Hey everyone! I'm part of a large company currently relying on traditional "Failover Clusters in Windows" and we're looking to transition to Kubernetes. However, we're completely new to it and have a few questions about implementing GitOps for our infrastructure. We specifically want to discuss this in the context of BareMetal or OnPrem VMs. Here are our questions: 1. How do you handle GitOps for things like defining a metrics server? 2. Is TalosOS the right choice for our on-prem setup? 3. When it comes to local storage for SQL Server, should we go with SMB, NFS, or other options? 4. We're also concerned about backups and quick recovery in case of disasters. How do you ensure safety in that regard? Looking forward to your insights! Thanks!
3 Answers
For GitOps in Kubernetes, you have two main tools to consider: ArgoCD and FluxCD. Both have their benefits and will help maintain a reconciliation loop, ensuring that what’s running in your cluster matches what’s defined in your Git repositories. For your questions, the metrics server setup would be part of your deployment process. You could structure it in different waves or with dependency modeling based on the tool you choose. Both Argo and Flux work well with Talos, or any other similar cluster setup. In terms of storage, think about getting fast, local storage or setting up Ceph for shared block storage through a decent network. As for backups, it can get tricky with stateful services, but automating the backup process is definitely achievable with the right setup! Good luck!
You might want to check out Rancher. It's got a free tier, and when combined with Harvester, it can manage everything for you, including quick cluster deployments. They also provide built-in cluster storage through Longhorn and GitOps tools through Fleet, which could cover a lot for your needs!
I've heard of Longhorn as a distributed file system. Is it straightforward to set up? And what exactly does Fleet do?
I totally relate to your situation! We also considered TalosOS but found that Flatcar Linux fits our needs better. Especially if you’re using Hyper-V VMs, Flatcar just worked smoother for us. For storage, we’re opting for IBM CSI for direct fiber channel storage access. Regarding backups, we're adding Veeam Kasten to our stack to enhance that capability. If you have more questions, let me know!
Thanks for the tips! I’m definitely curious about the differences between TalosOS and Flatcar. What are the quirks you experienced with Talos?
Thanks for the detailed explanation! Between ArgoCD and FluxCD, which one would you personally recommend for a team just starting out?