Why stress test At Manabie, we have already implemented bunch of features for the business, mostly to support ERP in education domain. The traffic on production is not that high. For each cluster (of each clients), it barely reaches 100 rpc on our Grafana dashboard, and everything looks safe. But no, we realize that in our roadmap, we have a plan to merge all the cluster into a multi-tenant cluster. We don’t know if the cluster merge will affect some resources, maybe there are extensively used features that we will not be aware of (features team are rapidly adding more APIs, as the business grows) and we may encouter performance issues unprepared.
Read more →This blog post focuses on how we can create a reusable workflow for GitHub Actions. The do’s and don’ts will be elaborated along the way.
1. What are the available options? There are two major types of workflows that can be reused in Github Actions:
Custom actions Reusable workflows Reusable workflows are quite new. In this blog, we decide to use custom actions (or more specifically, composite actions). The main reason is because we want to run the workflow inside a job, along with other steps (to share caches, tokens, etc…).
Read more →In this blog post, we will learn how to self-host Unleash in a Kubernetes cluster.
Why feature toggles? While developing new features for our end-users, we often encounter these 2 problems:
Features are not developed and rolled out in a single night. It often takes several days or even weeks before a feature is fully developed, tested, and deployed. In such cases, we usually have to deploy a piece of that feature to production, and we would need to hide that piece until the feature is fully completed and ready.
Read more →