Race condition on kube config when running in parallel

[logicbomb421]

When running this step in parallel workflows, a race condition on the config.lock file generated when changing the kube context is possible. This is very intermittent, but when it happens, one (or more) steps will error with: error: open /codefresh/volume/sensitive/.kube/config.lock: file exists.

This behavior can be reproduced by running two kubectl config use-context commands in parallel (e.g. two different terminals).

These calls are originating from the use-context command that is present in all install and promotion scripts.

Would it be possible to make this an optional param/create bypass param in order to support use cases where multiple releases are being installed into the same cluster, and the context has been preset?

Thanks!

[pampy]

I recently hit the same issue and had to go back to sequential steps.
It's funny that the documentation explicitly has an example which doesn't mention a race condition:
https://codefresh.io/docs/docs/deploy-to-kubernetes/custom-kubectl-commands/#example-of-parallel-deployment-with-kubectl

[ScottMillard]

Just experienced this also. If there is no solution will have to move back to sequential.