imranismail

Dynamic Elixir Node Discovery on Kubernetes

One way one would do it i to connect nodes together by having the sys.config/vm.args as suggested by chrismccord here

However, when deploying to platform such as Kubernetes or AWS ElasticBeanstalk whereby the cluster is elastic and the IPs that are attached to the nodes are ephemeral, you probably don't want to update sys.config/vm.args everytime you do a deployment now would you.

Now, to solve this one would need to use a Discovery Service such as Consul/etcd/Zookeeper, I opt to use Kubernetes as it's one of the best PaaS for containerized apps available right now and also has a RESTful API for querying resources such as node IPs.

How it's done is that the Iris.Kubernetes process will poll the API endpoint every 5 seconds for new IPs that are available in the cluster and attempt a connection.

imranismail

Batch No More, Stream Galore: Zero ETL Pipelines from Day 0

In the early stages, many startups rely on traditional batch-processing systems that perform adequately when data volumes are low. However, as organizations grow, these batch pipelines often become inefficient - resulting in slow processing times and complex troubleshooting when issues arise. In many cases, organizations face debates over selecting the optimal batch processing tool, which can consume valuable resources without resolving the underlying performance challenges. Moreover, when failures occur, the process of tracking down issues within a complex network of scheduled jobs can be arduous, sometimes necessitating the reprocessing of entire data batches and causing operational delays.

Recent advancements in the field of Extract, Transform, Load (ETL) have led to the emergence of Zero-ETL strategies. With these developments, streaming data processing is increasingly recognized as the preferred approach from the outset, rather than as a lat

imranismail

Motivation

The terraform helm provider is rather noisy printing unnecessary metadata in the plan output making it hard to review changes. This awk script redacts unnecessary field from the output to make it easier to review.

Usage

terraform plan -out plan.tfplan | awk -f supress_helm_release.awk
terraform show -no-color plan.tfplan | awk -f supress_helm_release.awk

imranismail

//SearchController.php
public function autocomplete(){
	$term = Input::get('term');
	
	$results = array();
	
	$queries = DB::table('users')
		->where('first_name', 'LIKE', '%'.$term.'%')
		->orWhere('last_name', 'LIKE', '%'.$term.'%')
		->take(5)->get();

imranismail

const FORMAT_DISTANCE_LOCALE = {
  lessThanXSeconds: {
    one: "'less than a second'",
    other: "'less than 's' seconds'"
  },

  xSeconds: {
    one: "s' second'",
    other: "s' seconds'"
  },

imranismail

flowchart BT
    Jedan[Jedan\nChristmas] -->|KomentarA| Dva[Neštonešto]
    Dva --> Tri{Let me think}
    Tri -->|Broj4| Četiri[LeKomentar]
    Tri -->|PremaDva| Pet[Tekst5]
    Tri -->|Tri| Šest[fa:fa-car Car]

Pet[<img src='https://iconscout.com/ms-icon-310x310.png' width='40' /> \nasd]

imranismail

apiVersion: apps/v1
kind: Deployment
metadata:
  name: myapp
spec:
  selector:
    matchLabels:
      app: myapp
  template:
    metadata:

imranismail

defmodule Queue do
  def insert(queue, item), do: :queue.in(item, queue)

  def insert_last(queue, item), do: :queue.in_r(item, queue)

  def member?(queue, item), do: :queue.member(item, queue)

  def filter(queue, fun), do: :queue.filter(fun, queue)

  def split(queue, n) do

imranismail

#! /usr/bin/env bash

# This script is used to patch all argocd apps in a cluster
# the patch will add a new annotation to all apps
# the annotation will be used to prevent the app from being refreshed unnecessarily
# the following annotation will be added to all apps
# argocd.argoproj.io/manifest-generate-paths: .

set -euo pipefail

imranismail

defmodule Web.InvalidParamsError do
  defexception [:changeset, plug_status: 422]

  def message(value) do
    """
    Invalid parameters

      #{inspect(error_messages(value.changeset))}
    """
  end