After we migrated our Kubernetes cluster at GKE to the containerd-based node image type, the amount of logs produced by container-runtime has increased drastically (see a screenshot). Most of those log records are of info type and log some containerd's operations. The amount of produced logs is huge - some millions records per day, that creates additional costs each day.
Screenshot: Increased amount of logs after migration to containerd
Is it possible somehow to configure the logging severity level of containerd at GKE? Or what could be the way of reducing the amount of such logs?
The Kubernetes cluster is running on the latest available at GKE version (from v1.21) 1.21.11-gke.1900 atm.
I would appreciate any assistance in these regards.