You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I run my gh workflows in a self-hosted runner that runs inside a container. The container is torn down at the end of the action, and a service immediately spawns another one, so that there's always a runner listening. This is for security reasons, so that the code tested remains contained and does not pose (much of) a security risk (there are additional guardrails, but that's not the point here).
Given that the container where the runner lives will be torn down right after the action completes, it is somewhat of a waste to perform the post checkout cleanup. In some cases, this can take >2min, which is somewhat annoying, given that it is ~10% of the whole action runtime.
Our development can be fast, with ppl sometimes pushing commits to the PR while testing is still ongoing (e.g., one CI job fails, so they immediately push a fix, which causes all jobs to be canceled due to concurrency settings).
Do you think it's possible to make the cleanup step optional? Of course, the default behavior should be to perform it...
The text was updated successfully, but these errors were encountered:
I run my gh workflows in a self-hosted runner that runs inside a container. The container is torn down at the end of the action, and a service immediately spawns another one, so that there's always a runner listening. This is for security reasons, so that the code tested remains contained and does not pose (much of) a security risk (there are additional guardrails, but that's not the point here).
Given that the container where the runner lives will be torn down right after the action completes, it is somewhat of a waste to perform the post checkout cleanup. In some cases, this can take >2min, which is somewhat annoying, given that it is ~10% of the whole action runtime.
Our development can be fast, with ppl sometimes pushing commits to the PR while testing is still ongoing (e.g., one CI job fails, so they immediately push a fix, which causes all jobs to be canceled due to concurrency settings).
Do you think it's possible to make the cleanup step optional? Of course, the default behavior should be to perform it...
The text was updated successfully, but these errors were encountered: