Hi on kubernetes channel I posted issues regarding slow etcd. At first I spotted messages like that:

Readiness probe failed: Error from server (InternalError): an error on the
  server ("[+]ping ok\n[+]log ok\n[-]etcd failed: reason

I started investigating the issue and I can see very high read iops ~6k. I found following process is reading heavily:

git -c credential.helper=/bin/sh -c 'echo "password=$GIT_PASSWORD"' -C /var/lib/rancher-data/local-catalogs/v2/rancher-charts/4b40cac65xxxx reset --hard FETCH_HEAD

This proces is starging by this process:

agent CATTLE_CLUSTER_AGENT_PORT_443_TCP_PORT=443 CATTLE_CLUSTER_REGISTRY=

Do you have idea why this cause so high reads? And how to mitigate the problem?