Worker Outage - Latest Version
TLDR : pulled the latest version and installed using Caprover. Works outage and logs saying Please compile your catalog first. Any ideas?
Asked GPT (Sorry Im not a dev) and this is what it says:
twentycrm/twenty:latest image includes dev source instead of production build (worker & i18n issues)
Inside the running container we can see:
/app/packages/twenty-server/src
/app/packages/twenty-server/test
jest.config.mjs
tsconfig.json
lingui.config.ts
This indicates the image packaged the raw TypeScript source, not the compiled dist/ output only.
Because of this:
• The server and worker processes do not initialize correctly.
• Worker queue never starts (no worker binary in the expected state).
• Lingui throws “Please compile your catalog first”.
• Runtime loads uncompiled catalogs instead of prebuilt ones.
• NODE_ENV=production does not fix the issue because the image content is not a production artifact.
Adding production envs like:
NODE_ENV=production
LINGUI_COMPILE_CATALOGS=true
does not solve it, since the problem is that the image itself contains development code, not a production build.
A proper production image should contain only:
/app/packages/twenty-server/dist
/app/node_modules
/app/package.json
and should not include /src, tests, tsconfigs, Lingui config files, etc.
Could you confirm whether the published image was built from the production Dockerfile, or if the build pipeline accidentally pushed a development artifact?
A corrected production build would also solve:
• Worker not running
• i18n runtime errors
• Slow startup
• Schema regeneration
Thanks in advance — happy to assist with logs or testing other images.
Asked GPT (Sorry Im not a dev) and this is what it says:
twentycrm/twenty:latest image includes dev source instead of production build (worker & i18n issues)
Inside the running container we can see:
/app/packages/twenty-server/src
/app/packages/twenty-server/test
jest.config.mjs
tsconfig.json
lingui.config.ts
This indicates the image packaged the raw TypeScript source, not the compiled dist/ output only.
Because of this:
• The server and worker processes do not initialize correctly.
• Worker queue never starts (no worker binary in the expected state).
• Lingui throws “Please compile your catalog first”.
• Runtime loads uncompiled catalogs instead of prebuilt ones.
• NODE_ENV=production does not fix the issue because the image content is not a production artifact.
Adding production envs like:
NODE_ENV=production
LINGUI_COMPILE_CATALOGS=true
does not solve it, since the problem is that the image itself contains development code, not a production build.
A proper production image should contain only:
/app/packages/twenty-server/dist
/app/node_modules
/app/package.json
and should not include /src, tests, tsconfigs, Lingui config files, etc.
Could you confirm whether the published image was built from the production Dockerfile, or if the build pipeline accidentally pushed a development artifact?
A corrected production build would also solve:
• Worker not running
• i18n runtime errors
• Slow startup
• Schema regeneration
Thanks in advance — happy to assist with logs or testing other images.
