Can't run development environment on Windows

[webwarrior-ws]

Is there an existing issue for this?

• I have searched the existing issues

Describe the bug

Running ./sndev start fails with an error. db image is created, but app and worker are not.

Screenshots

No response

Steps To Reproduce

Run ./sndev start in bash console on Windows with Docker Desktop running.

Expected behavior

Docker images are created and the app is running.

Logs

npm verbose cli /usr/local/bin/node /usr/local/bin/npm
npm info using npm@10.7.0
npm info using node@v18.20.4
npm verbose title npm install
npm verbose argv "install" "--loglevel" "verbose" "--legacy-peer-deps"
npm verbose logfile logs-max:10 dir:/home/apprunner/.npm/_logs/2025-08-12T11_26_30_758Z-
npm verbose logfile /home/apprunner/.npm/_logs/2025-08-12T11_26_30_758Z-debug-0.log
\|/-\|/-\|/-\|/-\|/-\npm verbose stack Error: ENOTEMPTY: directory not empty, rename '/app/node_modules/aws-sdk' -> '/app/node_modules/.aws-sdk-P9TfRRB5'
\npm verbose cwd /app
\npm verbose Linux 5.15.153.1-microsoft-standard-WSL2
\npm verbose node v18.20.4
\npm verbose npm  v10.7.0
\npm error code ENOTEMPTY
\npm error syscall rename
\npm error path /app/node_modules/aws-sdk
\npm error dest /app/node_modules/.aws-sdk-P9TfRRB5
\npm error errno -39
\npm error ENOTEMPTY: directory not empty, rename '/app/node_modules/aws-sdk' -> '/app/node_modules/.aws-sdk-P9TfRRB5'
\npm verbose exit -39
\npm verbose code -39
\
\npm error A complete log of this run can be found in: /home/apprunner/.npm/_logs/2025-08-12T11_26_30_758Z-debug-0.log

Device information

• OS: Windows 10

Additional context

No response

[ekzyis]

Mhh, not sure why it would want to rename node_modules/aws-sdk:

\npm error code ENOTEMPTY
\npm error syscall rename
\npm error path /app/node_modules/aws-sdk
\npm error dest /app/node_modules/.aws-sdk-P9TfRRB5
\npm error errno -39
\npm error ENOTEMPTY: directory not empty, rename '/app/node_modules/aws-sdk' -> '/app/node_modules/.aws-sdk-P9TfRRB5'

Can you try rm -rf node_modules and then run sndev start again?

[webwarrior-ws]

I managed to get rid of ENOTEMPTY: directory not empty.

But the app is still not running and is stuck in kind of infinite boot loop.
And when I press Enter in bash console, I get:

Gracefully stopping... (press Ctrl+C again to force)
dependency failed to start: container app is unhealthy

There is nothing that would indicate an error in app container logs. In db logs, there is this:

2025-08-13 12:31:21.718 UTC [354] LOG:  could not receive data from client: Connection reset by peer
2025-08-13 12:31:28.664 UTC [356] WARNING:  there is no transaction in progress
2025-08-13 12:31:28.753 UTC [356] FATAL:  terminating connection due to administrator command
2025-08-13 12:31:28.858 UTC [27] LOG:  checkpoint starting: immediate force wait
2025-08-13 12:31:28.943 UTC [27] LOG:  checkpoint complete: wrote 5 buffers (0.0%); 0 WAL file(s) added, 0 removed, 1 recycled; write=0.030 s, sync=0.014 s, total=0.085 s; sync files=4, longest=0.007 s, average=0.004 s; distance=18473 kB, estimate=18577 kB; lsn=0/B2861F50, redo lsn=0/B2861F18

[ekzyis]

Can you show the full app container logs during startup?

[webwarrior-ws]

Can you show the full app container logs startup?

https://gist.github.com/webwarrior-ws/c6552b25dc2cd5431c71c8bd6241a3ec

[ekzyis]

Mhh, I only see this but it shouldn't be failing because of this because we wrap the git call in next.config.js with try/catch

2025-08-12 14:48:43.860 | fatal: detected dubious ownership in repository at '/app'
2025-08-12 14:48:43.861 | To add an exception for this directory, call:
2025-08-12 14:48:43.861 |
2025-08-12 14:48:43.861 | git config --global --add safe.directory /app
2025-08-12 14:48:43.861 | could not get commit hash with git rev-parse HEAD ... using 0000

It just restarts after it's stuck at "Starting..." for 5 seconds

2025-08-12 14:48:43.894 | ✓ Starting...
2025-08-12 14:48:48.358 | \2025-08-12T12:48:49.111071949Z npm verbose cli /usr/local/bin/node /usr/local/bin/npm

Can you also show me the full db container logs during startup?

[webwarrior-ws]

db container logs: https://gist.github.com/webwarrior-ws/0aeaac1547e76bfc9052519e8579bb2f

[knocte]

I managed to get rid of ENOTEMPTY: directory not empty.

How?

[webwarrior-ws]

I managed to get rid of ENOTEMPTY: directory not empty.

How?

I'm not 100% sure. I tried to delete node_modules using Docker Desktop, and got errors. But maybe one of the attempts succeeded. I had to do it while the container is runnning.

[ekzyis]

db container logs: https://gist.github.com/webwarrior-ws/0aeaac1547e76bfc9052519e8579bb2f

I'm sorry, I forgot to reply!

But mhh, the db logs don't show much. And I can see in the app logs that the migration runs successfully. So I think it's not a db problem. The prisma client is also generated.

Did you try deleting the .next folder?