Scripts, variables, resources disappeared from web interface
Hi,
I can no longer see any of my resources, run history, variables or scripts within my self-hosted windmill web interface. I have multiple scheduled triggers and i can confirm that the automations are still working as expected and if i go to my (external) postgres database I can still see all the scripts/resources in there along with new logs coming in. So the issue is limited to not being able to see anything in the web interface. What i've tried so far:
1. Made sure i was in the correct workspace
2. Tried logging in as superuser and other users who have access to the scripts.
3. Made sure the database connection is fine.
4. Restarted containers, updated to the latest version.
5. Checked browser console logs and docker container logs for any errors, nothing there.
The only place where I can see any history at all is if i click 'view last job' on the workers tab which does indeed show me the last job. However, if i click 'schedule' link or the 'run_id' link for the job it just gives me the error 'Could not load script: Not found: Script not found at name 6d640b38d88e8cb2 (scripts.rs:1148)'.
I have been using self-hosted windmill since May 2023 and have never had any issues until now. Would appreciate any help!
1 Reply
For anyone else who comes across this issue: This happened due to missing
BYPASSRLS privilege on the windmill_admin role. I migrated my database to a new server using pg_dump and not sure why but that permission didn't import to new database correctly. Privileges can be tested with the command SELECT rolname, rolbypassrls FROM pg_roles WHERE rolname IN ('windmill_admin', 'windmill_user'); (it should turn TRUE for both).
Weirdly, there were no errors anywhere in the logs for windmill_server or the postgres so it was difficult to diagnose. Hopefully, this will saves someone else days of debugging!