

extent analysis

Fix Plan

Code Changes

Increase Logging Verbosity

Implement Retry Mechanism

Handle AI Service Overload

Verification

Extra Tips

Still need to ship something?

TRENDING

openclaw - 💡(How to fix) Fix Gateway freezes after nested subagent activity, stops all Telegram polling [1 participants]

Recommended Tools

GitHub issue graph ai analysis

Error Message

Root Cause

Fix Action

Workaround

Code Example

Description

Environment

Steps to Reproduce

Expected Behavior

Actual Behavior

Logs

Frequency

Workaround

extent analysis

Fix Plan

Code Changes

Increase Logging Verbosity

Implement Retry Mechanism

Handle AI Service Overload

Verification

Extra Tips

Still need to ship something?

RELATED_DISCOVERY

TRENDING