Bolt.diy hanging after prompt

[shawn]

New installation on a M1 Macbook Pro. Appears to be complete and should run but starts then hangs.

Using Ollama and Qwen2.5-coder:32b

After executing the following at the terminal - pnpm run dev
Getting a couple of warnings…

▲ [WARNING] Duplicate key “dev” in object literal [duplicate-object-key]

package.json:13:4:
  13 │     "dev": "remix vite:dev --host 0.0.0.0 --port 3000 --open",
     ╵     ~~~~~

The original key “dev” is here:

package.json:12:4:
  12 │     "dev": "node pre-start.cjs  && remix vite:dev",
     ╵     ~~~~~

and…

WARN Constants Failed to get LMStudio models: fetch failed
WARN Constants Failed to get LMStudio models: fetch failed
WARN Constants Failed to get LMStudio models: fetch failed

Also Ollama is local so don’t think the following should be displayed on the screen…?

Ollama API Key:
Not set (will still work if set in .env file)

Also I just realised when following an installation guide I failed to - Increase the Ollama Context Window

Is this a deal breaker…?

Initially it started working and built an index page then moved on to the next then hung.

I have been using Bolt.new for the last couple of weeks and churned though millions of tokens unnecessarily for silly errors that Bolt introduced then had to undo but it would make the same mistakes again and again costing a lot of my monthly balance.

Hoping to get this local copy of Bolt.diy up and running quickly and very excited to share some of my recent works with the community also.

Thanks so much to @ColeMedin for your incredible foresight and investment into this project so far.

[leex279]

hi @shawn,

can you paste the debug output so we get a bit more details (Settings => Debug Tab, enabled in Features first if not done).

Also paste a screenshot or the output of the terminal/shell you running bolt.

[shawn]

Thanks. I’ve switched it on and here’s the debug details…

{
“System”: {
“os”: “MacIntel”,
“browser”: “Safari/537.36”,
“screen”: “3440x1440”,
“language”: “en-AU”,
“timezone”: “Australia/Melbourne”,
“memory”: “2.05 GB”,
“cores”: 8
},
“Providers”: [
{
“name”: “Ollama”,
“enabled”: true,
“isLocal”: true,
“running”: false,
“error”: “No URL configured”,
“lastChecked”: “2024-12-18T08:32:56.538Z”,
“url”: null
},
{
“name”: “OpenAILike”,
“enabled”: true,
“isLocal”: true,
“running”: false,
“error”: “No URL configured”,
“lastChecked”: “2024-12-18T08:32:56.538Z”,
“url”: null
},
{
“name”: “LMStudio”,
“enabled”: true,
“isLocal”: true,
“running”: false,
“error”: “No URL configured”,
“lastChecked”: “2024-12-18T08:32:56.538Z”,
“url”: null
}
],
“Version”: “25fe15232fcd6cee83f179adbd1d3e7d6a90acca”,
“Timestamp”: “2024-12-18T08:33:25.173Z”
}

[shawn]

[shawn]

The guide I used to install is below which is for Windows but I’ve done my best to make changes for Macos…

OTTODEV + QWEN : FREE OPEN-SOURCE BOLT

[shawn]

I did update package.json to remove the original dev line so only one now as beloe…
“dev”: “remix vite:dev --host 0.0.0.0 --port 3000 --open”,

[leex279]

Thanks, were you got this instructions (google sheet) from? I dont there is a need for changing the dev line and you should stick to the provided docs from this project:
https://stackblitz-labs.github.io/bolt.diy/#setup

So dont change the package.json. It should work fine.

[shawn]

Thank you.

I have gone back and restarted as per guide you provided but still getting the following debug detail…
{
“System”: {
“os”: “macOS”,
“browser”: “Chrome 131.0.0.0”,
“screen”: “3440x1440”,
“language”: “en-AU”,
“timezone”: “Australia/Melbourne”,
“memory”: “2.05 GB (Used: 81.35 MB)”,
“cores”: 8,
“deviceType”: “Desktop”,
“colorDepth”: “24-bit”,
“pixelRatio”: 1,
“online”: true,
“cookiesEnabled”: true,
“doNotTrack”: false
},
“Providers”: [
{
“name”: “Ollama”,
“enabled”: true,
“isLocal”: true,
“running”: false,
“error”: “No URL configured”,
“lastChecked”: “2024-12-18T10:41:50.765Z”,
“url”: null
},
{
“name”: “OpenAILike”,
“enabled”: true,
“isLocal”: true,
“running”: false,
“error”: “No URL configured”,
“lastChecked”: “2024-12-18T10:41:50.765Z”,
“url”: null
},
{
“name”: “LMStudio”,
“enabled”: true,
“isLocal”: true,
“running”: false,
“error”: “No URL configured”,
“lastChecked”: “2024-12-18T10:41:50.765Z”,
“url”: null
}
],
“Version”: {
“hash”: “e064803”,
“branch”: “main”
},
“Timestamp”: “2024-12-18T10:42:09.007Z”
}

Please help.

[leex279]

Did you also set the context values. See in .env.example. You have to set according to your available VRAM.

Also => does Ollama work directly in the termimal with this model and provides output?

# Example Context Values for qwen2.5-coder:32b
# 
# DEFAULT_NUM_CTX=32768 # Consumes 36GB of VRAM
# DEFAULT_NUM_CTX=24576 # Consumes 32GB of VRAM
# DEFAULT_NUM_CTX=12288 # Consumes 26GB of VRAM
# DEFAULT_NUM_CTX=6144 # Consumes 24GB of VRAM
DEFAULT_NUM_CTX=

[shawn]

The guide does not suggest to do this that I can see and how do I do this for Mac please?

I set this to…
DEFAULT_NUM_CTX=6144

Restarted server and still same issue.

And yes Ollama is running…

[leex279]

Ok, but your Ollama is not running, at least that cant be seen in this screenshot:

it says error. Can you please try to use ollama directly in your terminal and send a prompt there. If this is working, you can go on with bolt.

See here for an example:

Ollama CLI tutorial: Running Ollama via the terminal

you should get the output of the prompt directly in the terminal

[shawn]

OK. I had Ollama running as an App and showing as running. I closed that and ran from the terminal as below…

Then let that run. Opened another terminal and Ollama seemed to take a little while to initialise. Then i entered this…

So yes it’s working here now. Just not in the GUI panel…

Still getting this in Debug.

[aliasfox]

Reading the tutorial…

Just an FYI, you can set the repo name (without rename later) with git clone {url} {name}. This is useful if you need a copy of a repo but git won’t let you by the same name (like stupidly, i sometime clone a repo twice, once with {repo}-clean for faster searching without all the node junk, lol).

The note is wrong, you don’t need to do anything, this is all set by remix. DON’T DO STEP 5. And I didn’t think you need to do any of that Ollama stuff, because I think that’s the default. But if you want to uncomment the flag in the .env for that, fine. But that is all I’d do.

Just follow the instructions on the repo.

Basically, not using docker:
git clone https://github.com/stackblitz-labs/bolt.diy
cd bolt.diy
npn install pnpm
pnpm install
pnpm run dev

Note: Make sure to copy .env.example to .env, add all your keys and/or baseUrl paths (but I believe this assumes their defaults because both Ollama and LMStudio worked for me).

That’s it, install Ollama and/or LMSudio with defaults and Bolt.diy should just pick them up.

Please take the lesson here not to just take any random instructions as gospel.

So maybe just try a clean install and see if it works.

[shawn]

The guide I followed was for OTTODEV and had a few changes. I get it but even the other guide is not complete and I have questions…

Should I load a URL for Ollama as it is running locally - how did you do this please?

Also the guide is different from your suggestion as it says…

Rename .env.example to .env.local

Is that correct or rename to .env please?

Also you have to start Ollama server and have running before starting the Dev server in another terminal with - pnpm run dev ???

Thanks for your patience as the guide isn’t very good covering all this.

[aliasfox]

I answered that. It’s default, the fallback is already set in the code. Just follow the instructions… Thinking that there is something missing does not mean it is. I believe you are over-troubleshooting a problem that does not exist. We’ve all been there, and it doesn’t matter what version you are setting up because they should all be the same, this has not changed.

My Ollama starts by default, I did not set this. But you can use ollama serve I believe. But if it’s in your system tray, or can visit the url endpoint, it’s already running.

And .env.local is fine, I think it looks for both.

[shawn]

Thank you.

I will remove the URL from the providers.

Problem I see is the Local LLM Status still shows ‘Not Running’ even after I had it running and entered prompt as you requested.

I will try clean install again. There is something not right even when I follow this guide exactly it still doesn’t work.

[aliasfox]

Also, another little note. The base URL should NOT end in “/”. I did that once on something and it resulted in a console error ...//v1/messages or something. Basically, it causes two slashes.

If you run into issues after a clean setup and making sure ollama is running (just browse to the URL/Port it uses, it should say Ollama is running or something), provide your console log errors (screenshot is fine).

And what operating system are you on? I’m assuming Windows (version…)?

[shawn]

Hi again.

I think the issue was I don’t have enough grunt on my machine to run this so I’ve swapped to a windows gamer machine and starting from scratch. I am installing in D drive and getting an issue as the path doesn’t have usr/local/bin

The guide doesn’t mention anything about managing this just says you should be good to go. I can update path through normal means but since I’m installing on D drive, I assume dependencies will be different too…

Installing at…
PS D:\coding\git\bolt.diy>
Mode LastWriteTime Length Name

---

d----- 19/12/2024 7:04 PM .github
d----- 19/12/2024 7:04 PM .husky
d----- 19/12/2024 7:04 PM app
d----- 19/12/2024 7:04 PM docs
d----- 19/12/2024 7:04 PM functions
d----- 19/12/2024 7:04 PM icons
d----- 20/12/2024 8:11 PM node_modules
d----- 19/12/2024 7:04 PM public
d----- 19/12/2024 7:04 PM types
-a---- 19/12/2024 7:04 PM 385 .dockerignore
-a---- 19/12/2024 7:04 PM 223 .editorconfig
-a---- 19/12/2024 7:04 PM 3181 .env.example
-a---- 19/12/2024 7:04 PM 439 .gitignore
-a---- 19/12/2024 7:04 PM 24 .prettierignore
-a---- 19/12/2024 7:04 PM 134 .prettierrc
-a---- 19/12/2024 7:04 PM 28 .tool-versions
-a---- 19/12/2024 7:04 PM 416 bindings.sh
-a---- 19/12/2024 7:04 PM 729 changelog.md
-a---- 19/12/2024 7:04 PM 6289 CONTRIBUTING.md
-a---- 19/12/2024 7:04 PM 2311 docker-compose.yaml
-a---- 19/12/2024 7:04 PM 2458 Dockerfile
-a---- 19/12/2024 7:04 PM 1776 eslint.config.mjs
-a---- 19/12/2024 7:04 PM 6111 FAQ.md
-a---- 19/12/2024 7:04 PM 1094 LICENSE
-a---- 19/12/2024 7:04 PM 217 load-context.ts
-a---- 20/12/2024 8:11 PM 892859 package-lock.json
-a---- 20/12/2024 7:23 PM 5167 package.json
-a---- 19/12/2024 7:04 PM 411854 pnpm-lock.yaml
-a---- 19/12/2024 7:04 PM 576 pre-start.cjs
-a---- 19/12/2024 7:04 PM 8827 README.md
-a---- 19/12/2024 7:04 PM 879 tsconfig.json
-a---- 19/12/2024 7:04 PM 9181 uno.config.ts
-a---- 19/12/2024 7:04 PM 2254 vite.config.ts
-a---- 19/12/2024 7:04 PM 517 worker-configuration.d.ts
-a---- 19/12/2024 7:04 PM 207 wrangler.toml

And path environment displays as…
PS D:\coding\git\bolt.diy> $env:PATH
C:\Python312\Scripts;C:\Python312;C:\Windows\system32;C:\Windows;C:\Windows\System32\Wbem;C:\Windows\System32\WindowsPowerShell\v1.0;C:\Windows\System32\OpenSSH;C:\Program Files (x86)\NVIDIA Corporation\PhysX\Common;D:\Program Files\Node;C:\ProgramData\chocolatey\bin;C:\Users\Admin\AppData\Local\pnpm;C:\Users\Admin\AppData\Local\Microsoft\WindowsApps;C:\Users\Admin\AppData\Local\GitHubDesktop\bin;C:\Users\Admin\AppData\Roaming\npm

Is this the additional entry I need to put in my path please…
D:\coding\git\bolt.diy

Thanks for your help.

[leex279]

Can you post the screenshot / exact error you get please?
It find it a bit confusing that you get “usr/local/bin” because this should be only on unix based systems, doesnt it? windows does not have these folder structure

[shawn]

You wanted me to use this guide…
https://stackblitz-labs.github.io/bolt.diy/#setup

I’m actually having issues with NPM and dependency issues but let me do some more investigation first.