You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Apparently, under these fast create/deletes, the operating system is reusing inodes for new files, and they are being detected as file moves, rather than creations:
However, the moves are arbitrary. For example, it detects a move from 87 to 83.processed. A possible workaround is looking for creation, modified events, and also moves (with dest_path in this case).
The text was updated successfully, but these errors were encountered:
I realize that that the problem of distinguishing remove/create with reused inodes vs file move is simply not solvable. But it seems something else is off in this scenario, where file modified events are also created for all files that were already deleted.
I'm observing missing events for newly created files in the following scenario:
X
, it createsX.processed
and removesX
.Here's a gist with code to reproduce the issue: https://gist.github.com/smola/d40614a9b546a650d87bc3588efc1014
Apparently, under these fast create/deletes, the operating system is reusing inodes for new files, and they are being detected as file moves, rather than creations:
watchdog/src/watchdog/utils/dirsnapshot.py
Lines 112 to 117 in 48c49a1
However, the moves are arbitrary. For example, it detects a move from
87
to83.processed
. A possible workaround is looking for creation, modified events, and also moves (withdest_path
in this case).The text was updated successfully, but these errors were encountered: