NFS supports a timeout, so when the remote storage isn't available, then it will only block for a short period. Also, you can use the automounter, not a full-time fstab mount with NFS.
If you have a 100 Mbps network or faster, forcing all NFS to use TCP would reduce the chance of data corruption.
autofs is the older method to use the client-side automounter. It has well-used, well-understood, setup but is more complex than the newer, systemd-automount method available in 16.04 and later. An example:
https://ubuntuforums.org/showthread....6#post13935156
I've been using autofs for a very long time for NFS, CIFS, and USB storage. Basically, any storage that isn't always connected. I wanted control over the mount options and especially the locations. I've been badly burned by the /media/ use, plus permanent mounts aren't supposed to go there according to the file system hierarchy standards.
Discovered the systemd-automount capability just a few months ago. Been using it a few months. Extremely solid and performs the same as autofs or simple fstab mounts. No issues with it at all. I'll migrate off autofs completely. This is all client-side config. The NFS server still uses the /etc/exports file in the normal way. My NFS servers never go off-line unless I take them down or there's a HW failure.
Guess we've beaten this topic. If you think it is SOLVED, please use the Thread Tools and mark it so.
Bookmarks