Page MenuHomePhabricator

Upstream caches: 404
Closed, ResolvedPublicBUG REPORT

Assigned To
Authored By
Yann
Mar 12 2023, 2:58 PM
Referenced Files
F36913184: missingfile.PNG
Mar 15 2023, 4:25 PM
F36913181: source-code.html
Mar 15 2023, 4:25 PM
F36912520: Screenshot_of_broken_thumbnails_on_Wikimedia_projects_01.png
Mar 15 2023, 10:58 AM
F36912521: Screenshot_of_broken_thumbnails_on_Wikimedia_projects_02.png
Mar 15 2023, 10:58 AM
Tokens
"The World Burns" token, awarded by TheresNoTime."Burninate" token, awarded by doctaxon."Burninate" token, awarded by Don-vip.

Description

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

What happens?:
Our servers are currently under maintenance or experiencing a technical problem. Please try again in a few minutes.
See the error message at the bottom of this page for more information.

Request from 81.185.168.223 via cp6002 cp6002, Varnish XID 432320098
Upstream caches: cp6002 int
Error: 404, Not Found at Sun, 12 Mar 2023 14:54:40 GMT

Other information (browser name/version, screenshots, etc.):
Chrome on Windows 10

Event Timeline

I fail to reproduce from Europe. Is this reproducible or a one-off?

I fail to reproduce from Europe. Is this reproducible or a one-off?

Apparently, just one-off.

I think I have the same problem on several files. Example one:
https://commons.wikimedia.org/wiki/File:NASA’s_SpaceX_Crew-5_Splashdown_(NHQ202303110021).jpg => image is not displayed
https://upload.wikimedia.org/wikipedia/commons/thumb/c/cf/NASA’s_SpaceX_Crew-5_Splashdown_(NHQ202303110021).jpg/1280px-NASA’s_SpaceX_Crew-5_Splashdown_(NHQ202303110021).jpg => HTTP 404

Request from 2a02:8428:7cba:5a01:f4a4:8d46:2615:ae62 via cp6008 cp6008, Varnish XID 555059990
Upstream caches: cp6008 int
Error: 404, Not Found at Tue, 14 Mar 2023 12:40:29 GMT

Firefox 110.0.1 (64 bits) on Windows 11, French ISP

Again here https://upload.wikimedia.org/wikipedia/commons/thumb/f/f1/Rolland_-_Beethoven,_tr._Hull,_1927.pdf/page9-375px-Rolland_-_Beethoven,_tr._Hull,_1927.pdf.jpg

Request from 90.112.34.87 via cp6006 cp6006, Varnish XID 675386332
Upstream caches: cp6006 int
Error: 404, Not Found at Tue, 14 Mar 2023 17:15:04 GMT

Chrome on Windows 10, French ISP

Aklapper triaged this task as Unbreak Now! priority.Mar 14 2023, 8:01 PM
Aklapper added projects: Thumbor, SRE.
Aklapper added a subscriber: GPSLeo.
#wikimedia-tech 2023-03-14
[16:39]  <  roy649_> When a do a get on https://upload.wikimedia.org/wikipedia/commons/thumb/b/b2/Pete_Sutherland_playing_at_Bristol_town_tribute_(cropped).jpg/280px-Pete_Sutherland_playing_at_Bristol_town_tribute_(cropped).jpg, I get "Our servers are currently under maintenance or experiencing a technical problem. Please try again in a few minutes."
[16:39]  <  roy649_> Just in case this isn't already a known issue.
[16:40]  <  roy649_> Request from 2600:4041:51ec:c500:fdcb:d9c7:57cb:3c21 via cp1090 cp1090, Varnish XID 716451937
[16:40]  <  roy649_> Upstream caches: cp1090 int
[16:40]  <  roy649_> Error: 404, Not Found at Tue, 14 Mar 2023 16:38:25 GMT
[16:58]  <AntiComposite> works for me

[17:18]  <  roy649_> Hmmm, working for me now too.  I'll blame this on everybody's favorite issue: cache invalidation :-)
[17:24]  <  roy649_> There's still something going on.  I'm getting similar results now for:
[17:24]  <  roy649_> https://upload.wikimedia.org/wikipedia/commons/thumb/archive/b/b2/20230314172235!Pete_Sutherland_playing_at_Bristol_town_tribute_(cropped).jpg/120px-Pete_Sutherland_playing_at_Bristol_town_tribute_(cropped).jpg
[17:24]  <  roy649_> and
[17:24]  <  roy649_> https://upload.wikimedia.org/wikipedia/commons/thumb/b/b2/Pete_Sutherland_playing_at_Bristol_town_tribute_(cropped).jpg/120px-Pete_Sutherland_playing_at_Bristol_town_tribute_(cropped).jpg?20230314172238
[17:25]  <  roy649_> Request from 2600:4041:51ec:c500:fdcb:d9c7:57cb:3c21 via cp1090 cp1090, Varnish XID 725720168
[17:25]  <  roy649_> Upstream caches: cp1090 int
[17:25]  <  roy649_> Error: 404, Not Found at Tue, 14 Mar 2023 17:25:14 GMT
[17:25]  <  roy649_> and
[17:25]  <  roy649_> Request from 2600:4041:51ec:c500:fdcb:d9c7:57cb:3c21 via cp1090 cp1090, Varnish XID 703262314
[17:25]  <  roy649_> Upstream caches: cp1090 int
[17:25]  <  roy649_> Error: 404, Not Found at Tue, 14 Mar 2023 17:25:32 GMT
WARNING: random speculation ahead!

Typically swift (the media storage system) would not return a 404 unless the source file actually does not exist. When it looks for a thumbnail and cannot find it swift should call Thumbor to create the thumb from the original and then store and return the thumb. This all sort of looks like the 404 is actually going out to the CDN edge as a response to the original caller, but also that the thumb is generated and stored for the next requestor. I haven't yet been able to recreate the 404 for any of the reported URLs.

Some logs (NDA, they have IPs in) for a particular request above - P45869.

It's notable that if thumbor gets an AUTH failure, it will return a 404. I suspect that thumbor doesn't cache its auth sessions very well, and so may be somewhat inclined to overload the swift auth service; I would generally expect thumbs to be regenerated next time round (if you see what I mean), modulo the edge caching the 404 for 10 minutes.

Hello, this problem causes difficulties on Wikisource, due to the impossibility to access page image while proofreading...

I'm experiencing some problems with files not displayed and broken thumbnails too, here are my screenshots (from discussion on Village pump "Files are not appearing.") I have a good internet connection, in general I have no problems with thumbnails, upload and download. Windows 10 Home, Chrome version 111.0.5563.65 (64 bits) and Firefox version 111.0 (64 bits).

Screenshot 2023-03-14 at 16-39-31 Caricamenti di Patafisik - Wikimedia Commons.png (1×1 px, 320 KB)

Screenshot 2023-03-14 at 16-31-59 Wikipedia Ciàcołe - Wikipèdia Vèneta ła ensiclopedia łìbara.png (1×1 px, 102 KB)

Screenshot_of_broken_thumbnails_on_Wikimedia_projects_02.png (1×2 px, 216 KB)

Screenshot_of_broken_thumbnails_on_Wikimedia_projects_01.png (313×844 px, 24 KB)

further three are missing when searching: https://de.wikipedia.org/w/index.php?fulltext=1&search="Auguste Bock"&title=Spezial:Suche&ns0=1

Picture links are missing in source code:

missingfile.PNG (1×2 px, 243 KB)

2 things can be true at the same time.

maybe it's unclear as rendered but some articles just don't have any images associated with them. or whatever algorithm is excluding some from being chosen as the "primary" image for an article. see e.g. https://de.wikipedia.org/wiki/Liste_der_Gedenktafeln_in_Berlin-Mitte?action=info&uselang=en

anyway if you want to pursue this further I think that's not relevant here, please open a new ticket.

It appears that there was a lingering switchover issue related to communication between thumbor and swift. We're still investigating what caused this but it appears that new files are being thumbnailled correctly now.

MatthewVernon claimed this task.

I'm closing this task now, as I think thumbnails are now being correctly generated.

That's not a 404 error but a 429 error, thus a different issue.

@Aklapper which bug task belongs to this 429 error issue?