I'm setting up RabbitMQ in my Kubernetes cluster and considering whether it's a good idea to use a shared NFS volume for persistent queues across different nodes. Is this a smart choice for a production environment?
3 Answers
It's not advisable to use NFS for RabbitMQ. Each node should have its own storage to ensure high availability and proper functioning of the cluster. Using NFS could lead to issues since it’s not designed for this kind of workload.
Generally, NFS isn’t recommended for Kubernetes workloads. If you still want to go down that route, you can look into the NFS CSI provider, which might make it easier to manage. Check out their documentation for guidance.
Sure! The NFS CSI provider allows you to mount NFS shares as volumes in your Pods, which can help if you're really in a pinch.
NFS tends to be a quick fix for legacy systems rather than a solid solution. You might want to rethink your architecture. If you really need to share storage, consider using advanced methods like caching solutions or data distribution strategies.
Could you explain how the NFS CSI provider works? I'm not familiar with it.