Before you start

• I have searched the forum and the existing issues for this problem.
• I have tested with the latest beta or stable release and can still reproduce the problem.
• This is a reproducible bug, not a usage question (questions belong on the forum).

Associated forum post URL

No response

What is the problem you are having with rclone?

rclone rc vfs/forget is meant to make an rclone mount pick up a file that changed on the remote,
but it only refreshes rclone's own cache — not the kernel's. So a read can come back truncated to the
file's old size for up to --attr-timeout.

How to reproduce

1. rclone mount a remote with --attr-timeout 60s.
2. Read a file through the mount — the kernel caches its size for 60s.
3. Change that file on the remote so it grows (another client, or a WebDAV backend whose store is edited).
4. Call rclone rc vfs/forget file=....
5. Read the file again any time in the next 60s: it's truncated to the old
   size. vfs/forget made no difference; only the timeout expiring eventually
   heals it. This is because the kernel answers getattr from its own cache.

Repro script: https://gist.github.com/thomasjm/a76fabd691911437874e155d12bd1cdd

rclone version

rclone v1.75.0-DEV

• os/version: nixos 25.11 (64 bit)
• os/kernel: 6.12.76 (x86_64)
• os/type: linux
• os/arch: amd64
• go/version: go1.25.7
• go/linking: dynamic
• go/tags: none

Operating system

Linux

Which cloud storage system are you using?

Local WebDAV serve

Output of rclone config redacted

N/A

The command you were trying to run

N/A

A log from the command with the -vv flag

N/A

Anything else?

I have a proposed fix here, which I'd be happy to open as a PR:

https://github.com/thomasjm/rclone/tree/vfs-forget-invalidate-kernel-cache