README.md

Command Guide and Reference Manual

kubectl apply

• kubectl apply -f [filename]:

  If you want to create api object in a declarative way, you can use this command.

  For example, if you have a pod template yaml file called pod.yaml in current directory, then you can type kubectl apply -f ./pod.yaml to create a pod according to your specified template.

kubectl get

• kubectl get pod [pod name]

  This command will show the status of the given pod in a table.

  For example, if you have a pod called example in the default namespace, then you can check its status by kubectl get pod example(since it's in the default namespace, so the namespace can be omitted. Otherwise, you must type namespace/name to specify a pod).

  Example

  

• kubectl get pods

  This command will show the status of all pods(in all namespaces) in a table.

  Example

  

• kubectl get node [node name]

  This command will show the status of the given node in a table.

  Example

  

• kubectl get nodes

  This command will show the status of all nodes(in all namespaces) in a table.

  Example

  

• kubectl get rs [replicaSet name]

  This command will show the status of the given replicaSet in a table.

  Example

  

• kubectl get rss

  This command will show the status of all replicaSets(in all namespaces) in a table.

  Example

  

• kubectl get hpa [hpa name]

  This command will show the status of the given horizontal pod autoscaler in a table.

  Example

  

• kubectl get hpas

  This command will show the status of all horizontal pod autoscalers(in all namespaces) in a table.

  Example

  

• kubectl get wf [workflow name]

  This command will show the status of the given workflow in a table.

  Example

  

• kubectl get wfs

  This command will show the status of all workflows(in all namespaces) in a table.

  Example

  

• kubectl get func [function name]

  This command will show the status of the given function in a table.

  Example

  

• kubectl get funcs

  This command will show the status of all functions in a table.

  Example

  

• kubectl get gpu [gpu job name]

  This command will show the status of the given gpu job in a table.

  Example

  

• kubectl get gpus

  This command will show the status of all gpu jobs in a table.

  Example

  

• kubectl get service [service name]

  This command will show the status of the given service in a table.

• kubectl get services

  This command will show the status of all services in a table.

• kubectl get dns [dns name]

  This command will show the status of the given dns in a table.

• kubectl get dnses

  This command will show the status of all dnses in a table.

kubectl delete

• kubectl delete [api object type] [name]
  Example

  

kubectl autoscale

• kubectl autoscale [hpa name]

Parameter	Description	Example
target	Required. Specify the target of hpa	--target=rs1
min	Optional, default value is 1. Specify the minimum number of replicas	--min=1
max	Optional, default value is 1. Specify the maximum number of replicas	--max=4
c	Optional. Choose cpu utilization as metrics, and the given value as its benchmark	-c 0.5
m	Optional. Choose memory utilization as metrics, and the given value as its benchmark	-m 0.5
i	Optional, default value is 15. Specify the interval of scaling.	-i 30

This command will create a hpa according to given parameters. Notice that if you specify both cpu and memory metrics, the hpa controller will choose cpu(cpu has higher priority).

You can also create a hpa in a declarative way: kubectl apply -f hpa.yaml.

We recommend you to use kubectl autoscale command, for it is more obvious.

Example

kubectl reset

This command is for test only(of course you can also feel free to use it). It will remove all the K-V pairs stored in etcd, thus resetting the status of the whole system.

kubectl gpu

• kubectl gpu [gpu job name] -d [directory] -f [file to download]

  This command will list or download the files throw the nginx-fileserver. Files in blue colors are directories.

  -d flag stands for the directory to hold the download files(only works when flag -f is used).

  -f flag stands for the file you want to download. It will be downloaded through GET http request into the directory you have specified in flag -d.

  If you don't use any flags, this command will list the files, just like ls command.

  List files example

  
  Download files example

  You can see that the file matrix-op.out has been successfully downloaded to the current directory.

  

kubectl func

• kubectl func add -f [function name] -p [function code path]

  This command will register a function to the knative.

  -f flag stands for the name of the function.

  -p flag stands for the filepath of the code.

  Example

  

• kubectl func rm -f [function name]

  This command will remove the function and its instances(pods and replicaSet).

  Example

  

• kubectl func update -f [function name] -p [function code path]

  This command will update a function to the knative.

kubectl trigger

• kubectl trigger [function name] -d [function params json]

  This command will call a serverless function through http trigger.

  -d flag stands for the json form of parameters the function needs.

  Example

  

kubectl wf

• kubectl wf apply -f [workflow.json filepath]

  This command will create a workflow in a declarative way.

  -f flag stands for the ${workflow_name}.json filepath.

  Example

  

• kubectl wf rm [workflow name]

  This command will remove the specified workflow.

  Example

  

kubectl label

• kubectl label nodes [node name] [label]

  This command can label a node with given labels.

  Example

  

kubectl cfg

• kubectl cfg sched=[schedule strategy]

  This command can dynamically change the schedule strategy.

strategy	Description
min-pods	Schedule pod to the node with minimum number of pods
max-pods	Schedule pod to the node with maximum number of pods
min-cpu	Schedule pod to the node with minimum cpu utilization
min-mem	Schedule pod to the node with minimum memory utilization
random	Schedule pod to a random node