Nfs

juju deploy nfs

18.04 LTS 16.04 LTS 14.04 LTS

Discuss this charm

Share your thoughts on this charm with the community on discourse.

Join the discussion

Overview

This is the NFS charm with support for trusty, xenial, and bionic. It creates an NFSv4(http://nfs.sourceforge.net/) server using the local filesystem as a backing store. There is no clustering or federation, so multiple units would result in multiple nfs servers. This is NOT a highly available server.

Usage

To deploy an NFS server:

juju deploy nfs

Each relation to the NFS charm will result in a new export for the units in that application. The default location of data is /srv/data

Note that the NFS charm will NOT destroy data. This means if you remove a relation to an application, you will need to manually destroy the data saved by that application in the local data location.

Kubernetes
conjure-up canonical-kubernetes
juju deploy nfs
juju add-relation nfs kubernetes-worker
Owncloud
juju deploy nfs
juju deploy mysql
juju deploy owncloud
juju add-relation mysql:owncloud
juju add-relation nfs:nfs owncloud:shared-fs

The above example deploys OwnCloud personal cloud storage, and provides remote storage via the NFS host.

Wordpress
juju deploy nfs
juju deploy mysql
juju deploy wordpress
juju add-relation mysql:db wordpress:db
juju add-relation nfs:nfs wordpress:nfs

Migrating Storage

To migrate storage from one NFS unit to another, first add the new unit in such a way as to avoid publishing it before it's ready:

juju config nfs active_units=<old unit ID>
juju add-unit nfs

Now start the downtime:

juju config nfs active_units=none

Wait for all clients to unmount, then move the underlying storage to the new unit in whatever way is appropriate for your deployment.

Finish the downtime by publishing the new unit:

juju config nfs active_units=<new unit ID>

After clients have mounted the new unit and you've checked that all is well, you can remove the old unit:

juju remove-unit <old unit ID>
juju config nfs active_units=

Note that the migration process is quite abrupt: the server does not wait for all clients on related units to unmount before stopping. Consider arranging separately for downtime of clients that might write to their NFS mounts.

Known Limitations and Issues

No high availability story

At present the charms consuming an NFS relationship only account for a single host. Most charms assume the first incoming NFS mount-point is the sole replacement, and subsequent NFS relationship-join requests are ignored.

If you are attempting to deploy NFS to an LXC container, such as the juju local provider, there are additional steps that need to be taken prior to deploying the NFS charm.

On the LXC host:

apt-get install nfs-common
modprobe nfsd
mount -t nfsd nfsd /proc/fs/nfsd

Edit /etc/apparmor.d/lxc/lxc-default and add the following three lines to it:

mount fstype=nfs,
mount fstype=nfs4,
mount fstype=nfsd,
mount fstype=rpc_pipefs,

after which:

sudo /etc/init.d/apparmor restart

Finally:

juju deploy nfs

Configuration

  • storage_root: The root path where exported directories will be created
  • export_options: The default export options. Ships with rw,sync,no_root_squash,no_all_squash
  • mount_options: The default client mount options

Contact Information

Mike Wilson mike.wilson@canonical.com

Upstream NFS Project