#Chat slow to respond/start typing [73862]

1 messages · Page 1 of 1 (latest)

languid vine
#

recently updated openclaw after not doing so for a month or so. chat over discord is really slow - both in the time it takes for the typing indicator to appear and the time of response. is this a known issue with recent versions of openclaw? could I have a bad config?

serene urchinBOT
#

To help others find answers, you can mark your question as solved via Right click solution message -> Apps -> ✅ Mark Solution

fervent oar
#

What version are you on?

languid vine
#

2026.4.29

unborn robin
#

i ahve he same issue on 4.29

maiden roost
#

I had to revert to 4.23 . Tried up to 4.27. In my case there is a delay of around 6+ seconds .

wintry ravine
#

Im having the same issue v2026.4.26 happens a lot more on this version for me. Updated and still same thing.

sage flicker
#

Treat 4.23 as the last stable version - anything after causes hideous delays/issues and high cpu.

maiden roost
sage flicker
#

just today mentioning cpu I count 12 issues in github

wintry ravine
#

And beautiful

#

Everything started working again

#

Still slow but not on the level of the new versions

wintry ravine
maiden roost
#

@wintry ravine, what hardware are you using?

#

Restore an old image, rather than reverting, if possible. 4.29 leaves some files such as installs.json in ~/.openclaw/plugins/ (source Gemini AI)

sage flicker
wintry ravine
#

16 gb ram

wintry ravine
maiden roost
#

Our issue will be fixed in the next version. Steipete took care of it. Respect.

wintry ravine
amber tinsel
wintry ravine
#

Downgraded and now its around 6-20 secs

#

better than before for sure

amber tinsel
# wintry ravine better than before for sure

I am maintain two machine, one trying out new versions and the other one that is my live agent. its running like 10second usually, which I can accept it as its "human" feel, if you get what i mean LOL.

wintry ravine
#

Yep true

mighty moon
#

have the same issue with 2026.4.29, is there anywhere a list of "stable" versions?

lapis trail
#

I have same issue on 4.29 T-T

#

I tried both on wsl arch and on windows 11 same issue

hollow nest
#

Oh your agents are replying to yall? I worked yesterday like 4 hours to get mine to even reply again

sage flicker
#

I'm surprised they've left .29 up for so long considering all the issues

proud jackal
jolly dove
#

Any Danger in Downgrading? from 4.29 > 4.23

twin nebula
#

Anyone know how to even fix .23 issues? I downgraded to 23 but its still not working. Is there a command to reset or something?

wintry ravine
wintry ravine
#

What’s the issue you’re facing )

candid quest
#

Having the same issue with 4.29, Discord is taking like 6 seconds to respond now, and telegram is lagging slightly more as well. Trying to fix it is costing me my whole weekly token limit, lol

hollow nest
#

Discord for me takes like solid 6-8 mins until reply

maiden roost
jolly dove
#

all my hope goes to 2026.4.30 sadblob

maiden roost
#

I was eagerly waiting for 4.30, but seems worse than the previous. 4.23 , with deleting previous installations is the only stable working for the time being. It has nothing to do with the model nor the hardware .

#

Usually there are a lot of beta versions till the final stable though . So , who knows.

craggy plaza
#

im somewhat getting the same thing. I thought it was an ollama thing but could be a combination of this and ollama.

peak grail
#

4.29 totally unusable for me

craggy plaza
#

im getting 14us timing reported for CPU. My openclaw is running on a VM with 4vcores assigned. The host system CPU is a threadripper. I might have to increase the amount of CPU available to the VM.

jolly dove
maiden roost
#

Problem remains in 2026.5.2 beta 1 - tested. It does not matter how strong your CPU is. People with strong machines, also are face issues.

jolly dove
#

For me 2026.5.2 works... not out of the box, after 3 hours with chatgpt my agent responds fast on discord and works again...

maiden roost
#

@jolly dove that's good news. What do you mean 3 hours with chatgpt? Will it break after the next update? What chatgpt concluded was the issue for the delay?

jolly dove
#

tried so many things it busted my personal kontext windows... some Cron jobs were stucked... so many things.
summary from Chatgpt:
Final follow-up / probable root cause:

The system is working again now.

In my case, this did not appear to be caused by cron jobs, skills, WhatsApp, an external database, or general CPU usage alone. Cron jobs did show up in the logs and some old jobs/recovery tasks added load, but they looked more like symptoms or amplifiers, not the primary cause.

The main issue seems to have been related to the OpenClaw 2026.5.2 upgrade and plugin/config migration.

After upgrading to OpenClaw 2026.5.2, my config still contained stale plugin entries for Discord and Brave:

  • plugins.entries.discord: plugin not found
  • plugins.entries.brave: plugin not found

At the same time, the config was initially invalid because the web search provider was still set to Brave, but the Brave plugin/provider was no longer available in the same way:

  • tools.web.search.provider: web_search provider is not available: brave

The gateway repeatedly restored/used last-known-good config states while reporting those validation problems.

Discord itself was configured correctly, and the token/env vars were present, but the Discord channel was not actually starting properly until I installed the Discord plugin again as an external plugin:

  • @openclaw/discord was installed into ~/.openclaw/npm
  • after restarting, Discord was loaded again as a plugin

After that, the status probe reported:

  • Discord default: enabled, configured, running, connected, bot:@Lisa, token:config, intents:content=limited, works

Native slash commands were disabled and cleared:

  • commands.native=false
  • commands.nativeSkills=false
  • channels.discord.commands.native=false
  • channels.discord.commands.nativeSkills=false
  • log: cleared native commands (commands.native=false)

So the final working state is:

  • OpenClaw 2026.5.2
  • Discord plugin installed again
  • stale/invalid Brave provider config removed or no longer used
  • normal minimax / venice / elevenlabs disabled
  • MiniMax Coding Plan models used instead:
    • primary: minimax-coding/MiniMax-M2.7
    • fallback: minimax-coding/MiniMax-M2.5
  • native Discord slash commands disabled
  • Discord text chat working again

There were also MiniMax M2.7 timeouts through OpenClaw, but a direct curl test to the MiniMax Anthropic-compatible endpoint answered in about 3.4 seconds. So I do not think MiniMax was globally down. It looked more like OpenClaw-side request timeout / model-run handling under load or during recovery. I increased model request timeouts for the MiniMax Coding models as a precaution.

My conclusion:

The most likely cause was not a single broken cron job or Discord token issue, but a bad post-upgrade state after 2026.5.2: stale plugin references, Discord/Brave plugin packaging/config changes, and invalid web search provider configuration. Reinstalling the Discord plugin, removing/avoiding stale plugin entries, disabling unwanted providers, and letting the gateway settle brought the system back.

Important lesson:
Do not diagnose this purely by CPU usage. The useful checks were:

  • openclaw config validate
  • openclaw channels status --probe discord
  • gateway logs for plugin loading
  • Discord channel/provider startup logs
  • model timeout / fallback logs

@maiden roost

boreal iris
#

Editing title