All-Hands-AI
[Bug]: UI gets stuck after one step of browsing
Is there an existing issue for the same bug?
- [X] I have checked the existing issues.
Describe the bug and reproduction steps
After one step of browsing, the UI gets stuck and doesn't update despite the fact that the agent was still working. Here's a prompt that I used:
Please look at this pull request and resolve the remaining issues: https://github.com/All-Hands-AI/openhands-resolver/pull/274
After I did this, the UI stopped working after the first agent reply, but after I reloaded the web page the remaining steps in the process were displayed.
OpenHands Installation
app.all-hands.dev
OpenHands Version
main
Operating System
None
Logs, Errors, Screenshots, and Additional Context
No response
I'm unable to replicate this bug. I was able to browse through our GitHub repo via OpenHands. Here is the trajectory, but it seems to be loading too much. I guess we never fixed that browser trajectory stuff?
Could you confirm the issue still persists? Maybe it was fixed from a recent merge?
The website has troubles opening that trajectory link so I can't see it. 😅 The browser trajectory stuff worked here, I think, maybe it broke?
We've made a change to how feedback is sent in #4597. We use to filter the data to remove the large browser data on the frontend, but seem to have missed that in the new change.
Hey @amanape , I was able to replicate the issue in the online version this morning, let me try again.
Here's an example (on slack because the movie was too big for github): https://openhands-ai.slack.com/archives/C07S376KASK/p1731357841319049
This sounds like the same that I saw tonight on the website. For some reason it was hanging with "Starting up" message after it started browsing. After some refreshes it suddenly displayed what it had done. No error, just, I don't know, slow?
I guess this is only a problem online, but this problem is pretty annoying… I’m also encountering it when not just browsing but also when the agent reads a large log file, etc.
