Convoy-nfs in production and host tags

Hello community

I have two seemingly small questions:

  1. Does anyone successfully use Convoy NFS via rancher in a production high throughput environment?

    We are slowly moving our old infrastructure to containers and are (unfortunately) very reliant on NFS for shared object storage.

  2. Can I limit where the system containers are deployed and where volume groups are accessible via host tags?

    As Convoy-NFS (as deployed via the catalog) runs as a system service, it is started on all my hosts. I’d like to limit the hosts it runs on.
    Also them the same for Volumes I then create - can I limit which hosts in the storage pool this volume will be available on?

Samuel

1 Like

I am still struggling with item 2, but may have a solution.

First more background: I have a possible situation where I have several hundred hosts in my rancher environment, and around 6 NFS mounts. However, only a handful of my hosts actually need to access this NFS mount.

If I leave the default options, I’ll end up with 6 x 300 extra containers which I blatantly do not need.

Would the solution be to build my own docker-compose/rancher-compose file where I only make the one simple change of adding the following extra label to the convoy-nfs service?

io.rancher.scheduler.affinity:host_label=mylabelforthisservice

Leading me to ask: Would it be difficult to put this host_label into the answer file for the default Convoy-NFS catalog item?

1 Like