Hi team, how do I solve the problem of the entire...
# questions
p
Hi team, how do I solve the problem of the entire history scrolling as soon as I open a dialogue window with a bot? The history scrolls for quite a long time. Is there an option where it automatically goes to the last dialogue without scrolling, which takes about 4 minutes?
➕ 1
👀 1
c
Does this happen if you use agent-canvas? https://github.com/OpenHands/agent-canvas#quickstart
You can access from it the Cloud too, if you were using the Cloud
The Canvas is the new OpenHands interface, we deprecated the old one; and it has pagination so it should load just the latest naturally
e
This is good feedback - we should optimize this in agent canvas
p
Yes I think I use last model canvas, but LLM is Gpt or DeepSeek Where I can double check model. Then error aper very often- Backend unable And can't use it from my iPhone UI is frozen nothing is working on mobile
w
Karina--how did you install OpenHands?
p
Mmmm …. It is in my server now Railway Or what do you mean
w
like what command did you run to install it?
npm install @openhands/agent-cavas
?
p
Yes it is Agent Canvas v 1.2.1 Docker ghcr .io/openhands/agent-canvas:1.2.1 OpenHands Agent Canvas (v1.2.1) deployed on Railway is very slow. Despite configuring the OpenAI API key in the UI, server logs consistently show has_llm_api_key=false. The agent remains in a 'Connecting' or 'Starting' state and fails to respond to messages, occasionally throwing ACP prompt failed: Internal error." Details: Platform: Railway (Docker deployment) Runtime: SANDBOX_RUNTIME=local Issue: LLM settings/API key not persisting or not being recognized by the backend.
w
This is probably an issue with your Railway setup
what port did you expose? I'm trying to repro
working just fine for me...
p
Port 8000 . Railway port environment variable is set to 8000. Very slow and when I send messages during the work of agent he continues to work on hes task and ignore my massages
w
it wouldn't respond on 8000 for me, I had to do 8080 for some reason
p
Hm same result in 8080 I guess we need one more bot 🤖 that can help us with installation and technical questions 😇