It recently came to my attention, that a fixed number of inodes is only a characteristic of some file system, not all of them.
In this case it was btrfs which allocates inodes dynamically (and where the remaining space is more difficult to measure).
The general problem of different filesystem characteristics is far more extensive, there might be questions about the consistency of the journal or quotas instead of "physical" limits.
In general there are some things to consider for now and in the future:
There has to some filesystem-specific path for some measurements/tests
What can go wrong in a specific filesystem and should therefore be monitored?
Describe the feature request
It recently came to my attention, that a fixed number of inodes is only a characteristic of some file system, not all of them. In this case it was btrfs which allocates inodes dynamically (and where the remaining space is more difficult to measure). The general problem of different filesystem characteristics is far more extensive, there might be questions about the consistency of the journal or quotas instead of "physical" limits. In general there are some things to consider for now and in the future: