C12s/c12s-kubespray
3
0
Fork 0
Code Issues Pull requests Packages Projects Releases Wiki Activity

Fixed issue #7112.  Created new API Server vars that replace defunct Controller Manager one (#7114)

Browse source 
Signed-off-by: Brendan Holmes <5072156+holmesb@users.noreply.github.com>
This commit is contained in:
holmesb 2021-01-08 15:20:53 +00:00 committed by GitHub
parent ab2bfd7f8c
commit b0ad8ec023
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
4 changed files with 25 additions and 13 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

25
docs/kubernetes-reliability.md
View file

@ -43,8 +43,10 @@ attempts to set a status of node.
At the same time Kubernetes controller manager will try to check
`nodeStatusUpdateRetry` times every `--node-monitor-period` of time. After
`--node-monitor-grace-period` it will consider node unhealthy. It will remove
its pods based on `--pod-eviction-timeout`
`--node-monitor-grace-period` it will consider node unhealthy. Pods will then be rescheduled based on the
[Taint Based Eviction](https://kubernetes.io/docs/concepts/scheduling-eviction/taint-and-toleration/#taint-based-evictions)
timers that you set on them individually, or the API Server's global timers:`--default-not-ready-toleration-seconds` &
``--default-unreachable-toleration-seconds``.
Kube proxy has a watcher over API. Once pods are evicted, Kube proxy will
notice and will update iptables of the node. It will remove endpoints from
@ -57,12 +59,14 @@ services so pods from failed node won't be accessible anymore.
If `-–node-status-update-frequency` is set to **4s** (10s is default).
`--node-monitor-period` to **2s** (5s is default).
`--node-monitor-grace-period` to **20s** (40s is default).
`--pod-eviction-timeout` is set to **30s** (5m is default)
`--default-not-ready-toleration-seconds` and ``--default-unreachable-toleration-seconds`` are set to **30**
(300 seconds is default). Note these two values should be integers representing the number of seconds ("s" or "m" for
seconds\minutes are not specified).
In such scenario, pods will be evicted in **50s** because the node will be
considered as down after **20s**, and `--pod-eviction-timeout` occurs after
**30s** more. However, this scenario creates an overhead on etcd as every node
will try to update its status every 2 seconds.
considered as down after **20s**, and `--default-not-ready-toleration-seconds` or
``--default-unreachable-toleration-seconds`` occur after **30s** more. However, this scenario creates an overhead on
etcd as every node will try to update its status every 2 seconds.
If the environment has 1000 nodes, there will be 15000 node updates per
minute which may require large etcd containers or even dedicated nodes for etcd.
@ -75,7 +79,8 @@ minute which may require large etcd containers or even dedicated nodes for etcd.
## Medium Update and Average Reaction
Let's set `-–node-status-update-frequency` to **20s**
`--node-monitor-grace-period` to **2m** and `--pod-eviction-timeout` to **1m**.
`--node-monitor-grace-period` to **2m** and `--default-not-ready-toleration-seconds` and
``--default-unreachable-toleration-seconds`` to **60**.
In that case, Kubelet will try to update status every 20s. So, it will be 6 * 5
= 30 attempts before Kubernetes controller manager will consider unhealthy
status of node. After 1m it will evict all pods. The total time will be 3m
@ -90,9 +95,9 @@ etcd updates per minute.
## Low Update and Slow reaction
Let's set `-–node-status-update-frequency` to **1m**.
`--node-monitor-grace-period` will set to **5m** and `--pod-eviction-timeout`
to **1m**. In this scenario, every kubelet will try to update the status every
minute. There will be 5 * 5 = 25 attempts before unhealthy status. After 5m,
`--node-monitor-grace-period` will set to **5m** and `--default-not-ready-toleration-seconds` and
``--default-unreachable-toleration-seconds`` to **60**. In this scenario, every kubelet will try to update the status
every minute. There will be 5 * 5 = 25 attempts before unhealthy status. After 5m,
Kubernetes controller manager will set unhealthy status. This means that pods
will be evicted after 1m after being marked unhealthy. (6m in total).

3
docs/large-deployments.md
View file

@ -30,7 +30,8 @@ For a large scaled deployments, consider the following configuration changes:
* Tune ``kubelet_status_update_frequency`` to increase reliability of kubelet.
``kube_controller_node_monitor_grace_period``,
``kube_controller_node_monitor_period``,
``kube_controller_pod_eviction_timeout`` for better Kubernetes reliability.
``kube_apiserver_pod_eviction_not_ready_timeout_seconds`` &
``kube_apiserver_pod_eviction_unreachable_timeout_seconds`` for better Kubernetes reliability.
Check out [Kubernetes Reliability](kubernetes-reliability.md)
* Tune network prefix sizes. Those are ``kube_network_node_prefix``,

3
roles/kubernetes/master/defaults/main/main.yml
View file

@ -86,9 +86,10 @@ audit_webhook_batch_max_wait: 1s
kube_controller_node_monitor_grace_period: 40s
kube_controller_node_monitor_period: 5s
kube_controller_pod_eviction_timeout: 5m0s
kube_controller_terminated_pod_gc_threshold: 12500
kube_apiserver_request_timeout: "1m0s"
kube_apiserver_pod_eviction_not_ready_timeout_seconds: "300"
kube_apiserver_pod_eviction_unreachable_timeout_seconds: "300"
# 1.10+ admission plugins
kube_apiserver_enable_admission_plugins: []

7
roles/kubernetes/master/templates/kubeadm-config.v1beta2.yaml.j2
View file

@ -100,6 +100,12 @@ certificatesDir: {{ kube_cert_dir }}
imageRepository: {{ kube_image_repo }}
apiServer:
extraArgs:
{% if kube_apiserver_pod_eviction_not_ready_timeout_seconds is defined %}
default-not-ready-toleration-seconds: "{{ kube_apiserver_pod_eviction_not_ready_timeout_seconds }}"
{% endif %}
{% if kube_apiserver_pod_eviction_unreachable_timeout_seconds is defined %}
default-unreachable-toleration-seconds: "{{ kube_apiserver_pod_eviction_unreachable_timeout_seconds }}"
{% endif %}
{% if kube_api_anonymous_auth is defined %}
anonymous-auth: "{{ kube_api_anonymous_auth }}"
{% endif %}
@ -256,7 +262,6 @@ controllerManager:
extraArgs:
node-monitor-grace-period: {{ kube_controller_node_monitor_grace_period }}
node-monitor-period: {{ kube_controller_node_monitor_period }}
pod-eviction-timeout: {{ kube_controller_pod_eviction_timeout }}
node-cidr-mask-size: "{{ kube_network_node_prefix }}"
profiling: "{{ kube_profiling }}"
terminated-pod-gc-threshold: "{{ kube_controller_terminated_pod_gc_threshold }}"