Error while running backstage application in Kubernetes

[ravargs]

I deployed the backstage application through the helm charts and the template creation was failing with with a docker not found error. This was fixed through docker in docker but now it fails with this error .

2021-03-18T13:05:10.764Z {"status":"Extracting","progressDetail":{"current":23264053,"total":23264053},"progress":"[==================================================\u003e] 23.26 MB/23.26 MB","id":"681cae0a8229"}
2021-03-18T13:05:10.798Z {"status":"Pull complete","progressDetail":{},"id":"681cae0a8229"}
2021-03-18T13:05:10.947Z {"status":"Digest: sha256:0e2891bebc903eeedf7293cf75e9faad7539874e9d4e5e37da91760e2c8fab60"}
2021-03-18T13:05:10.990Z {"status":"Status: Downloaded newer image for spotify/backstage-cookiecutter:latest"}
2021-03-18T13:05:12.975Z DEBUG cookiecutter.config: User config not found. Loading default config.
2021-03-18T13:05:12.976Z A valid repository for "/input" could not be found in the following locations:
/input
/input
2021-03-18T13:05:13.554Z Error: Docker container returned a non-zero exit code (1)
at Object.runDockerContainer (/app/node_modules/@backstage/backend-common/dist/index.cjs.js:1811:11)
at runMicrotasks ()
at processTicksAndRejections (internal/process/task_queues.js:93:5)
at async CookieCutter.run (/app/node_modules/@backstage/plugin-scaffolder-backend/dist/index.cjs.js:706:7)
at async Object.handler (/app/node_modules/@backstage/plugin-scaffolder-backend/dist/index.cjs.js:1626:9)
at async TaskWorker.runOneTask (/app/node_modules/@backstage/plugin-scaffolder-backend/dist/index.cjs.js:2180:11)
at async /app/node_modules/@backstage/plugin-scaffolder-backend/dist/index.cjs.js:2121:9

Can you please confirm if
Backstage application is fully supported in Kubernetes platform? If so Docker in Docker setup is necessary?

[Rugvip]

Hi! See #2842, I believe this is the issue you're facing. /template was renamed to /input not long ago

[ravargs]

@Rugvip - Its looks like similar issue but I don't see this occurring when run from local development box.
Where should I change the /input directory?

[Rugvip]

@ravargs sorry for the confusion, was just highlighting that the /template directory referred to in the logs in #2842 has been renamed to /input, meaning the logs are comparable. You do not need to rename any directory.

You'll want to try out #2842 (comment) as see if that fixes the issue for you

[niallmccullagh]

Just wanted to tag onto this issue and note that comment in #2842 about sharing /tmp worked for me. I created a shared volume and mounted it to /tmp on both the backend and docker dind sidecar.

[ravargs]

@niallmccullagh - Thanks, the same steps worked for me as well.

[ravargs]

Issue is fixed by creating /tmp shared volume.