Rancher-nfs equivalent in Rancher 2.0?

I’m testing out Rancher 2.0 and the first thing I need is to connect to my persistant storage.

It doesn’t look like there is anything like rancher-nfs in the standard catalog, nfs-provisioner seems to be only to create a new NFS server in a container to provision persistant storage off of it.

It seems that what I want is nfs-client, from incubator (https://github.com/kubernetes-incubator/external-storage/tree/master/nfs-client), but I couldn’t find a way to deploy it from the web interface. I can probably do it manually but really don’t see the point of Rancher 2.0 if you start off CLI’ing basic stuff.

Now it’s GA, I’m confused, is there a roadmap for actual feature parity, or am I missing something?

1 Like

I am looking for the same, Completely stuck right now

There’s no catalog item for it because it’s a built in volume driver in k8s. You create an (ephemeral) volume or persistent volume and point it at a server and path.
Volumes | Kubernetes

What this does not do is dynamically create a subdirectory of that NFS server and give the pod only access to that subdirectory. That’s what external-storage/nfs-client adds.
external-storage/nfs creates NFS servers, as you said.

Thanks for your help @vincent. I missed this.
On that note however I’m trying to setup nfs server using nfs-provisioner from the incubator. I have been trying for several days now. No luck. It just says ‘pending’ all the time when I try to create a volume. I dont see any errors when I inspect the deployment. I think I’m not setting this up correctly. I would appreciate if you could point me in the right direction. My setup is just bare machines

1 Like

So anyone having success deploying nfs-client ? Looks like that’s what replaces rancher-nfs. I’m wondering where to start to deploy what’s in the source tree.

Not really working for me, I added the persistent volume, and then a workload with said volume. ANd it just keeps timing out on Unavailable Containers with unready status

I’m using the k8s NFS client persistent volume with efs and no problem at all. Just make sure to have the options set and add the volume claim in the correct namespace. From that point the workloads in the same namespace have access. You do need to make sure the folder exist.

Thank you, the whole point of rancher-nfs though, was that you pointed it to the upper level directory and it auto provisioned directories for containers as you created it.

I’m not familiar with manually installing things like storage providers but probably once nfs-client is installed it should work the way I expect it to work.

I think if you want that same type of behavior you have to look at storage classes. Although I haven’t seen any of them for nfs :confused: it does have gluster and others that would work similarly.

looking for the same thing just like that i have use in 1.6 hmmm

hope rancher labs will add rancher-nfs in rancher 2.0 and other networking features from 1.6 library

i’ve tried launching nfs-client-provisioner and nfs-server-provisioner the nfs-client keeps on installing in status and then nfs-server keeps on deploying in status

error pops up in nfs-client-provisioner

Kubectl apply failed. Error: error: error validating "app-template": error validating data: ValidationError(Deployment.spec.template.spec.volumes[0].nfs): missing required field "server" in io.k8s.api.core.v1.NFSVolumeSource; if you choose to ignore these errors, turn validation off with --validate=false : exit status 1

and in nfs-server-provisioner saids active and unavailable…

ImagePullBackOff: Back-off pulling image "quay.io/kubernetes_incubator/nfs-provisioner:v1.0.9"