I started playing with convoy NFS to solve a challenge I’m facing with moving some of our environments over to docker. This is all within the context of Rancher.
I notice that, when I create a new convoy-nfs stack in Rancher, each volume I create within this new driver creates a directory with that name. I realized that I can create a volume with a name where the directory already exists. However, I am concerned about convoy-nfs wanting to manage this top level directory.
I have several systems which have to access data stored on NFS. And this NFS directory must never ever be removed, this would be extremely bad. But applications need to be able to have RW access to the directory.
The basic question:
Can I not use convoy-nfs to point to an existing directory and mount this into a container without convoy-nfs wanting to “completely take it over”?
For example, imagine my NFS server exporting the following: /var/storage/somedata
I want this “somedata” NFS folder to be accessible to a container. The way I understand this so far means, I have to change my export completely in that I now export “/var/storage” instead of “/var/storage/somedata”. Then, when creating a container, I create a volume called “somedata”.
This would work, but I am still concerned that if anyone ever removed/purged this volume, convoy-nfs would attempt to remove the “somedata” directory.
Is there no way where I can use convoy-nfs (within Rancher) to simply point to an existing directory without having to change my exports on the NFS server to export one level higher (which may not always be an option on our storage array).