Page MenuHomePhabricator

File not found: /v1/AUTH_mw/wikipedia-commons-local-public ... for 3 files
Closed, DeclinedPublicBUG REPORT

Description

The "user:Ra Boe" asked me to report this bug for 3 of his files

Steps to replicate the issue (include links if applicable):

Go to:

What happens?:
Click on the full resolution link.

  • File not found: /v1/AUTH_mw/wikipedia-commons-local-public.1d/1/1d/Triumph_Herald_13-60_Nettuno_by-RaBoe_02.jpg
  • File not found: /v1/AUTH_mw/wikipedia-commons-local-public.b7/b/b7/Triumph_Herald_13-60_Nettuno_by-RaBoe_03.jpg
  • File not found: /v1/AUTH_mw/wikipedia-commons-local-public.5b/5/5b/Triumph_Herald_13-60_Nettuno_by-RaBoe_05.jpg

Event Timeline

I'm afraid that these three images are long gone. They are not found in either swift cluster, nor in either site's backups, which tells me they've been absent for at least 5 years (since our media backups start in about 2019).

To answer the obvious "why now?" question - we've been clearing out old thumbnails (per T379942); previously we tended to keep all thumbnails forever, which meant that if an original was lost/damaged/missing but we had thumbnails, then those thumbnails would always be visible, so the loss of the original would not be evident.

If the reporting user (who I think is the original photographer) still has the original photographs, I suggest they re-upload them. But I'm afraid there's nothing more I can do from an object storage perspective.

As there are likely many more of these cases is there a possibility to scan over all files on Commons to find all files affected?

As there are likely many more of these cases is there a possibility to scan over all files on Commons to find all files affected?

In a way that's been done while taking backups, the issue is that some files have been deleted on purpose (think illegal stuff), other never were uploaded (e.g. imports from wikivoyage) and others may be lost. A larger audit would be needed to check which are on the third category.

As there are likely many more of these cases is there a possibility to scan over all files on Commons to find all files affected?

My feeling is that there are likely not many (more) of these - we've been deleting thumbnails for months now, and it has only so far uncovered a very few previously-unknown-to-be-lost images.

As there are likely many more of these cases is there a possibility to scan over all files on Commons to find all files affected?

In a way that's been done while taking backups, the issue is that some files have been deleted on purpose (think illegal stuff), other never were uploaded (e.g. imports from wikivoyage) and others may be lost. A larger audit would be needed to check which are on the third category.

See T17889: Write and run script to find non-existent images on Wikimedia wikis for dedicated task for this. For files we can not host for legal reasons they should first be deleted in normal way, so such missing file should only occurred in filearchive.

This is happening now for https://upload.wikimedia.org/wikipedia/commons/6/64/2025-11-16_ONEW_concert_032.jpg :

File not found: /v1/AUTH_mw/wikipedia-commons-local-public.64/6/64/2025-11-16_ONEW_concert_032.jpg

The file was uploaded on 2025-12-11 with UploadWizard and I'm positive that at the time all the files correctly showed a thumbnail on https://commons.wikimedia.org/wiki/Category:Onew_Percent_in_Helsinki . No logged action appears to have happened since.

The file has been reuploaded from backups:
https://commons.wikimedia.org/wiki/File:2025-11-16_ONEW_concert_032.jpg

But please, when reporting new cases, create a new ticket with the same tags. It is ok to do it here, but there is a high change it may be missed.

Pppery subscribed.

Closing as declined so that this ticket doesn't remain open forever. The files in question don't exist, and there seems to be nothing anyone can do about it now. I've tagged the lingering file pages on Commons for speedy deletion as corrupt.

@Pppery: please also notice the file uploader (who is still active) so they can know the issue and potentially reupload the file.

Feel free to do that yourself. But they're already aware and made the initial report so I saw no need to.