Small drives are shown in red even when there is plenty of space

[metov]

When I create a 1 GB partition on my drive, and write 100 MB of data to it, duf highlights it in red. I assume this means low space and indicates that "only" 900 MB free space is left. However, a partition being 10% full hardly seems like something to be alarmed over.

I suppose it is ultimately subjective whether people would care more about absolute or relative free space. Sounds like duf is going with the latter. Is there any possibility of focusing on relative space instead?

• In my use case, the partition in question is /boot and hence the 900 MB of free space is actually plenty.
• Using relative thresholds is self-updating, and obviates the need to perpetually increase the threshold of what counts as "low" as hard drives get bigger and user expectations change.

If it is desirable to keep the current behavior of going by absolute sizes, maybe we could have an option --highlighting {relative|absolute}?

[Justinzobel]

This is an issue for me also as it shows boot and efi partitions in red even though there's plenty of space:

[muesli]

This kind of works as intended. You'll still get to see a green value for the usage in percent. As I see it, unifying the color coding for both columns would only remove information. I do understand that everyone has different needs however, so maybe duf's recently added --avail-threshold and --usage-threshold flags help remedy this a bit?

# Availability threshold, for example in gigabytes
$ duf --avail-threshold="10G,1G"

# Usage-threshold in percent between 0.0 and 1.0
$ duf --usage-threshold="0.5,0.9"

[Justinzobel]

The problem with the thresholds is they are global I imagine, against all disks.

I don't want it to tell me when only 559.7M is available on my root partition, which could easily fill up in moments with caches and regular activities.

[metov]

@muesli Thank you for the response! Those options resolve my issue, so feel free to close.

[C0rn3j]

@muesli allowing to use --avail-threshold and --usage-threshold repeatedly with some kind of optional identifiers would solve this issue, more detailed description on dysk repo: Canop/dysk#18 (comment)