#Problem with Update

1 messages · Page 1 of 1 (latest)

sleek ingotBOT
#

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

brisk daggerBOT
#

If the web UI suddenly stops working after an “update”, 99% of the time the Gateway process isn’t running anymore (or it’s running but RPC probe/auth fails). The dashboard is just a client — it can’t work without a healthy gateway.

Can you SSH into the Hostinger VPS and paste the output of this (redact any tokens/keys if they show up):

openclaw --version
openclaw status
openclaw gateway status
openclaw logs --follow

What we’re looking for:

  • openclaw gateway status should say Runtime: running and RPC probe: ok
  • If it says stopped or probe failed, the UI won’t connect.

If it’s stopped, try:

# 
openclaw gateway start
# or (if it exists on your version)
openclaw gateway restart

Docs runbook (same “command ladder” as above):
https://docs.openclaw.ai/gateway/troubleshooting

Once you paste those outputs, I can tell you if this is a crash-loop, a port/bind issue, or just an auth/token mismatch after the update.

barren flume
#

my UI works again but openclaw dont start still.

Last time it was because of the telegram and whatsapp channel settings in the config.

brisk daggerBOT
#

If the UI loads but “OpenClaw doesn’t start”, we need to separate two cases:

  1. Gateway isn’t running / crash-looping (most common after a config change / update)
  2. Gateway is running, but channels (Telegram/WhatsApp) aren’t connected, so it looks “dead”

First: what version are you on now?

openclaw --version
openclaw gateway status
openclaw logs --follow
  • If gateway status is stopped (or RPC probe: failed), the logs will usually spell out the exact reason (often “config key unknown”, “missing allowlist”, auth mismatch, etc.).
  • Since you said last time it was Telegram/WhatsApp config: please paste the first error stacktrace you see in openclaw logs --follow right after you try:
# 
openclaw gateway start

If gateway status is running, then check channel connectivity:

openclaw channels status --probe

Also helpful (redact tokens/phone numbers):

openclaw doctor
openclaw config get channels
#

Post those outputs and I’ll tell you exactly which config bit is breaking startup (or which channel is failing and why).

barren flume
#

root@:~# openclaw --version
openclaw gateway status
openclaw logs --follow
openclaw: command not found
openclaw: command not found
openclaw: command not found

brisk daggerBOT
#

openclaw: command not found means you’re not actually running the OpenClaw CLI on that shell/user anymore (either it’s not installed, or it’s installed under a different Linux user and not in root’s PATH).

On VPS installs, the Gateway is often a systemd user service (openclaw-gateway*.service), so using root is a common gotcha. (Docs: https://docs.openclaw.ai/platforms)

Run these and paste the output:

whoami
ls /home
command -v node && node -v
command -v npm && npm -v
which -a openclaw || true

Then try the likely install user (pick one that exists under /home, e.g. openclaw, ubuntu, etc.):

# 
sudo -iu <that_user> which -a openclaw
sudo -iu <that_user> openclaw --version
sudo -iu <that_user> openclaw gateway status

If openclaw is missing for all users, it probably got uninstalled or you originally ran it via an ephemeral method and the daemon now points at a non-existent path. In that case tell me:

  • which user should own the install (the one that used to run the gateway),
  • and what node -v / npm -v are,

and I’ll give you the clean reinstall steps without wiping your ~/.openclaw state.

barren flume
#

can a person help that also knows hostinger?

brisk daggerBOT
#

Hostinger-specific knowledge usually isn’t needed here — for OpenClaw it mostly matters what kind of Hostinger plan you’re on:

  • Hostinger “Web Hosting” (shared hosting): OpenClaw won’t run there (no persistent Node daemon / no systemd).
  • Hostinger VPS (KVM): totally fine — it’s just a normal Linux box, and we can debug it.

So first question: are you on Hostinger VPS and do you have SSH access?

If yes, the key clue from your last paste was openclaw: command not found. That means either:

  1. OpenClaw isn’t installed anymore, or
  2. it’s installed under a different user (common: the gateway runs as a non-root user, and you’re logged in as root).

Please paste the output of:

whoami
ls /home
# 
which -a openclaw || true

Then try (replace <user> with one of the users you see under /home, often openclaw/ubuntu):

sudo -iu <user> which -a openclaw
sudo -iu <user> openclaw --version

If you tell me whether it’s VPS vs shared hosting + those outputs, I can give you the exact next steps.

barren flume
#

its on VPS KVM and Its in a docker, thats why the direct command doesnt work

brisk daggerBOT
#

Yep — if you’re running OpenClaw in Docker, then openclaw ... on the VPS host will often be “command not found” (that’s expected). You need to run the CLI inside the container (or via the openclaw-cli helper container if you’re using the official compose setup).

1) First: are the containers even up?

From the folder that has your docker-compose.yml:

docker compose ps

2) Check why the gateway isn’t starting (this is the money command)

docker compose logs --tail 200 openclaw-gateway

If it’s crash-looping because of Telegram/WhatsApp config, the error will be in there.

3) Run “openclaw commands” the Docker way

#

These mirror the normal troubleshooting ladder, but via the CLI container:

docker compose run --rm openclaw-cli --version
docker compose run --rm openclaw-cli gateway status
docker compose run --rm openclaw-cli doctor
docker compose run --rm openclaw-cli channels status --probe

If you don’t have the openclaw-cli service in your compose

Then do it manually:

docker ps
docker logs --tail 200 <gateway_container_name>
docker exec -it <gateway_container_name> openclaw --version