In the second part of the DevOps project, the focus is on deploying monitoring tools like ArgoCD, Prometheus, and Grafana to a Kubernetes cluster. The blog post covers installing ArgoCD, deploying Prometheus using Helm charts, setting up monitoring for ArgoCD, visualizing ArgoCD metrics using Grafana dashboards, and continuous deployment of applications using ArgoCD. A useful tool, K8sgpt, is recommended to analyze the cluster for errors and potential issues. The next blog post will discuss configuring Alert Manager for notifications, setting up Slack alerts, and installing Loki for logs, enhancing the monitoring solution.

https://blog.devgenius.io/optimizing-kubernetes-deployments-with-argocd-and-prometheus-aa86c11e2bba

Don't forget about security

https://dzone.com/articles/container-security-top-5-best-practices-for-devops

A new terraform version has been released. Import already existed infrastructure to the terraform state become easier.
https://www.hashicorp.com/blog/terraform-1-5-brings-config-driven-import-and-checks

Streaming alert evaluation offers better scalability than traditional polling time-series databases, overcoming high dimensionality/cardinality limitations. This enables engineers to have more reliable and real-time alerting systems. The transition to the streaming path has opened doors for supporting more exciting use-cases and has allowed multiple platform teams at Netflix to generate and maintain alerts programmatically without affecting other users. The streaming paradigm may help tackle correlation problems in observability and offer new opportunities for metrics and events verticals, such as logs and traces.

https://netflixtechblog.com/improved-alerting-with-atlas-streaming-eval-e691c60dc61e

In this post, the author discusses potential PostgreSQL pitfalls that may not affect small databases, but can cause issues when databases grow.
https://philbooth.me/blog/nine-ways-to-shoot-yourself-in-the-foot-with-postgresql

Pipedrive Infra manages numerous Kubernetes clusters across different clouds, including AWS and on-premise OpenStack. They had been experiencing intermittent failing pod health checks, which became more frequent over time. After an extensive investigation, the team discovered that Kubelet was initiating TCP sessions to pods using random source ports within the same range reserved by Kubernetes nodeports. This caused the TCP SYN-ACK to be redirected to other pods, leading to failed health checks. The solution was to disallow the use of the nodeport range as the source port for TCP sessions with a single line of code, effectively resolving the issue.

https://medium.com/pipedrive-engineering/solving-the-mystery-of-pods-health-checks-failures-in-kubernetes-55b375493d03

Efficient GPU utilization is crucial for minimizing infrastructure expenses, especially in large Kubernetes clusters running AI and HPC workloads. NVIDIA MIG enables partitioning GPUs into smaller slices, but using MIG in Kubernetes through the NVIDIA GPU Operator alone has limitations due to static configurations. Dynamic MIG Partitioning addresses these limitations by automating the creation and deletion of MIG profiles based on real-time workload requirements, ensuring optimal GPU utilization. The nos module works alongside the NVIDIA GPU Operator to implement dynamic MIG partitioning, simplifying the management of MIG configurations and reducing operational costs.

https://towardsdatascience.com/dynamic-mig-partitioning-in-kubernetes-89db6cdde7a3

Roadmapper - A Roadmap as Code (Rac) python library. Generate professional roadmap diagram using python code.

https://github.com/csgoh/roadmapper

Open-Source Tracing Platform


https://github.com/teletrace/teletrace

rustic - fast, encrypted, deduplicated backups powered by Rust

https://github.com/rustic-rs/rustic

Have you ever heard that company migrate from microservice architecture to monolith?
Moving our service to a monolith reduced our infrastructure cost by over 90%. It also increased our scaling capabilities. Today, we’re able to handle thousands of streams and we still have capacity to scale the service even further. Moving the solution to Amazon EC2 and Amazon ECS also allowed us to use the Amazon EC2 compute saving plans that will help drive costs down even further.
https://www.primevideotech.com/video-streaming/scaling-up-the-prime-video-audio-video-monitoring-service-and-reducing-costs-by-90