Subin sub-mod

OpenShift Pipelines

https://openshift.github.io/pipelines-docs/docs/docs/index.html
OpenShift Pipelines are cloud-native, continuous integration and delivery solutions based on Kubernetes resources. It uses Tekton Pipelines. https://github.com/openshift/pipelines-tutorial

Install via OperatorHub

Slide from Google : http://files.informatandm.com/uploads/2019/5/Apr_18_Dan_Lorenc_The_Future_of_Cloud_Native_CI_CD.pdf Blog IBM : https://developer.ibm.com/tutorials/knative-build-app-development-with-tekton/

Setup Tekton on GKE

Install Pipelines

Slide from Google : http://files.informatandm.com/uploads/2019/5/Apr_18_Dan_Lorenc_The_Future_of_Cloud_Native_CI_CD.pdf
Blog IBM : https://developer.ibm.com/tutorials/knative-build-app-development-with-tekton/

Openshift Namespace

namespace: tekton-demo  
sa:  tekton-demo

Kubernetes Reflector

Reflector is a key component for Kubernetes clients, kube-scheduler and Replication Controller. Reflector is in the middle between client and Kubernetes API-server. It provides a framework to monitor the changes of the Kubernetes cluster.

Definition of Reflector

Here is the definition of Reflector.

As shown in the definition, there are two important compoents of a Reflector:

Kubernetes SharedInformer

Overview

SharedInformer has a shared data cache, and is able to distribute notifications for changes to the cache to multiple listeners who have registered to it. SharedInformer is also a "Producer-Consumer" framework, using Controller as Producer to produce the Events (, or changes to the cache).

Setup CodeReady Containers -1.14.0 on Remote Server and connect from Laptop.
These steps work for Fedora 31.

Setup On the remote host
Setup On the Laptop

On-the-remote-host

Use cuda:10.0-cudnn7-devel-centos7 container

docker run -it -u 0 docker.io/nvidia/cuda:10.0-cudnn7-devel-centos7 /bin/bash

Setup LD_LIBRARY_PATH and Links

yum install -y centos-release-scl
yum install -y devtoolset-7 rh-python36 mlocate
source scl_source enable devtoolset-7 rh-python36

some thing like this would cause invalid input params: param types don''t match the user-specified type in pipeline webhook controller which simply can't parse the type

---
apiVersion: tekton.dev/v1alpha1
kind: PipelineRun
metadata:
  name: params-example
spec:

	INFO:tensorflow:Restoring parameters from ./models/model.ckpt-500
	I0106 15:55:19.536532 140690754516544 saver.py:1284] Restoring parameters from ./models/model.ckpt-500
	INFO:tensorflow:Running local_init_op.
	I0106 15:55:20.502651 140690754516544 session_manager.py:500] Running local_init_op.
	INFO:tensorflow:Done running local_init_op.
	I0106 15:55:20.614528 140690754516544 session_manager.py:502] Done running local_init_op.
	INFO:tensorflow:Performing evaluation on 22 images.
	I0106 15:55:23.292874 140683104995072 coco_evaluation.py:205] Performing evaluation on 22 images.
	creating index...
	index created!

	// use nvidia/cuda:10.0-cudnn7-devel-centos7 as BASE


	// Ensure below paths are set to access nvcc at $CUDA_HOME/bin
	export CUDA_HOME=/usr/local/cuda
	export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/cuda/lib64:/usr/local/cuda/extras/CUPTI/lib64
	export PATH=$PATH:$CUDA_HOME/bin

	// Install mentions GCC>=4.9. update to devtoolset-7
	// use python 3.6