Right now the API will send an error when someone tries to call stop or kill on a paused container.
This is because signals alone don't do anything to processes that have been frozen by cgroup freezer.
However once the process is unfrozen the kernel will go ahead and send those signals.
We should be able to update the API to go ahead and send the requested signal to the container, then automatically unpause it.
This is better than asking the user to unpause it first as the container may be paused due to some bad behavior in the container.
Introducing this behavior would also allow us to remove the work-around added in #13304
Right now the API will send an error when someone tries to call
stoporkillon a paused container.This is because signals alone don't do anything to processes that have been frozen by cgroup freezer.
However once the process is unfrozen the kernel will go ahead and send those signals.
We should be able to update the API to go ahead and send the requested signal to the container, then automatically unpause it.
This is better than asking the user to unpause it first as the container may be paused due to some bad behavior in the container.
Introducing this behavior would also allow us to remove the work-around added in #13304