#I've updated from 2026.4.22 to 2026.5.2 and my openclaw is not working anymore.

1 messages · Page 1 of 1 (latest)

lament basin
#

Setup: discord + gemini API + cron jobs

Hosted on Ubuntu VM

An unknown error occurred
gateway disconnected: closed | idle

I'm seeing LLM timeouts as well, but the Gemini calls do not have any issues.

16:33:27+00:00 error diagnostic {"subsystem":"diagnostic"} lane task error: lane=session:agent:main:discord:channel:1487379402215133205 durationMs=129181 error="FailoverError: LLM request timed out."

I'm not sure what part of the config is relevant here.

lament basin
#

@atomic cave any idea?

crisp wasp
#

Same issues

lament basin
#

even roolback didn't fix anything. So, went back to 5.2 and waiting now

dim storm
#

Same story here. Discord won‘t start. Brave somehow missing. Doctor says deactivated plugins. I can‘t really take this anymore, working with alpha software is really hard.

rugged vapor
#

are you only complaining here? or are you opening issues in github?

dim storm
#

@rugged vapor as far as I know this is a help channel. Not "open github issue channel" and Krill usually instructs you to do so when it‘s appropriate.

#

But it seems that Krill is down

crisp wasp
slow marlin
#

Force install the Discord plugin

openclaw plugins install @openclaw/discord@2026.5.2 --force --pin
openclaw plugins registry --refresh
openclaw gateway restart

Then verify:

openclaw doctor
openclaw plugins list | grep -A2 -B1 discord
tropic pasture
rugged vapor
#

I understand this is a help forum, I also have used it a lot of times. But I also believe that openins issues shows OpenClaw team where to put attention.

flat hatch
slow marlin
hexed loom
#

I just reinstalled openclaw, somehow discord is working again

novel ice
lament basin
#

Upgraded to 5.3:

19:27:45+00:00 error diagnostic {"subsystem":"diagnostic"} lane task error: lane=main durationMs=128762 error="FailoverError: LLM request timed out."

and

Config warnings ────────────────────────────────────────────────────────────────────╮
│ │
│ - plugins: plugin: installed plugin package requires compiled runtime output for │
│ TypeScript entry setup-entry.ts: expected ./dist/setup-entry.js, │
│ ./dist/setup-entry.mjs, ./dist/setup-entry.cjs, setup-entry.js, setup-entry.mjs, │
│ setup-entry.cjs │
│ - plugins: plugin: installed plugin package requires compiled runtime output for │
│ TypeScript entry index.ts: expected ./dist/index.js, ./dist/index.mjs, │
│ ./dist/index.cjs, index.js, index.mjs, index.cjs

south sandal
#

I think I found a cleaner repro for the openai/gpt-5.5 issue.

This is not an OpenAI model latency problem. In the OpenAI Platform logs, the Responses API request completes successfully and the assistant output is literally:

OK

Request metadata in OpenAI Platform:

model: gpt-5.5-2026-04-23
openclaw_session_id: test-openai-gpt55-config
openclaw_transport: stream

But OpenClaw CLI returns:

{
"status": "timeout",
"summary": "aborted",
"stopReason": "rpc",
"durationMs": 247294,
"agentMeta": {
"provider": "openai",
"model": "gpt-5.5",
"agentHarnessId": "pi"
},
"livenessState": "blocked"
}

The prompt was minimal:

Reply with only: OK

So the response is produced by OpenAI, but it does not seem to propagate back through OpenClaw’s stream handling into the session / CLI / UI.

My current guess:

OpenAI Responses stream completes
→ OpenClaw openai-transport / PI harness does not mark the run completed
→ session/UI/CLI waits until idle timeout

Also worth noting: earlier I saw this log line:

OpenAI responses tool schema strict mode downgraded to strict=false for openai/gpt-5.5 because 20 tool schema(s) are not strict-compatible

But in this minimal repro OpenAI still returns the answer, so the main issue looks like stream/result propagation rather than the model failing to answer.