After renaming in webdav folder, error on reload folder for few minutes

using command: rclone serve dlna idriveE2:/bucket --use-server-modtime --fast-list --no-checksum --log-level INFO
When i connect with webdav client, if i rename a file in a folder, then i get "error loading folder" when the client tries to refresh the folder after the rename for 30sec-5minutes, even if i try to manually refresh. After that everything is ok, but it's very annoying that it happens 70% of renames.
In the same time of error refresh retries i see these error in rclone log: http: superfluous response.WriteHeader call from*Handler).ServeHTTP (webdav.go:75)

It happened sometimes when i used google drive as backend, but happens much more frequently since i use idrive e2 S3 compatible cloud.

The folder where it happens seems to be random, sometimes happens in same folder, sometimes not.
Same happens without --use-server-modtime --fast-list but using --vfs-fast-fingerprint

What is the problem? Any hints?

This is the directory cache.

The default time is 5 minutes.

You can change the time or clear the directory cache using the remote control.

Not at computer to look up the flags but hopefully that is helpful!

But it happens after every single rename in a folder. Its hard to use as a usual drive with usual file operations if the refreshing of folder takes minutes after every change

Adjust the dircache time with this on the serve command.

  --dir-cache-time Duration                Time to cache directory entries for (default 5m0s)

To set the time period you want. The smaller you make it, the more often rclone will query the source.

I know this setting, but it isnt solves the problem of renaming

It should make the time you have to wait shorter. Does it?

sometimes i don't need to wait, i rename 3 files in 2 minutes, and after few seconds the name changes in webdav folder without problem. But mostly i rename, then error, and it stays for minutes, even if i refresh the webdav folder.
Why it is random?

BTW it solves nothing. Same waiting time if waiting is needed, even if i set it to 5s

Can you make a way for me to reproduce the problem locally?

If you put a lot of (mostly video) files in a folder, and thare are a lot of other files around that folder in the same remote folder, that can reproduce it. But as i wrote, its't totally random, in the same share, same files.

But the initial loading of filetree isn't takes so long at the iniital load, just around 5-10 seconds. Even if this error happens, i can move to upper folder and open the files, but if i enter the previous folder i see the error until it loads after some time.


which webdav client that connect to dlna?

can you:
--- post the full output of rclone version?
--- change --log-level INFO to --log-level DEBUG, rerun the command, post the top thirty lines?

rclone v1.61.1
- os/version: ubuntu 20.04 (64 bit)
- os/kernel: 5.4.0-135-generic (x86_64)
- os/type: linux
- os/arch: amd64
- go/version: go1.19.4
- go/linking: static
- go/tags: none

DEBUG log level causes a lot of spam, for example logging in each minute, even in idle what is NOT deleted from VFS cache :smiley:

But i'll try to catch log when the problem happens.

Also the webdav client (CX file manager on android) where this problem exists, there are a lot of different operation parallel to the rename, as it tries to load preview for each video, and image file.

Even the problem is not the same, as sometimes that happens what i wrote, but sometimes i do a rename, the client writes that its successfully renamed, and shows is like it. But after few minutes, after the filelist refreshed, i see that the renamed files are duplicated, one with old name, and one with the new name, and i see the same with other clients even on the cloud service's own web file browser.

So some days when rename, the client shows fail, and the changes do not appear until the next directory cache expire, but the rename in the background is done correctly. Some days the client shows its done, but in the storage, the files are duplicated.
Some days it's just working correctly.
Always same client, same rclone settings.

This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.