hkubectl

Install

curl -Lo hkubectl https://github.com/kube-HPC/hkubectl/releases/latest/download/hkubectl-linux \
&& chmod +x hkubectl \
&& sudo mv hkubectl /usr/local/bin/  

Usage

hkubectl

---

$ hkubectl [ command ]  

Options:

option	description	type	required	default
--version	Show version number	boolean
--rejectUnauthorized	set to false to ignore certificate signing errors. Useful for self signed TLS certificate	boolean
--endpoint	url of hkube api endpoint	string
--pathPrefix	path prefix url of hkube api endpoint	string		/hkube/api-server/
--verbose	verbose logging	boolean
-j, --json	output json to stdout	boolean
--help	Show help	boolean

exec

---

$ hkubectl exec < command >  

Execution pipelines as raw or stored

get

$ hkubectl exec get < jobId >  

Returns the executed pipeline data
Options:

option	description	type	required	default
jobId	The jobId to get the result	string	true

raw

$ hkubectl exec raw   

execute raw pipeline from file
Options:

option	description	type	required	default
-f, --file	file path/name for running pipeline. use - to read from stdin	string	true
--noWait	if true, does not wait for the execution to finish	boolean		false
--noResult	if true, does not show the result of the execution	boolean		false

stored

$ hkubectl exec stored [ name ]  

execute pipeline by name
Options:

option	description	type	required	default
name	The name of the algorithm	string
-f, --file	file path/name for running pipeline	string
--noWait	if true, does not wait for the execution to finish	boolean		false
--noResult	if true, does not show the result of the execution	boolean		false

stop

$ hkubectl exec stop   

call to stop pipeline execution
Options:

option	description	type	required	default
jobId	The jobId to get the result	string	true
reason	Reason for stopping the pipeline	string

status

$ hkubectl exec status < jobId >  

Returns a status for the current pipeline
Options:

option	description	type	required	default
jobId	The jobId to get the result	string	true

result

$ hkubectl exec result < jobId >  

returns result for the execution of a specific pipeline run
Options:

option	description	type	required	default
jobId	The jobId to get the result	string	true

algorithm

$ hkubectl exec algorithm [ name ]  

execute algorithm
Options:

option	description	type	required	default
name	The name of the algorithm	string
-f, --file	file path/name for extra data	string
--noWait	if true, does not wait for the execution to finish	boolean		false
--noResult	if true, does not show the result of the execution	boolean		false

algorithm

---

$ hkubectl algorithm < command >  

Manage loaded algorithms

apply

$ hkubectl algorithm apply [ name ]  

apply an algorithm
Options:

option	description	type	required	default
name	The name of the algorithm	string
-f, --file	the algorithm file	string
--env	the algorithm env [choices: "python", "nodejs", "java"]	string
--codePath	the code path for the algorithm	string
--codeEntryPoint, --entryPoint	the code entry point for the algorithm	string
--image, --algorithmImage	set algorithm image	string
--cpu	CPU requirements of the algorithm in cores	number
--gpu	GPU requirements of the algorithm in cores	number
--mem	memory requirements of the algorithm. Possible units are ['Mi', 'Gi']. Minimum is 4Mi	string
--noWait	if true, does not wait for the build to finish	boolean		false
--setCurrent	if true, sets the new version as the current version	boolean		false

list

$ hkubectl algorithm list   

Lists all registered algorithms

get

$ hkubectl algorithm get < name >  

Gets an algorithm by name
Options:

option	description	type	required	default
name	The name of the algorithm	string	true

delete

$ hkubectl algorithm delete < name >  

Deletes an algorithm by name
Options:

option	description	type	required	default
name	The name of the algorithm	string	true

version

$ hkubectl algorithm version < name >  

Gets versions of algorithm
Options:

option	description	type	required	default
name	The name of the algorithm	string	true
--setCurrent, --set	Sets the current version	string
--force	If true forces the change of the version (might stop running pipelines)	boolean

template

$ hkubectl algorithm template [ name ]  

Create algorithm template for builds
Options:

option	description	type	required	default
name	The name of the algorithm	string
--codePath	the code path for the algorithm [required]	string	true
--codeEntryPoint, --entryPoint	the code entry point for the algorithm	string		main
--env	the algorithm env [required] [choices: "python", "nodejs"]	string	true
--overwrite	overwrite an existing folder	boolean
--cpu	CPU requirements of the algorithm in cores	number		0.1
--gpu	GPU requirements of the algorithm in cores	number		0
--mem	memory requirements of the algorithm. Possible units are ['Ki', 'M', 'Mi', 'Gi', 'm', 'K', 'G', 'T', 'Ti', 'P', 'Pi', 'E', 'Ei']. Minimum is 4Mi	string		512Mi

pipeline

---

$ hkubectl pipeline < command >  

Manage loaded algorithms

get

$ hkubectl pipeline get [ name ]  

Gets an pipeline by name
Options:

option	description	type	required	default
name	The name of the algorithm	string

store

$ hkubectl pipeline store   

Store pipeline
Options:

option	description	type	required	default
-f, --file	path for descriptor file	string	true
--readmeFile	path for readme file. example: --readmeFile="./readme.md	string

sync

---

$ hkubectl sync < command >  

sync local source folder into algorithm container in the cluster

watch

$ hkubectl sync watch   

watch a local folder
Options:

option	description	type	required	default
-a, --algorithmName	The name of the algorithm to sync data into [required]	string	true
-f, --folder	local folder to sync.	string		./
--bidirectional, --bidi	Sync files in both ways	boolean		false

create

$ hkubectl sync create   

creates the algorithm for development.
Options:

option	description	type	required	default
-a, --algorithmName	The name of the algorithm	string	true
-f, --folder	local folder to build from.	string		./
--env	algorithm runtime environment [choices: "python", "nodejs"]	string
-e, --entryPoint	the main file of the algorithm	string
--baseImage	base image for the algorithm	string

config

---

$ hkubectl config [ command ]  

Set configuration options for hkubectl

set

$ hkubectl config set   

Sets configuration options.

get

$ hkubectl config get   

Gets the current configuration.

datasource

---

$ hkubectl datasource < command >  

Execution pipelines as raw or stored

sync

$ hkubectl datasource sync   

should be called after push updates to git, creates a new version entry on the datasource service

prepare

$ hkubectl datasource prepare   

should be called before commiting to git, scans the directory and updates all the.dvc files

push

$ hkubectl datasource push   

calls dvc, git push and hkube datasource hkube datasource prepare and sync