bug: Database Manager infinite spinner → 500 Internal Server Error (timeout after 600s)

[juwy]

Describe the bug

Setup

• Self-hosted Docker standalone mode (single container + PostgreSQL 16)
• Image: ghcr.io/windmill-labs/windmill:latest
• Version: CE v1.648.0-9-g4a14e9436e
• DATABASE_URL=postgres://windmill:***@db/windmill?sslmode=disable
• Workers alive: 1, database healthy: true
• Regular scripts execute fine

Browser console

Failed to load resource: the server responded with a status of 500 (Internal Server Error)
ApiError: Internal Server Error

Excerpt from logs (filtered for non-INFO)

Failed to install rustls crypto provider
WARN: oauth.json not found, no OAuth clients loaded
WARN: SMTP not configured

ERROR windmill-common/src/error.rs:267: timeout after 600s
error=ExecutionErr("timeout after 600s") method=POST
uri=/api/w/my-first-workspace/jobs/run_wait_result/preview
traceId="e300044e-b8e4-4f14-83a4-2a989888cf96"

ERROR windmill-api/src/tracing_init.rs:39: response
latency=606957 status=500 method=POST
uri=/api/w/my-first-workspace/jobs/run_wait_result/preview

ERROR windmill-api/src/tracing_init.rs:50: response failure
latency=606957 method=POST
uri=/api/w/my-first-workspace/jobs/run_wait_result/preview

ERROR windmill-queue/src/jobs.rs:767: job 019cc8c4-7412-5753-b66b-bfde7d7daafb
in my-first-workspace did not succeed:
{"error":{"message":"Job canceled: http connection broke by admin",
"name":"Canceled","reason":"http connection broke","canceler":"admin"}}

Additional observations

• pg_hba.conf: default + host all all all scram-sha-256

Possibly related

- #7451

To reproduce

Steps to reproduce

1. Create a Data Table
2. Click "Manage" → Database Manager opens
3. Infinite loading spinner appears
4. After ~10 minutes: 500 error in browser console

Expected behavior

Database Manager opens successfully

Screenshots

No response

Browser information

No response

Application version

No response

Additional Context

No response

[rubenfiszel]

try to connect from psql within a worker to that same database, you should be able to reproduce the error. Your credentials/connection method are not accepted by the db.

[juwy]

Wow, that was a fast response 😃

I tried to connect from within the Windmill worker container, but the image doesn't include psql or pg_isready, so I couldn't test directly. I connected from the DB container instead, which worked fine.

However, I'm using the built-in "Custom instance database" feature for Data Tables, so I assumed Windmill handles all credentials internally.

Quote from docs: "Windmill has a separate user called custom_instance_user with a randomly generated password, which is stored in global_settings.custom_instance_pg_databases.user_pwd."

And since Windmill is generally working (scripts execute fine, worker is alive, database is healthy), I would expect the internal database connection to work as well.

Is there any additional configuration needed for the Database Manager to work with the custom instance database?

[Badisse]

Hi ! I have the same issue. Did you manage to find a solution ?