OS/Platform name and version: Linux, Fedora 40, Kernel 6.8
Rust version (if building from source): rustc --version: rustc 1.80.0 (051478957 2024-07-21)
Notify version (or commit hash if building from git): 6.1.1
If you're coming from a project that makes use of Notify, what it is, and a link to the downstream issue if there is one: My personal file manager
Filesystem type and options: zfs, tmpfs, and zfs over nfs
On Linux: Kernel version: 6.8
What you did (as detailed as you can)
I first noticed this when I noticed some of my notify watches would stop producing events for changes from the local machine on some directories. Eventually I traced it to NFS automatically unmounting idle file systems, which is annoying, but if I had gotten notified of the issue I could have worked around it by resubscribing.
This bug is not specific to NFS, and the same problems occur locally as well when unmounting filesystems. I tested this with zfs and tmpfs locally.
What you expected
The events are sent to the event handler so I can decide how to handle them.
What happened
I see these logs, but no events are handled because they're not covered by any cases in handle_inotify. Ideally I'd have some way to tie that id back to a specific directory I was watching and reconnect or otherwise handle it.
System details
rustc --version
: rustc 1.80.0 (051478957 2024-07-21)What you did (as detailed as you can)
I first noticed this when I noticed some of my notify watches would stop producing events for changes from the local machine on some directories. Eventually I traced it to NFS automatically unmounting idle file systems, which is annoying, but if I had gotten notified of the issue I could have worked around it by resubscribing.
This bug is not specific to NFS, and the same problems occur locally as well when unmounting filesystems. I tested this with zfs and tmpfs locally.
What you expected
The events are sent to the event handler so I can decide how to handle them.
What happened
I see these logs, but no events are handled because they're not covered by any cases in
handle_inotify
. Ideally I'd have some way to tie that id back to a specific directory I was watching and reconnect or otherwise handle it.