如何列出库伯内特节点上的污点？

小开

You can use kubectl's go-template output options to help you here,

kubectl get nodes -o go-template='\{\{range .items}}\{\{if $x := index .metadata.annotations "scheduler.alpha.kubernetes.io/taints"}}\{\{with $x := index .metadata.name}}\{\{.}}\{\{printf "\n"}}\{\{end}}\{\{end}}\{\{end}}'

On my cluster, this prints my masters, which are tainted:

kubemaster-1.example.net
kubemaster-2.example.net
kubemaster-3.example.net

小开

In Kubernetes 1.6.x the node taints have moved into the spec. Therefore the above answer by jaxxstorm will not work. Instead, you can use the following template.

\{\{printf "%-50s %-12s\n" "Node" "Taint"}}
\{\{- range .items}}
\{\{- if $taint := (index .spec "taints") }}
\{\{- .metadata.name }}\{\{ "\t" }}
\{\{- range $taint }}
\{\{- .key }}=\{\{ .value }}:\{\{ .effect }}\{\{ "\t" }}
\{\{- end }}
\{\{- "\n" }}
\{\{- end}}
\{\{- end}}

I have that saved into a file and then reference it like so:

kubectl get nodes -o go-template-file="./nodes-taints.tmpl"

You'll get output like so:

Node                                            Taint
ip-xxx-xxx-xxx-xxx.us-west-2.compute.internal   dedicate=etcd:NoSchedule
ip-xxx-xxx-xxx-xxx.us-west-2.compute.internal   dedicate=jenkins:NoSchedule
ip-xxx-xxx-xxx-xxx.us-west-2.compute.internal   dedicate=etcd:NoSchedule
ip-xxx-xxx-xxx-xxx.us-west-2.compute.internal   dedicate=containerlinux-canary-channel-workers:NoSchedule
ip-xxx-xxx-xxx-xxx.us-west-2.compute.internal   dedicate=jenkins:NoSchedule
ip-xxx-xxx-xxx-xxx.us-west-2.compute.internal   dedicate=etcd:NoSchedule
ip-xxx-xxx-xxx-xxx.us-west-2.compute.internal   dedicate=etcd:NoSchedule
ip-xxx-xxx-xxx-xxx.us-west-2.compute.internal   dedicate=etcd:NoSchedule
ip-xxx-xxx-xxx-xxx.us-west-2.compute.internal   dedicate=jenkins:NoSchedule

I'm not a huge go template user so I'm sure there are some things I could have done better but it is what it is.

Same as above but all in one line:

kubectl get nodes -o go-template='\{\{printf "%-50s %-12s\n" "Node" "Taint"}}\{\{- range .items}}\{\{- if $taint := (index .spec "taints") }}\{\{- .metadata.name }}\{\{ "\t" }}\{\{- range $taint }}\{\{- .key }}=\{\{ .value }}:\{\{ .effect }}\{\{ "\t" }}\{\{- end }}\{\{- "\n" }}\{\{- end}}\{\{- end}}'

小开

To find taints of node can just run:

kubectl describe nodes your-node-name

Output:

Name:                   your-node-name
...
Taints:                 node-role.kubernetes.io/master:NoSchedule
CreationTimestamp:      Wed, 19 Jul 2017 06:00:23 +0800

小开

I was looking to get the list of nodes that have a specific Taint. I only found this SO answer, so if anybody is looking for this answer, here is the solution:

kubectl get nodes -o go-template='\{\{range $item := .items}}\{\{with $nodename := $item.metadata.name}}\{\{range $taint := $item.spec.taints}}\{\{if and (eq $taint.key "node-role.kubernetes.io/master") (eq $taint.effect "NoSchedule")}}\{\{printf "%s\n" $nodename}}\{\{end}}\{\{end}}\{\{end}}\{\{end}}'

On my cluster, the output is:

preprod-master
preprod-proxy

小开

PowerShell:\> kubectl describe nodes | findstr "Taint Hostname"

or

Bash# kubectl describe nodes | egrep -hi "Taint|Hostname"

This command is mad easy to remember

Output looks like this:

Taints:             <none>
Hostname:   aks-agentpool-30208295-0
Taints:             <none>
Hostname:  aks-agentpool-30208295-1
...

小开

最佳答案

kubectl get nodes -o json | jq '.items[].spec'

which will give the complete spec with node name, or:

kubectl get nodes -o json | jq '.items[].spec.taints'

will produce the list of the taints per each node

小开

The CMD kubectl provides an arguments jsonpath to search and format the output after getting. You can check the doc k8s jsonpath for detail.

 kubectl get node  -o=jsonpath='{range .items[*]}{.metadata.name}{"\t"}{.spec.taints}{"\n"}{end}'

For further info, you can check the source code that reflect the source data with the method FindResults

小开

kubectl describe nodes [node_name] | grep 'Taints'


kubectl get nodes -o json | jq '.items[].spec.taints'

--> this last step will require jq installed ( sudo apt install jq)

小开

The simplest way to do this without using any extra tools such as JQ is to use the custom-columns output option.

$ kubectl get nodes -o custom-columns=NAME:.metadata.name,TAINTS:.spec.taints --no-headers

Output:

master-11   [map[effect:PreferNoSchedule key:node-role.kubernetes.io/master]]
master-12   [map[effect:PreferNoSchedule key:node-role.kubernetes.io/master]]
master-13   [map[effect:PreferNoSchedule key:node-role.kubernetes.io/master]]

With something like Taints where it is a map or list and you want it to look clean for parsing with some other tool you can clean them up using you can use something similar to the answer by Edwin Tai but with a little extra smarts to extract the keys.

kubectl get nodes -o=jsonpath='{range .items[*]}{.metadata.name}{"\t"}{.spec.taints[*].key}{"\n"}{end}'

Output:

master-11   node-role.kubernetes.io/master
master-12   node-role.kubernetes.io/master
master-13   node-role.kubernetes.io/master
worker-21   thegoldfish.org/storage thegoldfish.org/compute
worker-22   thegoldfish.org/storage thegoldfish.org/compute
worker-23   thegoldfish.org/compute
worker-24   thegoldfish.org/storage thegoldfish.org/compute

Extra examples:

Using this method you can easily create custom outputs

Quick overview of nodes:

kubectl get nodes -o custom-columns=NAME:.metadata.name,ARCH:.status.nodeInfo.architecture,KERNEL:.status.nodeInfo.kernelVersion,KUBLET:.status.nodeInfo.kubeletVersion,CPU:.status.capacity.cpu,RAM:.status.capacity.memory

Output:

NAME        ARCH    KERNEL                       KUBLET    CPU   RAM
master-11   amd64   3.10.0-1062.9.1.el7.x86_64   v1.17.0   6     7910096Ki
master-12   amd64   3.10.0-1062.9.1.el7.x86_64   v1.17.0   6     7910096Ki
master-13   amd64   3.10.0-1062.9.1.el7.x86_64   v1.17.0   6     7910096Ki

Overview of pods and where to find them sorted by creation time:

kubectl get pods -A -o custom-columns=NAMESPACE:.metadata.namespace,NAME:.metadata.name,NODE:.spec.nodeName,HOSTIP:.status.hostIP,PHASE:.status.phase,START_TIME:.metadata.creationTimestamp --sort-by=.metadata.creationTimestamp

Output:

NAMESPACE              NAME                                                  NODE        HOSTIP            PHASE       START_TIME
kube-system            kube-proxy-rhmrz                                      master-11   192.168.121.108   Running     2019-12-26T14:22:03Z
kube-system            coredns-6955765f44-777v9                              master-11   192.168.121.108   Running     2019-12-26T14:22:03Z
kube-system            coredns-6955765f44-w7rch                              master-11   192.168.121.108   Running     2019-12-26T14:22:03Z
kube-system            kube-scheduler-master-11                              master-11   192.168.121.108   Running     2019-12-26T14:22:05Z
kube-system            kube-controller-manager-master-11                     master-11   192.168.121.108   Running     2019-12-26T14:22:05Z
kube-system            etcd-master-11                                        master-11   192.168.121.108   Running     2019-12-26T14:22:05Z
kube-system            kube-apiserver-master-11                              master-11   192.168.121.108   Running     2019-12-26T14:22:05Z
kube-system            calico-node-sxls8                                     master-11   192.168.121.108   Running     2019-12-26T14:55:41Z
kube-system            calico-kube-controllers-6d85fdfbd8-dnpn4              master-11   192.168.121.108   Running     2019-12-26T14:55:41Z
kubernetes-dashboard   dashboard-metrics-scraper-76585494d8-jx9cg            master-11   192.168.121.108   Running     2019-12-26T16:10:16Z
kubernetes-dashboard   kubernetes-dashboard-5996555fd8-5z5p2                 master-11   192.168.121.108   Running     2019-12-26T16:10:16Z

The documentation for this is https://kubernetes.io/docs/reference/kubectl/overview/#custom-columns

小开

Try this one:

    kubectl get nodes -o=custom-columns=NAME:.metadata.name,TAINTS:.spec.taints

小开

#Check Node Taints
kubectl get nodes -o=custom-columns=NodeName:.metadata.name,TaintKey:.spec.taints[*].key,TaintValue:.spec.taints[*].value,TaintEffect:.spec.taints[*].effect

Let me try and explain what this first one means and then rest should fall in place:

NodeName:.metadata.name

ColumnName: JSONPATH to the attribute you are looking for.

ColumnName can be anything you want it to be.

Something like NodeName:items[*].metadata.name is equivalent to running $kubectl get nodes -o=jsonpath='{.items[*].metadata.name}' but with custom-columns flag you get values in rows and columns format.

Note: You don't need to start with .items[*]. It already parses that with custom-column flag

so now all the columns explained:

NodeName:.metadata.name - Get Node Names and put it under NodeName Column

TaintKey:.spec.taints[*].key - return all the keys for taints by looking under taints map and put it under TaintKey custom-column

TaintValue:.spec.taints[*].value - same as key but you are returning the value from the taints map.

TaintEffect:.spec.taints[*].effect - same as key but you are returning the effect from the taints map.

You set it under and alias like

alias get-nodetaints="kubectl get nodes -o=custom-columns=NodeName:.metadata.name,TaintKey:.spec.taints[*].key,TaintValue:.spec.taints[*].value,TaintEffect:.spec.taints[*].effect"

and you have your own nice command to get all taints and your output should look something like below

sample output for the command

小开

Below commands worked for me:

If you have the node IP, You can try

kubectl get node $node_ip -o json | jq '.spec.taints'
Output:
[
{
"effect": "NoSchedule",
"key": "dedicated"
}
]


(OR)


kubectl describe node $node_ip  | grep -i Taints
Output:
Taints:             dedicated:NoSchedule

To get all Taint config on namespace

kubectl get nodes -o json | jq '.items[].spec.taints'
Output:
[
{
"effect": "NoSchedule",
"key": "dedicated"
}
]

小开

kubectl describe node node01 | grep -i taints