snapshot limit?

Is there a limit to the number of snapshots that can exist on a file
system concurrently?

Did I read something here recently about a limit to the number of paths by which a file could be known?

If there are limits, then what happens when I bump into or exceed that limit?

--rich

Yes. But if there's a limit to the number of paths that can point to a single file, then that's not strictly true. Rather, there's a limit based on the number of snapshots pointing to the same file.

Since a snapshot is a copy of an existing file system, the only time a snapshot would not have files in common with previous snapshots would be in snapshots of empty file systems. Making snapshots of empty file systems begs the question of why one would be bothering with snapshots anyway.

Hence, I suspect that in the vast majority of practical cases, there is indeed a current limit to the number of snapshots that can be made concurrently based on the limit to the number of paths to a single file.
I was looking for validation of this theory and perhaps a way to
detect, avoid, and/or recover from this situation.

I'm experiencing at least one recurring error condition resulting in a polluted file system. I'm using a lot of snapshots in this particular application. So I'm wondering if such a limit exists and/or how to determine if I am running into this situation.

Can anyone offer any clarification?

--rich

Would that limit also apply to de-duplicated copies of a file ?

Suppose I have tree under a btrfs file-system with lots of identical files. (eg zero length lock files), and I run a de-duplication script on it, to turn all those files into one, and then make a series of snapshots.

100 identical files multiplied by 100 snapshots (eg. daily for a few months), comes to a rather large number of links pointing to the same file.

David Pottage