We make use of a mutex to guard access to some internal maps used to map container cgroups and policy settings.
We have to do a quick research to understand:
- Can our NRI plugin be called concurrently by the container runtime? If X pods are started at the same time, do we get X concurrent requests?
- Can we use a different approach for guarding these internal structures? Maybe using a RW mutex or something else?
We make use of a mutex to guard access to some internal maps used to map container cgroups and policy settings.
We have to do a quick research to understand: