Recently I’ve had the experience of reconfiguring the popular Kubernetes Service Mesh Istio (using it’s Gateway ingress model) to work with an AWS Application Load Balancer with a degree of automation and scalability. This is a challenging deployment to say the least and whilst documentation exists to varying degrees for the separate components, it’s scant. I’m less than impressed with the official Istio documentation (though it has gotten way better) . . .
Last year I wrote about automating Elastic Kubernetes Service role configuration (direct modification of the aws-auth ConfigMap) using Terraform, and a somewhat clunky method of injecting ARN data by looking it up from a secret management service (in this case Hashicorp Vault). Whilst the solution works well it comes with a serious built in issue when we want to provision a new deployment from scratch, namely the need to import . . .
Recently I’ve been presented with the same question from a couple of readers so I’m going to run through it quickly. A while back I looked at integrating Azure KeyVault with Ansible Tower (a horribly documented scenario in my experience), but I didn’t really cover how to call multiple KeyVault Secrets and assign them to a single Ansible Tower Credential for use in a Playbook. Please take a look at . . .
Ansible is a big favourite of mine as anyone that knows me will tell you and has become one of the biggest players in the DevOps world, inevitably if you’re going to use it at any real scale you’ll need to start thinking about tags. Tags are an essential part of life in the cloud, given the scale and complexity we can encounter they really become the only way to . . .
EDIT: A few days after publishing this article, Hashicorp’s official AWS provider was updated to support default tags directly from the provider (which is very simple and saves all of the work detailed in this article). This only works with AWS so if you’re working in another cloud keep reading on, if you’re only working in AWS take a look at the Hashicorp blog post here which provides some very . . .
In the previous post we looked at how to build Chartmuseum on Ubuntu Linux with an S3 backend, however out of the box this system presents a number of problems; specifically it isn’t TLS encrypted and the service runs on an unprivileged TCP port. I could see no guides suggesting how to do this, so lets take a look at how to solve this problem by performing by proxying our . . .
Helm is an incredibly popular package manager for Kubernetes, however despite it’s incredibly widespread use there isn’t a huge amount of information or options out there for creating private repositories using Open Source platforms. Chartmuseum seeks to solve this problem by offering us just that. In this post I’m looking at how to deploy and bootstrap Chartmuseum on Ubuntu Linux 18.04, using a secure AWS S3 backend. Getting Started Chartmuseum . . .
When I first started using Ansible, querying JSON was a source of constant frustration. Most of the articles I could find on the topic seem particularly interested in a long lesson on the topic of how JSON is structured. Whilst that is important to understand I couldn’t really find a guide that just broke down a few simple queries like I wanted. I’m not even going to attempt to talk . . .
Recently I’ve been working with Ansible in GCP to try and automate the process of provisioning a bare metal Kubernetes cluster. A good find in this process was the Ansible gcp_compute plugin which allows for the construction of Dynamic Inventories based on your existing GCE resources. What Are We Working With? As is often the case with Ansible, the documentation is great but can be confusing without context or to . . .
Recently I’ve been looking AWS’ Elastic File Service platform, which allows for the provisioning of highly available PaaS storage which can accessed via NFS by multiple services at at very low cost. Whilst this is good, what’s even better is templating and automating the provisioning. In this post we’ll look at how to provision HA EFS storage using Terraform. What Do We Want? We have the option to create EFS . . .