-
Notifications
You must be signed in to change notification settings - Fork 665
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Virtual File Sync no longer working properly #8295
Comments
I can confirm the behavior of file deletions on the server if the virtual file sync is used. |
I have done more testing. The problem seems to be related to a Windows 10 update.
Both Windows are on the 20H2 version. As a result of Windows update policy, all computers at my office updated at the beginning of this week causing also the update of the client (due to the reboot of the OS). |
I can confirm that, we are trying to find what changed in the "cloud files" windows API. But if there was a major change in how windows 10 handle files, they would have announced it... |
Thanks! We have identified the issue and are working on a fix. |
Fixed in 2.7.3 https://github.com/owncloud/client/releases/tag/v2.7.3 |
Is there a script to restore all files since a given date? Should be doable using bash & curl... An moving files on the server CLI and then running occ file:scan would assign new file ids to the deleted files and all shares and metadata would be lost... |
@hodyroff Sorry for comment on a closed issue. Can you share which API is changed in Windows 2020H2? It might affect other software use the same API. Thanks. |
|
@efrenbg1 @welkineins @voerg @Tenivar @butonic Restore script is here |
I'm a bit scared now to use ownCloud in production at my company, where I am responsible for the infrastructure. Could that happen again? |
Well. You might want to refer to microsoft/Windows-classic-samples#164 to understand how that could happen at all and what levels are involved. As much as I can't assure you that nifty things will never happy again, I can assure you that we do everything for our part to avoid that. This has been the most severe bug in the desktop syncing since years. I am grateful that we have mechanisms in place to help (trashbin). |
@dragotin Just out of curiosity. Don’t changes to the cloud file API show first in the Insiders program? All changes made to the cloud file API (or others) are first tested and published to the Insiders channel, but considering they are making constant changes to that API I'm not really surprised that has passed under the radar. |
I'm running the beta insiders preview and the problematic change 19041.685 was released at the same time as for the other Windows users. |
Thanks, i did try and run the restore.php, Having a warning on composer install, and an error on php restore.php execution. Warning on installation : Installing sabre/uri (2.2.1): Downloading (100%) php restore.php |
@EdouardRt some permission errors, please try to follow the steps here: |
thank you for your help @janackermann ! it does go through the cache warning, but the error while restoring remains :
My detailed configuration is available here. |
@EdouardRt This error usually occures if the script can't connect to the server
Totaly "normal", please don't try to rename them |
@janackermann Yes, variables are defined, all 4 between single quotes : 'name' ; 'https://website.xt' ; 'p@sword' ; 'yyyy-mm-dd' . |
@EdouardRt please try the following:
|
Thanks for the updated script !
|
@EdouardRt you are welcome! 404 usually means that the requested resource does not exist.
Is the server reachable from the client you want to execute the script? There is a new version of the script, please download the new version, do a composer install
An additional / at the end of the url can lead to problems |
@janackermann While : Trying the updated script right away. |
in order to get the script run this must be fixed Edithttps://server.xt/remote.php/dav/trash-bin/<your_username> |
@janackermann I tried an occ repair > no change Browsing through the files, i see them in data/user/files_trashbin/files Any hint on where to look ? i did not find any relevant case elsewhere... |
@janackermann |
@EdouardRt |
The 2.7.2 client deleted hundreds of thousands of files from our server. We could not access the "Trash" via the web interface: The recovery script was unusable because it would take hundreds of days for it to recover the files. I could not find a way to use Approximately 92% of the files succeeded. We now face other problems: First, when I try the TrashBin API restore on one of the remaining files, it says that the file already exists. It does not. I cannot find it anywhere in the database (except the trash and the history, showing it was deleted). It's not in the filesystem. Second, the web interface works now (presumably due to the lower number of files) but has no way of selecting files before/after a certain date. I don't want to restore files that were intentionally deleted before the client bug started deleting everything. I decided to try using the TrashBin API to delete files from the trash, before a certain date. That brings me to the third problem. Apparently, I can only delete one file at a time via the API. I get an error about it being "locked" if I try to run concurrent calls. It's taking 70-80 seconds per delete, which means it will take days to complete this relatively small operation. What can we do to solve this? |
This is still an issue even with Windows client v2.7.6: |
Could you open a new issue and fill in the details? |
Thanks Edouard for sharing this issue and the fix. at the end, the files didn't get restored. I don't know if this is because : Anyway, I found on the new owncloud web interface a "select all" checkbox in the "Trash" page. I just clicked it and pressed Restore. The operation takes a while, but it finally works. Thanks Jan Akermann, thanks owncloud team, thanks Edouard. |
Does this assume everyone saves their users passwords so they can run these scripts? I don't like having to ask users for their passwords. Is there no system wide deleted files access that an admin can access and do bulk restores from? This seems like a huge oversight! As a side note I'm also spending today restoring files that were deleted as a result of VFS being enable on a clients system. Client is using 2.9.1 and is moving from non-VFS to VFS. It looks like the client decided that all the files the client has not previously synced didn't need to be on the ownCloud server either. User made the change last night and managed to delete well over a TB of data. Enormous pain in the ***. |
No, you can also use impersonate as an admin to do what is needed here. The script also doesn't need the user PWs ... but as stated above it is slow. If you have a reproduceable case - how it is still possible to delete by accident with 2.9.1 we would be happy to look into it and try to find a fix, please open a new issue in that case. Sorry vor any invoncinience. If the user has pointed ownCloud to an existing sync folder where selective sync was activated, this sounds possible, but there is a clear warning. We do recommend branding and using MECM or similar for larger instances to avoid any user interaction mistakes. Ideas for better usability always welcome as well. Thanks for your feedback! |
I am afraid this issue occured to us somehow again with the latest desktop client on one of our few Windows 11 computers last week. The user whose credentials were logged by ownCloud server is not aware at all of manipulating in any way with ownCloud files in the last weeks yet basically all the contents of all the folders shared to that user vanished last week from the server, residing in the recycle bins of the dozens of users owning various shared folders. Win11 Czech Edition Pro, ownCloud desktop client 2.10.1, server 10.9.1 on Ubuntu Server 18.04. Unfortunately I had logging limited to errors and not notices etc so no data about the DELETE commands or so from the owncloud.log are available. The only mention I see is in the Activity tab where it is saved which particular user was responsible for the mass deletion. |
Could you check if you have a file called |
And, @brozkeff, just to be sure: The sync uses Virtual Files (VFS), correct? |
I am sorry but it was a computer of my colleague who asked me to disable the owncloud sync completely to not mess with the files, since she did not need the files to be synced at all for her work. At least I discovered Apache access logs and detected the very first files being deleted, however nothing fancy to see there, just the DELETE commands sent by the client (IP and username redacted):
|
@brozkeff thanks for your help. Let's keep the eyes open on this one. |
Since the beginning of this week all clients updated automatically to client version 2.7.2. This made Virtual File Sync to stop working properly. I have observed two behaviors:
Expected behaviour
Working with Virtual File Sync as expected.
Actual behaviour
All files are missing from the client folder, causing a complete wipe of data on the server.
Steps to reproduce
Server configuration
Operating system: debian 10 (Linux 4.19.0 - amd64)
Web server: nginx/1.14.2
Database: 10.3.25-MariaDB
PHP version: 7.2.34
ownCloud version: 10.5.0
Storage backend (external storage): none
Client configuration
Client version: 2.7.2
Operating system: Windows 10 Pro 20H2
OS language: Spanish
Qt version used by client package (Linux only, see also Settings dialog): none
Client package (From ownCloud or distro) (Linux only): none
Installation path of client: C:\Program Files (x86)\ownCloud
Logs
Client logfile: (Nothing on the log file)
Web server error log: (Nothing on the log file)
Server logfile: (Nothing on the log file)
What I've tried
Full clean install of the client removing all %appdata% files
Dowgrading to previous client versions (only 2.5.4 seems to work as of right now)
Clean install of ownCloud server
Update Windows 10 to latest version
~ Thanks in advanced for your hard work ~
The text was updated successfully, but these errors were encountered: