Merge pull request #816 from bogdando/paths_units

Systemd units, limits, and bin path fixes

docs/large-deployments.md

@@ -20,5 +20,12 @@ For a large scaled deployments, consider the following configuration changes:
   ``dns_cpu_requests``, ``dns_memory_limit``, ``dns_memory_requests``.
   Please note that limits must always be greater than or equal to requests.
 
+* Tune CPU/memory limits and requests. Those are located in roles' defaults
+  and named like ``foo_memory_limit``, ``foo_memory_requests`` and
+  ``foo_cpu_limit``, ``foo_cpu_requests``. Note that 'Mi' memory units for K8s
+  will be submitted as 'M', if applied for ``docker run``, and cpu K8s units will
+  end up with the 'm' skipped for docker as well. This is required as docker does not
+  understand k8s units well.
+
 For example, when deploying 200 nodes, you may want to run ansible with
 ``--forks=50``, ``--timeout=600`` and define the ``retry_stagger: 60``.

inventory/group_vars/all.yml

@@ -188,6 +188,7 @@ docker_daemon_graph: "/var/lib/docker"
 ## An obvious use case is allowing insecure-registry access
 ## to self hosted registries like so:
 docker_options: "--insecure-registry={{ kube_service_addresses }} --graph={{ docker_daemon_graph }}"
+docker_bin_dir: "/usr/bin"
 
 ## Uncomment this if you want to force overlay/overlay2 as docker storage driver
 ## Please note that overlay2 is only supported on newer kernels

roles/docker/handlers/main.yml

@@ -27,7 +27,7 @@
   pause: seconds=10 prompt="Waiting for docker restart"
 
 - name: Docker | wait for docker
-  command: /usr/bin/docker images
+  command: "{{ docker_bin_dir }}/docker images"
   register: docker_ready
   retries: 10
   delay: 5

roles/docker/templates/docker.service.j2

@@ -18,7 +18,7 @@ Environment=GOTRACEBACK=crash
 ExecReload=/bin/kill -s HUP $MAINPID
 Delegate=yes
 KillMode=process
-ExecStart=/usr/bin/docker daemon \
+ExecStart={{ docker_bin_dir }}/docker daemon \
           $DOCKER_OPTS \
           $DOCKER_STORAGE_OPTIONS \
           $DOCKER_NETWORK_OPTIONS \

roles/download/tasks/main.yml

@@ -79,7 +79,7 @@
 
 #NOTE(bogdando) this brings no docker-py deps for nodes
 - name: Download containers if pull is required or told to always pull
-  command: "/usr/bin/docker pull {{ pull_args }}"
+  command: "{{ docker_bin_dir }}/docker pull {{ pull_args }}"
   register: pull_task_result
   until: pull_task_result|success
   retries: 4
@@ -115,7 +115,7 @@
   tags: facts
 
 - name: Download | save container images
-  shell: docker save "{{ pull_args }}" | gzip -{{ download_compress }} > "{{ fname }}"
+  shell: "{{ docker_bin_dir }}/docker save {{ pull_args }} | gzip -{{ download_compress }} > {{ fname }}"
   delegate_to: "{{ download_delegate }}"
   register: saved
   run_once: true
@@ -145,6 +145,6 @@
   tags: [upload, upgrade]
 
 - name: Download | load container images
-  shell: docker load < "{{ fname }}"
+  shell: "{{ docker_bin_dir }}/docker load < {{ fname }}"
   when: (ansible_os_family != "CoreOS" and inventory_hostname != groups['kube-master'][0] or download_delegate == "localhost") and download_run_once|bool and download.enabled|bool and download.container|bool
   tags: [upload, upgrade]

roles/download/tasks/set_docker_image_facts.yml

@@ -8,7 +8,7 @@
       {%- if pull_by_digest|bool %}{{download.repo}}@sha256:{{download.sha256}}{%- else -%}{{download.repo}}:{{download.tag}}{%- endif -%}
 
 - name: Register docker images info
-  shell: "{% raw %}/usr/bin/docker images -q | xargs /usr/bin/docker inspect -f '{{.RepoTags}},{{.RepoDigests}}'{% endraw %}"
+  shell: "{{ docker_bin_dir }}/docker images -q | xargs {{ docker_bin_dir }}/docker inspect -f {% raw %}'{{.RepoTags}},{{.RepoDigests}}'{% endraw %}"
   register: docker_images_raw
   failed_when: false
   when: not download_always_pull|bool

roles/etcd/defaults/main.yml

@@ -6,3 +6,7 @@ etcd_cert_dir: "{{ etcd_config_dir }}/ssl"
 etcd_cert_group: root
 
 etcd_script_dir: "{{ bin_dir }}/etcd-scripts"
+
+# Limits
+etcd_memory_limit: 512M
+etcd_cpu_limit: 300m

roles/etcd/tasks/install.yml

@@ -12,10 +12,10 @@
 
 #Plan A: no docker-py deps
 - name: Install | Copy etcdctl binary from container
-  command: sh -c "/usr/bin/docker rm -f etcdctl-binarycopy;
+  command: sh -c "{{ docker_bin_dir }}/docker rm -f etcdctl-binarycopy;
-           /usr/bin/docker create --name etcdctl-binarycopy {{ etcd_image_repo }}:{{ etcd_image_tag }} &&
+           {{ docker_bin_dir }}/docker create --name etcdctl-binarycopy {{ etcd_image_repo }}:{{ etcd_image_tag }} &&
-           /usr/bin/docker cp etcdctl-binarycopy:{{ etcd_container_bin_dir }}etcdctl {{ bin_dir }}/etcdctl &&
+           {{ docker_bin_dir }}/docker cp etcdctl-binarycopy:{{ etcd_container_bin_dir }}etcdctl {{ bin_dir }}/etcdctl &&
-           /usr/bin/docker rm -f etcdctl-binarycopy"
+           {{ docker_bin_dir }}/docker rm -f etcdctl-binarycopy"
   when: etcd_deployment_type == "docker"
   register: etcd_task_result
   until: etcd_task_result.rc == 0

roles/etcd/tasks/pre_upgrade.yml

@@ -26,12 +26,12 @@
     - /etc/init.d/etcd-proxy
 
 - name: "Pre-upgrade | find etcd-proxy container"
-  command: docker ps -aq --filter "name=etcd-proxy*"
+  command: "{{ docker_bin_dir }}/docker ps -aq --filter 'name=etcd-proxy*'"
   register: etcd_proxy_container
   failed_when: false
 
 - name: "Pre-upgrade | remove etcd-proxy if it exists"
-  command: "docker rm -f {{item}}"
+  command: "{{ docker_bin_dir }}/docker rm -f {{item}}"
   with_items: "{{etcd_proxy_container.stdout_lines}}"
 
 - name: "Pre-upgrade | check if member list is non-SSL"

roles/etcd/templates/deb-etcd-docker.initd.j2

@@ -15,7 +15,7 @@ set -a
 PATH=/sbin:/usr/sbin:/bin/:/usr/bin
 DESC="etcd k/v store"
 NAME=etcd
-DAEMON={{ docker_bin_dir | default("/usr/bin") }}/docker
+DAEMON={{ docker_bin_dir }}/docker
 DAEMON_EXEC=`basename $DAEMON`
 DAEMON_ARGS="run --restart=on-failure:5 --env-file=/etc/etcd.env \
 --net=host \
@@ -50,7 +50,7 @@ do_status()
 #
 do_start()
 {
-    {{ docker_bin_dir | default("/usr/bin") }}/docker rm -f {{ etcd_member_name | default("etcd-proxy") }} &>/dev/null || true
+    {{ docker_bin_dir }}/docker rm -f {{ etcd_member_name | default("etcd") }} &>/dev/null || true
     sleep 1
     start-stop-daemon --background --start --quiet --make-pidfile --pidfile $PID --user $DAEMON_USER --exec $DAEMON -- \
         $DAEMON_ARGS \

roles/etcd/templates/etcd-docker.service.j2

@@ -6,7 +6,7 @@ After=docker.service
 [Service]
 User=root
 PermissionsStartOnly=true
-ExecStart={{ docker_bin_dir | default("/usr/bin") }}/docker run --restart=on-failure:5 \
+ExecStart={{ docker_bin_dir }}/docker run --restart=on-failure:5 \
 --env-file=/etc/etcd.env \
 {# TODO(mattymo): Allow docker IP binding and disable in envfile
    -p 2380:2380 -p 2379:2379 #}
@@ -14,14 +14,15 @@ ExecStart={{ docker_bin_dir | default("/usr/bin") }}/docker run --restart=on-fai
 -v /etc/ssl/certs:/etc/ssl/certs:ro \
 -v {{ etcd_cert_dir }}:{{ etcd_cert_dir }}:ro \
 -v /var/lib/etcd:/var/lib/etcd:rw \
+--memory={{ etcd_memory_limit|regex_replace('Mi', 'M') }} --cpu-shares={{ etcd_cpu_limit|regex_replace('m', '') }} \
 --name={{ etcd_member_name | default("etcd") }} \
 {{ etcd_image_repo }}:{{ etcd_image_tag }} \
 {% if etcd_after_v3 %}
 {{ etcd_container_bin_dir }}etcd
 {% endif %}
-ExecStartPre=-{{ docker_bin_dir | default("/usr/bin") }}/docker rm -f {{ etcd_member_name | default("etcd-proxy") }}
+ExecStartPre=-{{ docker_bin_dir }}/docker rm -f {{ etcd_member_name | default("etcd") }}
-ExecReload={{ docker_bin_dir | default("/usr/bin") }}/docker restart {{ etcd_member_name | default("etcd-proxy") }}
+ExecReload={{ docker_bin_dir }}/docker restart {{ etcd_member_name | default("etcd") }}
-ExecStop={{ docker_bin_dir | default("/usr/bin") }}/docker stop {{ etcd_member_name | default("etcd-proxy") }}
+ExecStop={{ docker_bin_dir }}/docker stop {{ etcd_member_name | default("etcd") }}
 Restart=always
 RestartSec=15s
 

roles/kubernetes-apps/ansible/defaults/main.yml

@@ -20,6 +20,12 @@ exechealthz_image_tag: "{{ exechealthz_version }}"
 calico_policy_image_repo: "calico/kube-policy-controller"
 calico_policy_image_tag: latest
 
+# Limits for calico apps
+calico_policy_controller_cpu_limit: 100m
+calico_policy_controller_memory_limit: 256M
+calico_policy_controller_cpu_requests: 30m
+calico_policy_controller_memory_requests: 128M
+
 # Netchecker
 deploy_netchecker: false
 netchecker_port: 31081
@@ -29,5 +35,19 @@ agent_img: "quay.io/l23network/mcp-netchecker-agent:v0.1"
 server_img: "quay.io/l23network/mcp-netchecker-server:v0.1"
 kubectl_image: "gcr.io/google_containers/kubectl:v0.18.0-120-gaeb4ac55ad12b1-dirty"
 
+# Limits for netchecker apps
+netchecker_agent_cpu_limit: 30m
+netchecker_agent_memory_limit: 100M
+netchecker_agent_cpu_requests: 15m
+netchecker_agent_memory_requests: 64M
+netchecker_server_cpu_limit: 100m
+netchecker_server_memory_limit: 256M
+netchecker_server_cpu_requests: 50m
+netchecker_server_memory_requests: 128M
+netchecker_kubectl_cpu_limit: 30m
+netchecker_kubectl_memory_limit: 128M
+netchecker_kubectl_cpu_requests: 15m
+netchecker_kubectl_memory_requests: 64M
+
 # SSL
 etcd_cert_dir: "/etc/ssl/etcd/ssl"

roles/kubernetes-apps/ansible/templates/calico-policy-controller.yml.j2

@@ -25,6 +25,13 @@ spec:
         - name: calico-policy-controller
           image: {{ calico_policy_image_repo }}:{{ calico_policy_image_tag }}
           imagePullPolicy: {{ k8s_image_pull_policy }}
+          resources:
+            limits:
+              cpu: {{ calico_policy_controller_cpu_limit }}
+              memory: {{ calico_policy_controller_memory_limit }}
+            requests:
+              cpu: {{ calico_policy_controller_cpu_requests }}
+              memory: {{ calico_policy_controller_memory_requests }}
           env:
             - name: ETCD_ENDPOINTS
               value: "{{ etcd_access_endpoint }}"

roles/kubernetes-apps/ansible/templates/netchecker-agent-ds.yml

@@ -23,3 +23,10 @@ spec:
             - name: REPORT_INTERVAL
               value: '{{ agent_report_interval }}'
           imagePullPolicy: {{ k8s_image_pull_policy }}
+          resources:
+            limits:
+              cpu: {{ netchecker_agent_cpu_limit }}
+              memory: {{ netchecker_agent_memory_limit }}
+            requests:
+              cpu: {{ netchecker_agent_cpu_requests }}
+              memory: {{ netchecker_agent_memory_requests }}

roles/kubernetes-apps/ansible/templates/netchecker-agent-hostnet-ds.yml

@@ -24,3 +24,10 @@ spec:
             - name: REPORT_INTERVAL
               value: '{{ agent_report_interval }}'
           imagePullPolicy: {{ k8s_image_pull_policy }}
+          resources:
+            limits:
+              cpu: {{ netchecker_agent_cpu_limit }}
+              memory: {{ netchecker_agent_memory_limit }}
+            requests:
+              cpu: {{ netchecker_agent_cpu_requests }}
+              memory: {{ netchecker_agent_memory_requests }}

roles/kubernetes-apps/ansible/templates/netchecker-server-pod.yml

@@ -11,11 +11,25 @@ spec:
       image: "{{ server_img }}"
       env:
       imagePullPolicy: {{ k8s_image_pull_policy }}
+      resources:
+        limits:
+          cpu: {{ netchecker_server_cpu_limit }}
+          memory: {{ netchecker_server_memory_limit }}
+        requests:
+          cpu: {{ netchecker_server_cpu_requests }}
+          memory: {{ netchecker_server_memory_requests }}
       ports:
         - containerPort: 8081
           hostPort: 8081
     - name: kubectl-proxy
       image: "{{ kubectl_image }}"
       imagePullPolicy: {{ k8s_image_pull_policy }}
+      resources:
+        limits:
+          cpu: {{ netchecker_kubectl_cpu_limit }}
+          memory: {{ netchecker_kubectl_memory_limit }}
+        requests:
+          cpu: {{ netchecker_kubectl_cpu_requests }}
+          memory: {{ netchecker_kubectl_memory_requests }}
       args:
         - proxy

roles/kubernetes/master/defaults/main.yml

@@ -13,4 +13,16 @@ kube_apiserver_node_port_range: "30000-32767"
 etcd_config_dir: /etc/ssl/etcd
 etcd_cert_dir: "{{ etcd_config_dir }}/ssl"
 
-
+# Limits for kube components
+kube_controller_memory_limit: 512M
+kube_controller_cpu_limit: 250m
+kube_controller_memory_requests: 170M
+kube_controller_cpu_requests: 100m
+kube_scheduler_memory_limit: 512M
+kube_scheduler_cpu_limit: 250m
+kube_scheduler_memory_requests: 170M
+kube_scheduler_cpu_requests: 100m
+kube_apiserver_memory_limit: 2000M
+kube_apiserver_cpu_limit: 800m
+kube_apiserver_memory_requests: 256M
+kube_apiserver_cpu_requests: 300m

roles/kubernetes/master/tasks/main.yml

@@ -3,7 +3,7 @@
   tags: k8s-pre-upgrade
 
 - name: Copy kubectl from hyperkube container
-  command: "/usr/bin/docker run --rm -v {{ bin_dir }}:/systembindir {{ hyperkube_image_repo }}:{{ hyperkube_image_tag }} /bin/cp /hyperkube /systembindir/kubectl"
+  command: "{{ docker_bin_dir }}/docker run --rm -v {{ bin_dir }}:/systembindir {{ hyperkube_image_repo }}:{{ hyperkube_image_tag }} /bin/cp /hyperkube /systembindir/kubectl"
   register: kube_task_result
   until: kube_task_result.rc == 0
   retries: 4

roles/kubernetes/master/templates/manifests/kube-apiserver.manifest.j2

@@ -12,6 +12,13 @@ spec:
   - name: kube-apiserver
     image: {{ hyperkube_image_repo }}:{{ hyperkube_image_tag }}
     imagePullPolicy: {{ k8s_image_pull_policy }}
+    resources:
+      limits:
+        cpu: {{ kube_apiserver_cpu_limit }}
+        memory: {{ kube_apiserver_memory_limit }}
+      requests:
+        cpu: {{ kube_apiserver_cpu_requests }}
+        memory: {{ kube_apiserver_memory_requests }}
     command:
     - /hyperkube
     - apiserver

roles/kubernetes/master/templates/manifests/kube-controller-manager.manifest.j2

@@ -11,6 +11,13 @@ spec:
   - name: kube-controller-manager
     image: {{ hyperkube_image_repo }}:{{ hyperkube_image_tag }}
     imagePullPolicy: {{ k8s_image_pull_policy }}
+    resources:
+      limits:
+        cpu: {{ kube_controller_cpu_limit }}
+        memory: {{ kube_controller_memory_limit }}
+      requests:
+        cpu: {{ kube_controller_cpu_requests }}
+        memory: {{ kube_controller_memory_requests }}
     command:
     - /hyperkube
     - controller-manager

roles/kubernetes/master/templates/manifests/kube-scheduler.manifest.j2

@@ -11,6 +11,13 @@ spec:
   - name: kube-scheduler
     image: {{ hyperkube_image_repo }}:{{ hyperkube_image_tag }}
     imagePullPolicy: {{ k8s_image_pull_policy }}
+    resources:
+      limits:
+        cpu: {{ kube_scheduler_cpu_limit }}
+        memory: {{ kube_scheduler_memory_limit }}
+      requests:
+        cpu: {{ kube_scheduler_cpu_requests }}
+        memory: {{ kube_scheduler_memory_requests }}
     command:
     - /hyperkube
     - scheduler

roles/kubernetes/node/defaults/main.yml

@@ -9,6 +9,18 @@ kube_proxy_mode: iptables
 # If using the pure iptables proxy, SNAT everything
 kube_proxy_masquerade_all: true
 
+# Limits for kube components and nginx load balancer app
+kubelet_memory_limit: 512M
+kubelet_cpu_limit: 100m
+kube_proxy_memory_limit: 2000M
+kube_proxy_cpu_limit: 500m
+kube_proxy_memory_requests: 256M
+kube_proxy_cpu_requests: 150m
+nginx_memory_limit: 512M
+nginx_cpu_limit: 300m
+nginx_memory_requests: 64M
+nginx_cpu_requests: 50m
+
 # kube_api_runtime_config:
 #   - extensions/v1beta1/daemonsets=true
 #   - extensions/v1beta1/deployments=true

roles/kubernetes/node/templates/deb-kubelet.initd.j2

@@ -39,7 +39,7 @@ DAEMON_USER=root
 #
 do_start()
 {
-        /usr/bin/docker rm -f kubelet &>/dev/null || true
+        {{ docker_bin_dir }}/docker rm -f kubelet &>/dev/null || true
         sleep 1
         # Return
         #   0 if daemon has been started

roles/kubernetes/node/templates/kubelet-container.j2

@@ -1,5 +1,5 @@
 #!/bin/bash
-/usr/bin/docker run --privileged \
+{{ docker_bin_dir }}/docker run --privileged \
 --net=host --pid=host --name=kubelet --restart=on-failure:5 \
 -v /etc/cni:/etc/cni:ro \
 -v /opt/cni:/opt/cni:ro \
@@ -9,6 +9,7 @@
 -v {{ docker_daemon_graph }}:/var/lib/docker \
 -v /var/run:/var/run \
 -v /var/lib/kubelet:/var/lib/kubelet \
+--memory={{ kubelet_memory_limit|regex_replace('Mi', 'M') }} --cpu-shares={{ kubelet_cpu_limit|regex_replace('m', '')  }} \
 {{ hyperkube_image_repo }}:{{ hyperkube_image_tag}} \
 nsenter --target=1 --mount --wd=. -- \
 ./hyperkube kubelet \

roles/kubernetes/node/templates/kubelet.service.j2

@@ -23,8 +23,8 @@ ExecStart={{ bin_dir }}/kubelet \
 		$DOCKER_SOCKET \
 		$KUBELET_NETWORK_PLUGIN \
 		$KUBELET_CLOUDPROVIDER
-ExecStartPre=-/usr/bin/docker rm -f kubelet
+ExecStartPre=-{{ docker_bin_dir }}/docker rm -f kubelet
-ExecReload=/usr/bin/docker restart kubelet
+ExecReload={{ docker_bin_dir }}/docker restart kubelet
 Restart=always
 RestartSec=10s
 

roles/kubernetes/node/templates/manifests/kube-proxy.manifest.j2

@@ -11,6 +11,13 @@ spec:
   - name: kube-proxy
     image: {{ hyperkube_image_repo }}:{{ hyperkube_image_tag }}
     imagePullPolicy: {{ k8s_image_pull_policy }}
+    resources:
+      limits:
+        cpu: {{ kube_proxy_cpu_limit }}
+        memory: {{ kube_proxy_memory_limit }}
+      requests:
+        cpu: {{ kube_proxy_cpu_requests }}
+        memory: {{ kube_proxy_memory_requests }}
     command:
     - /hyperkube
     - proxy

roles/kubernetes/node/templates/manifests/nginx-proxy.manifest.j2

@@ -11,6 +11,13 @@ spec:
   - name: nginx-proxy
     image: {{ nginx_image_repo }}:{{ nginx_image_tag }}
     imagePullPolicy: {{ k8s_image_pull_policy }}
+    resources:
+      limits:
+        cpu: {{ nginx_cpu_limit }}
+        memory: {{ nginx_memory_limit }}
+      requests:
+        cpu: {{ nginx_cpu_requests }}
+        memory: {{ nginx_memory_requests }}
     securityContext:
       privileged: true
     volumeMounts:

roles/network_plugin/calico/defaults/main.yml

@@ -19,3 +19,17 @@ global_as_num: "64512"
 # not be specified in calico CNI config, so Calico will use built-in
 # defaults. The value should be a number, not a string.
 # calico_mtu: 1500
+
+# Limits for apps
+calico_rr_memory_limit: 1000M
+calico_rr_cpu_limit: 300m
+calico_rr_memory_requests: 500M
+calico_rr_cpu_requests: 150m
+calico_node_memory_limit: 500M
+calico_node_cpu_limit: 300m
+calico_node_memory_requests: 256M
+calico_node_cpu_requests: 150m
+calicoctl_memory_limit: 170M
+calicoctl_cpu_limit: 100m
+calicoctl_memory_requests: 70M
+calicoctl_cpu_requests: 50m

roles/network_plugin/calico/rr/templates/calico-rr.service.j2

@@ -5,8 +5,8 @@ Requires=docker.service
 
 [Service]
 EnvironmentFile=/etc/calico/calico-rr.env
-ExecStartPre=-/usr/bin/docker rm -f calico-rr
+ExecStartPre=-{{ docker_bin_dir }}/docker rm -f calico-rr
-ExecStart=/usr/bin/docker run --net=host --privileged \
+ExecStart={{ docker_bin_dir }}/docker run --net=host --privileged \
  --name=calico-rr \
  -e IP=${IP} \
  -e IP6=${IP6} \
@@ -16,12 +16,13 @@ ExecStart=/usr/bin/docker run --net=host --privileged \
  -e ETCD_KEY_FILE=${ETCD_KEY_FILE} \
  -v /var/log/calico-rr:/var/log/calico \
  -v {{ calico_cert_dir }}:{{ calico_cert_dir }}:ro \
+ --memory={{ calico_rr_memory_limit|regex_replace('Mi', 'M') }} --cpu-shares={{ calico_rr_cpu_limit|regex_replace('m', '') }} \
  {{ calico_rr_image_repo }}:{{ calico_rr_image_tag }}
 
 Restart=always
 RestartSec=10s
 
-ExecStop=-/usr/bin/docker stop calico-rr
+ExecStop=-{{ docker_bin_dir }}/docker stop calico-rr
 
 [Install]
 WantedBy=multi-user.target

roles/network_plugin/calico/tasks/main.yml

@@ -41,7 +41,7 @@
   notify: restart calico-node
 
 - name: Calico | Copy cni plugins from hyperkube
-  command: "/usr/bin/docker run --rm -v /opt/cni/bin:/cnibindir {{ hyperkube_image_repo }}:{{ hyperkube_image_tag }} /usr/bin/rsync -a /opt/cni/bin/ /cnibindir/"
+  command: "{{ docker_bin_dir }}/docker run --rm -v /opt/cni/bin:/cnibindir {{ hyperkube_image_repo }}:{{ hyperkube_image_tag }} /usr/bin/rsync -a /opt/cni/bin/ /cnibindir/"
   register: cni_task_result
   until: cni_task_result.rc == 0
   retries: 4
@@ -50,7 +50,7 @@
   tags: [hyperkube, upgrade]
 
 - name: Calico | Copy cni plugins from calico/cni container
-  command: "/usr/bin/docker run --rm -v /opt/cni/bin:/cnibindir {{ calico_cni_image_repo }}:{{ calico_cni_image_tag }} sh -c 'cp -a /opt/cni/bin/* /cnibindir/'"
+  command: "{{ docker_bin_dir }}/docker run --rm -v /opt/cni/bin:/cnibindir {{ calico_cni_image_repo }}:{{ calico_cni_image_tag }} sh -c 'cp -a /opt/cni/bin/* /cnibindir/'"
   register: cni_task_result
   until: cni_task_result.rc == 0
   retries: 4

roles/network_plugin/calico/templates/calico-node.service.j2

@@ -5,8 +5,8 @@ Requires=docker.service
 
 [Service]
 EnvironmentFile=/etc/calico/calico.env
-ExecStartPre=-/usr/bin/docker rm -f calico-node
+ExecStartPre=-{{ docker_bin_dir }}/docker rm -f calico-node
-ExecStart=/usr/bin/docker run --net=host --privileged \
+ExecStart={{ docker_bin_dir }}/docker run --net=host --privileged \
  --name=calico-node \
  -e HOSTNAME=${CALICO_HOSTNAME} \
  -e IP=${CALICO_IP} \
@@ -24,12 +24,13 @@ ExecStart=/usr/bin/docker run --net=host --privileged \
  -v /lib/modules:/lib/modules \
  -v /var/run/calico:/var/run/calico \
  -v {{ calico_cert_dir }}:{{ calico_cert_dir }}:ro \
+ --memory={{ calico_node_memory_limit|regex_replace('Mi', 'M') }} --cpu-shares={{ calico_node_cpu_limit|regex_replace('m', '') }} \
  {{ calico_node_image_repo }}:{{ calico_node_image_tag }}
 
 Restart=always
 RestartSec=10s
 
-ExecStop=-/usr/bin/docker stop calico-node
+ExecStop=-{{ docker_bin_dir }}/docker stop calico-node
 
 [Install]
 WantedBy=multi-user.target

roles/network_plugin/calico/templates/calicoctl-container.j2

@@ -1,13 +1,14 @@
 #!/bin/bash
-/usr/bin/docker run -i --privileged --rm \
+{{ docker_bin_dir }}/docker run -i --privileged --rm \
 --net=host --pid=host \
 -e ETCD_ENDPOINTS={{ etcd_access_endpoint }} \
 -e ETCD_CA_CERT_FILE=/etc/calico/certs/ca_cert.crt \
 -e ETCD_CERT_FILE=/etc/calico/certs/cert.crt \
 -e ETCD_KEY_FILE=/etc/calico/certs/key.pem \
--v /usr/bin/docker:/usr/bin/docker \
+-v {{ docker_bin_dir }}/docker:{{ docker_bin_dir }}/docker \
 -v /var/run/docker.sock:/var/run/docker.sock \
 -v /var/run/calico:/var/run/calico \
 -v /etc/calico/certs:/etc/calico/certs:ro \
+--memory={{ calicoctl_memory_limit|regex_replace('Mi', 'M') }} --cpu-shares={{ calicoctl_cpu_limit|regex_replace('m', '') }} \
 {{ calicoctl_image_repo }}:{{ calicoctl_image_tag}} \
 $@

roles/network_plugin/canal/defaults/main.yml

@@ -13,3 +13,13 @@ canal_log_level: "info"
 # Etcd SSL dirs
 canal_cert_dir: /etc/canal/certs
 etcd_cert_dir: /etc/ssl/etcd/ssl
+
+# Limits for apps
+calico_node_memory_limit: 500M
+calico_node_cpu_limit: 200m
+calico_node_memory_requests: 256M
+calico_node_cpu_requests: 100m
+flannel_memory_limit: 500M
+flannel_cpu_limit: 200m
+flannel_memory_requests: 256M
+flannel_cpu_requests: 100m

roles/network_plugin/canal/tasks/main.yml

@@ -43,7 +43,7 @@
     dest: "{{kube_config_dir}}/canal-node.yaml"
 
 - name: Canal | Copy cni plugins from hyperkube
-  command: "/usr/bin/docker run --rm -v /opt/cni/bin:/cnibindir {{ hyperkube_image_repo }}:{{ hyperkube_image_tag }} /usr/bin/rsync -a /opt/cni/bin/ /cnibindir/"
+  command: "{{ docker_bin_dir }}/docker run --rm -v /opt/cni/bin:/cnibindir {{ hyperkube_image_repo }}:{{ hyperkube_image_tag }} /usr/bin/rsync -a /opt/cni/bin/ /cnibindir/"
   register: cni_task_result
   until: cni_task_result.rc == 0
   retries: 4
@@ -52,7 +52,7 @@
   tags: [hyperkube, upgrade]
 
 - name: Canal | Copy cni plugins from calico/cni
-  command: "/usr/bin/docker run --rm -v /opt/cni/bin:/cnibindir {{ calico_cni_image_repo }}:{{ calico_cni_image_tag }} sh -c 'cp -a /opt/cni/bin/* /cnibindir/'"
+  command: "{{ docker_bin_dir }}/docker run --rm -v /opt/cni/bin:/cnibindir {{ calico_cni_image_repo }}:{{ calico_cni_image_tag }} sh -c 'cp -a /opt/cni/bin/* /cnibindir/'"
   register: cni_task_result
   until: cni_task_result.rc == 0
   retries: 4

roles/network_plugin/canal/templates/canal-node.yml.j2

@@ -49,6 +49,13 @@ spec:
         - name: flannel
           image: "{{ flannel_image_repo }}:{{ flannel_image_tag }}"
           imagePullPolicy: {{ k8s_image_pull_policy }}
+          resources:
+            limits:
+              cpu: {{ flannel_cpu_limit }}
+              memory: {{ flannel_memory_limit }}
+            requests:
+              cpu: {{ flannel_cpu_requests }}
+              memory: {{ flannel_memory_requests }}
           env:
             # Cluster name
             - name: CLUSTER_NAME
@@ -119,6 +126,13 @@ spec:
         - name: calico-node
           image: "{{ calico_node_image_repo }}:{{ calico_node_image_tag }}"
           imagePullPolicy: {{ k8s_image_pull_policy }}
+          resources:
+            limits:
+              cpu: {{ calico_node_cpu_limit }}
+              memory: {{ calico_node_memory_limit }}
+            requests:
+              cpu: {{ calico_node_cpu_requests }}
+              memory: {{ calico_node_memory_requests }}
           env:
             # The location of the etcd cluster.
             - name: ETCD_ENDPOINTS

roles/network_plugin/cloud/tasks/main.yml

@@ -1,7 +1,7 @@
 ---
 
 - name: Cloud | Copy cni plugins from hyperkube
-  command: "/usr/bin/docker run --rm -v /opt/cni/bin:/cnibindir {{ hyperkube_image_repo }}:{{ hyperkube_image_tag }} /bin/cp -r /opt/cni/bin/. /cnibindir/"
+  command: "{{ docker_bin_dir }}/docker run --rm -v /opt/cni/bin:/cnibindir {{ hyperkube_image_repo }}:{{ hyperkube_image_tag }} /bin/cp -r /opt/cni/bin/. /cnibindir/"
   register: cni_task_result
   until: cni_task_result.rc == 0
   retries: 4

roles/network_plugin/flannel/defaults/main.yml

@@ -10,3 +10,9 @@ flannel_public_ip: "{{ access_ip|default(ip|default(ansible_default_ipv4.address
 # You can choose what type of flannel backend to use
 # please refer to flannel's docs : https://github.com/coreos/flannel/blob/master/README.md
 flannel_backend_type: "vxlan"
+
+# Limits for apps
+flannel_memory_limit: 500M
+flannel_cpu_limit: 300m
+flannel_memory_requests: 256M
+flannel_cpu_requests: 150m

roles/network_plugin/flannel/handlers/main.yml

@@ -32,7 +32,7 @@
   pause: seconds=10 prompt="Waiting for docker restart"
 
 - name: Flannel | wait for docker
-  command: /usr/bin/docker images
+  command: "{{ docker_bin_dir }}/docker images"
   register: docker_ready
   retries: 10
   delay: 5

roles/network_plugin/flannel/templates/flannel-pod.yml

@@ -19,6 +19,13 @@
       - name: "flannel-container"
         image: "{{ flannel_image_repo }}:{{ flannel_image_tag }}"
         imagePullPolicy: {{ k8s_image_pull_policy }}
+        resources:
+          limits:
+            cpu: {{ flannel_cpu_limit }}
+            memory: {{ flannel_memory_limit }}
+          requests:
+            cpu: {{ flannel_cpu_requests }}
+            memory: {{ flannel_memory_requests }}
         command:
           - "/bin/sh"
           - "-c"
@@ -26,9 +33,6 @@
         ports:
           - hostPort: 10253
             containerPort: 10253
-        resources:
-          limits:
-            cpu: "100m"
         volumeMounts:
           - name: "subnetenv"
             mountPath: "/run/flannel"

roles/network_plugin/weave/defaults/main.yml

@@ -0,0 +1,4 @@
+---
+# Limits
+weave_memory_limit: 500M
+weave_cpu_limit: 300m

roles/network_plugin/weave/tasks/main.yml

@@ -1,6 +1,6 @@
 ---
 - name: Weave | Copy cni plugins from hyperkube
-  command: "/usr/bin/docker run --rm -v /opt/cni/bin:/cnibindir {{ hyperkube_image_repo }}:{{ hyperkube_image_tag }} /bin/cp -r /opt/cni/bin/. /cnibindir/"
+  command: "{{ docker_bin_dir }}/docker run --rm -v /opt/cni/bin:/cnibindir {{ hyperkube_image_repo }}:{{ hyperkube_image_tag }} /bin/cp -r /opt/cni/bin/. /cnibindir/"
   register: cni_task_result
   until: cni_task_result.rc == 0
   retries: 4

roles/network_plugin/weave/templates/weave.j2

@@ -1,3 +1,4 @@
+WEAVE_DOCKER_ARGS="--memory={{ weave_memory_limit|regex_replace('Mi', 'M') }} --cpu-shares={{ weave_cpu_limit|regex_replace('m', '') }}"
 WEAVE_PEERS="{% for host in groups['k8s-cluster'] %}{{ hostvars[host]['access_ip'] | default(hostvars[host]['ip'] | default(hostvars[host]['ansible_default_ipv4']['address'])) }}{% if not loop.last %} {% endif %}{% endfor %}"
 WEAVEPROXY_ARGS="--rewrite-inspect --without-dns"
 WEAVE_SUBNET="--ipalloc-range {{ kube_pods_subnet }}"

roles/network_plugin/weave/templates/weave.service.j2

@@ -6,12 +6,13 @@ After=docker.service docker.socket
 
 [Service]
 EnvironmentFile=-/etc/weave.env
-ExecStartPre=-/usr/bin/docker rm -f weave
+ExecStartPre=-{{ docker_bin_dir }}/docker rm -f weave
 ExecStartPre={{ bin_dir }}/weave launch-router \
             $WEAVE_SUBNET \
             $WEAVE_PEERS
-ExecStart=/usr/bin/docker attach weave
+ExecStart={{ docker_bin_dir }}/docker attach weave
 ExecStop={{ bin_dir }}/weave stop
+Restart=on-failure
 
 [Install]
 WantedBy=multi-user.target

roles/network_plugin/weave/templates/weaveproxy.service.j2

@@ -7,11 +7,11 @@ After=docker.service docker.socket
 [Service]
 EnvironmentFile=-/etc/weave.%H.env
 EnvironmentFile=-/etc/weave.env
-ExecStartPre=-/usr/bin/docker rm -f weaveproxy
+ExecStartPre=-{{ docker_bin_dir }}/docker rm -f weaveproxy
 ExecStartPre={{ bin_dir }}/weave launch-proxy $WEAVEPROXY_ARGS
-ExecStart=/usr/bin/docker attach weaveproxy
+ExecStart={{ docker_bin_dir }}/docker attach weaveproxy
 Restart=on-failure
-ExecStop=/opt/bin/weave stop-proxy
+ExecStop={{ bin_dir }}/weave stop-proxy
 
 [Install]
 WantedBy=weave-network.target

roles/reset/tasks/main.yml

@@ -21,7 +21,7 @@
   when: ansible_service_mgr == "systemd" and services_removed.changed
 
 - name: reset | remove all containers
-  shell: docker ps -aq | xargs -r docker rm -fv
+  shell: "{{ docker_bin_dir }}/docker ps -aq | xargs -r docker rm -fv"
 
 - name: reset | gather mounted kubelet dirs
   shell: mount | grep /var/lib/kubelet | awk '{print $3}' | tac

scripts/collect-info.yaml

@@ -10,7 +10,7 @@
       - name: kernel_info
         cmd: uname -r
       - name: docker_info
-        cmd: docker info
+        cmd: "{{ docker_bin_dir }}/docker info"
       - name: ip_info
         cmd: ip -4 -o a
       - name: route_info
@@ -34,9 +34,11 @@
       - name: weave_info
         cmd: weave report
       - name: weave_logs
-        cmd: docker logs weave
+        cmd: "{{ docker_bin_dir }}/docker logs weave"
       - name: kube_describe_all
         cmd: kubectl describe all --all-namespaces
+      - name: kube_describe_nodes
+        cmd: kubectl describe nodes
       - name: kubelet_logs
         cmd: journalctl -u kubelet --no-pager
       - name: kubedns_logs