I have a pod which has memory leak problems. As we can see in the picture, there are a lot of buff/cach memory.

We used to use docker swarm as the cluster orchestration and we can directly access to the cluster node. And we write a crontab on the cluster node to clear the buff/cache every day in the morning.

15 00 * * * /usr/sbin/sysctl -w vm.drop_caches=3 >> /tmp/clean_cache.log 2>&1

However, we can’t access to the cluster node directly in k8s, can we clear the buff/cache still?