Nick Brandaleone nbrandaleone

Docs

curl https://raw.githubusercontent.com/aws-samples/amazon-cloudwatch-container-insights/master/k8s-yaml-templates/quickstart/cwagent-fluentd-quickstart.yaml | sed "s/{{cluster_name}}/cluster-name/;s/{{region_name}}/cluster-region/" | kubectl apply -f -

Replace the "cluster-name", and "cluster-region" with proper info.

Delete Container Insights: curl https://raw.githubusercontent.com/aws-samples/amazon-cloudwatch-container-insights/master/k8s-yaml-templates/quickstart/cwagent-fluentd-quickstart.yaml | sed "s/{{cluster_name}}/cluster-name/;s/{{region_name}}/cluster-region/" | kubectl delete -f -

Setup

Add this to your .profile, .bashrc, .zshrc...

decode_base64_url() {
  local len=$((${#1} % 4))
  local result="$1"
  if [ $len -eq 2 ]; then result="$1"'=='
  elif [ $len -eq 3 ]; then result="$1"'=' 
  fi
 echo "$result" | tr '_-' '/+' | openssl enc -d -base64

This gist shows how to setup the new preview feature called Firelens, which runs on Amazon ECS. Detailed instructions on the product is located here.

Firelens is a fluentbit log shipper, which is extremely light-weight in terms of memory footprint. It has been altered to support AWS CloudWatch and AWS Firehose, along with its traditional destinations. Firelens operates as a "side-car" container, enriches the logs, and then ships them off to its destination. The side-car can operate on 25 MB of RAM.

Firelens can work with ECS / Fargate. I have not tested it with Kubernetes, so YMMV.

Here is an example of an enriched log:

	The only necessary role is the Container Instance IAM role. This role allows the ECS agent (running on your EC2 instance) to communicate with Amazon ECS.

	There are five other roles that you may also find useful, for different purposes:

	ECS Service-Linked role (SLR) - This role enables Amazon ECS to manage a variety of AWS resources associated with your application on your behalf. When using a Service, this role allows Amazon ECS to manage the load balancer (Classic Load Balancers, Application Load Balancers, and Network Load Balancers) and service discovery (with Route 53) associated with your service. When using task networking, this role allows Amazon ECS to attach and detach Elastic Network Interfaces (ENIs) to your tasks. This role is required when using AWS Fargate.
	Service Scheduler IAM role - Prior to the introduction of the ECS Service-Linked role (SLR), this role was used in conjunction with a Service to enable Amazon ECS to manage the load balancer associated with your service. If you want

	I think I can provide an inaccurate but short version of the relationships among them.

	At first Clojure used Maven for building.
	Maven is boring, people made Leiningen, configuring with maps, and the plugin lein-cljsbuild can build ClojureScript.
	Leiningen contains only configurations, no logics. People want to compose tasks, so they made Boot, and provided boot-cljs and boot-reload for ClojureScript.
	Webpack is cool with hot module replacement, we got lein-figwheel for that.
	All tools above relies on JVM heavily, in order to run ClojureScript on devices without JVM, like iOS, Planck was created based on self-hosted ClojureScript and JavaScriptCode, which is a ClojureScript runtime.
	Planck does not support modules from npm, then Lumo was created to make full use of V8 engine.
	We still feel the toolchain is too heavy for developers from JavaScript ecosystem, shadow-cljs was refactored to be more friendly, supporting npm modules, relies on JVM but in the background, also with some f

	#!/usr/bin/env bash

	# Nick Brandaleone - September 2018
	#
	# Based upon a script by: https://github.com/scottslowe/scripts/blob/master/kcs
	# I found that the copying of config files to be unreliable, so I simply
	# echo the proper command that needs to be executed by the user.

	# Set some variables that will be used later
	KUBEDIR="$HOME/.kube"

	(ns ttt-mcts.core
	(:gen-class))

	(comment
	This program implements a monte carlo search algorithm to play tic-tac-toe.
	)
	;The "MCTS" algorithm has four steps
	;1) Randomly pick a branch of the tree (in practice, the more
	; successful branches are usually weighted to be chosen more).
	;2) Expand the branch (create a node).

	(ns ttt-minmax.core
	(:gen-class))

	(comment
	This program implements a min-max algorithm to play tic-tac-toe.

	The program generates all possible moves by examining the empty slots in the playing board.
	It creates a tree, growing out from the current board position. It assigns a number to each leaf node
	based upon the outcome of the game. It propogates this back up to the current board position, and
	makes the move which maximizes its 'utility' function, and minimizes that of its oponent.

	//
	// ViewController.swift
	// SwiftLife
	//
	// Created by Nick Brandaleone on 6/1/15.
	// Copyright (c) 2015 Nick Brandaleone. All rights reserved.
	//

	import UIKit

	;;
	;; Source code originally from "Clojure Programming" (2012) by Chas Emerick, Brian Carper and Christophe Grand
	;;
	(ns gol.core
	(:gen-class))

	(defn neighbors
	"find our cell neighbors"
	[[x y]]
	(for [dx [-1 0 1] dy [-1 0 1] :when (not= 0 dx dy)]