Compatibility with fstab

[dannycjones]

Some users of file connectors may rely on fstab to manage mounting file systems in their Linux operating system.

More info on fstab: https://www.redhat.com/sysadmin/etc-fstab

[dannycjones]

I gave this a go just now; I'm getting the following error as the connector doesn't accept arguments via -o yet. (I also dump the args before parsing.)

ubuntu@ip-172-31-37-193:~$ sudo mount /mnt/s3-file-connector
["s3-file-connector", "djonesoa-eu-west-1", "/mnt/s3-file-connector", "-o", "rw,endpoint_url=https://s3.eu-west-1.amazonaws.com,dev,suid"]
error: Found argument '-o' which wasn't expected, or isn't valid in this context

	If you tried to supply `-o` as a value rather than a flag, use `-- -o`

USAGE:
    s3-file-connector [OPTIONS] <BUCKET_NAME> <MOUNT_POINT>

For more information try --help

Using this in /etc/fstab:

djonesoa-eu-west-1 /mnt/s3-file-connector fuse.s3-file-connector defaults,endpoint_url=https://s3.eu-west-1.amazonaws.com 0 0

[dannycjones]

If we did want to support the use of fstab, we'd need to allow -o options to be set and also overwrite existing options that might have been set in fstab, such as the endpoint URL.

We might use argument delimiter to support -o, but handling unexpected options or ones that are normally full arguments is unclear.

[StarlightSyndrome]

This is somehow an additional info/request for implementing this feature.

I just have setup mount-s3 with systemd automount with a light wrapper script that does the actual mounting. A line in fstab would look like this

my-s3-bucket /s3/my-s3-bucket fuse.mount-s3.sh noauto,x-systemd.automount,--read-only,--allow-other

where the /usr/local/bin/mount-s3.sh script extracts the source, target, and the double dashed options and calls mount-s3 with it.

While this approach works well with autofs it does not with systemd automount since the latter already creates a mount on the target:

$ cat /proc/mounts
...
systemd-1 on /s3/my-s3-bucket type autofs (rw,relatime,fd=55,pgrp=1,timeout=0,minproto=5,maxproto=5,direct)
...

Subsequently mount-s3 fails on the actual mounting during triggered by accessing the mount point with:
Error: mount point /s3/my-s3-bucket is already mounted.

Basically it shall be enough to check the fs type on the mountpoint. If it is autofs it can be ignored.

Also I suggest that either we could filter the (comma separated) options passed by -o for those options with double dashes. Or we introduce our own parameter that can hold all our options like so:

my-s3-bucket /s3/my-s3-bucket fuse.mount-s3 noauto,x-systemd.automount,mount-s3-options=read-only,allow-other,max-threads=24

This parameter has to be put at the end of the the fstab options so everything following the mount-s3-options= can be considered mount-s3 arguments and parameters.

Alternatively without mount-s3-options= all parameters in the fourth argument could be tested if they are valid mount-s3 parameters and used if true.

Also note that the fuse.mount-s3-do wrapper script will not be needed any more. Just a fuse.mount-s3 will do!