GCP COS VM docker logs rotation

I have deployed GCP VM with cos-stable-121-18867-90-97 as boot disk. It is running systemd service on boot that is executing script. Script is executing docker image like this:

#!/bin/bash

# Script to manage task runner container
# Actions: start, stop, cleanup

set -e -o pipefail

get_access_token () {
  local endpoint=http://metadata.google.internal/computeMetadata/v1/instance/service-accounts/default/token
  curl -s -H 'Metadata-Flavor: Google' "$endpoint" | cut -d'"' -f 4
}

container_name="task-runner"

case "$1" in
  start)
    echo "Logging into registry..."

    docker login -u oauth2accesstoken -p "$(get_access_token)" europe-docker.pkg.dev

    echo "Starting task runner container..."
    # Run the container with the specified configuration
    docker run \
      --name "$container_name" \
      -v "/etc/secrets/dotenv_file:/secrets/.env:ro" \
      -v "/etc/app/cloudsql_client_ssl_identity_pkcs12:/cloudsql-client-certs/client_identity.p12:ro" \
      -v "/etc/app/cloudsql_client_ssl_ca:/cloudsql-client-certs/client_ca.cert:ro" \
      --env-file "/etc/app/container_env_file" \
      "europe-docker.pkg.dev/xx/xx/app:x.y.z" \
      /usr/local/bin/supercronic -json "/etc/production.crontab"
    ;;

  stop)
    echo "Stopping task runner container..."
    # Stop the container if it's running
    docker stop "$container_name" 2>/dev/null || true
    ;;

  cleanup)
    echo "Cleaning up task runner container..."
    # Remove the container
    docker rm -f "$container_name" 2>/dev/null || true
    ;;

  *)
    echo "Usage: $0 {start|stop|cleanup}"
    exit 1
    ;;
esac

exit 0

All is running fine for around two weeks until VM disk gets full of docker logs from the running container. I've tried to set /etc/docker/daemon.json like this

{
  "live-restore": true,
  "log-opts": {
    "tag": "{{.Name}}",
    "max-size": "100m",
    "max-file": "3"
  },
  "storage-driver": "overlay2",
  "mtu": 1460
}

but it still keeps happening. It is like docker is not using this configuration at all.

Now I'm testing adding --log-opt max-size=10m and --log-opt max-file=5 to docker run command in the script. But will take some time to see if it works or not.

Any idea why docker configuration was not loaded or maybe was wrong from the beginning? Is there a nice way to limit how much logs is stored on disk by docker with COS boot disk?

//EDIT

Passing --log-opt to docker run seems to work, but why not daemon.json?