This is a enhancement proposal to add a new way to filter mount points.

Our current use case is that we collect node_filesystem_avail_bytes metric  filtered by certain prefix on the mount point paths. However, if there are any inconsistencies in the naming, it can lead into difficulties to track all the wanted mount points.

I am proposing two alternative new ways filter mount points:

1. Add a way for node-exporter to read file system UUIDs so that then the mount point filtering could be done based on the UUID. This would be the preferred option because the UUID is supposed to be unique and works well with LVMs and regular disks alike.

2. Add a way for node-exporter to read file system labels so that then the mount point filtering could be done based on the label.

Do you think implementing one of these would be possible and which one you would prefer? If it would be possible, I would propose extending the filesystemCollector to include one of these filters.