User talk:Zhuyifei1999
This is a Wikimedia Commons user talk page.
This is not an article, file or the talk page of an article or file. If you find this page on any site other than the Wikimedia Commons you are viewing a mirror site. Be aware that the page may be outdated and that the user to whom this talk page belongs may have no personal affiliation with any site other than the Wikimedia Commons itself. The original page is located at https://commons.wikimedia.org/wiki/User_talk:Zhuyifei1999.
This is the user talk page of Zhuyifei1999, where you can send messages and comments to Zhuyifei1999.
- Be polite.
- Be friendly.
- Assume good faith.
- No personal attacks.
- Please sign and date your entries by clicking on the appropriate button or by typing four tildes (
~~~~
) at the end. - Put new text under old text.
- New to Wikimedia Commons? Welcome! Ask questions, get answers as soon as possible.
- Click here to start a new topic.
|
FlickreviewR 2 getting 403'd by Flickr
[edit]Since 21:30 today, the FlickreviewR 2 has stopped reviewing the files and flr-framework.err
is full of ex.
[2025-01-14 22:12:37] Sleeping for 60 seconds, received 403 from https://live.staticflickr.com/7281/16727651966_6951d76d15_o_d.jpg
It seems like I can curl files from the bastion, but from a k8s shell I got
$ curl https://live.staticflickr.com/7281/16727651966_6951d76d15_o_d.jpg
<html><body><h1>403 Forbidden</h1>
Request forbidden by administrative rules.
</body></html>
Seems like this will probably require some contact with them to resolve. AntiCompositeNumber (talk) 22:24, 14 January 2025 (UTC)
- FWIW, I apparently got a better response once with
curl -H 'User-Agent: [my email address]'
(curl said it didn’t want to write binary output to my terminal); but after that I now get the 403 error again, even if I vary the user agent a bit. (Unfortunately I usedcurl -v
and mosh ate the scrollback, so I can’t say for sure what the status code of the seemingly-successful response was.) I agree with your assessment, but just to be sure, does the bot use a good user agent? Lucas Werkmeister (talk) 22:37, 14 January 2025 (UTC)- I wonder if there is some hidden rate limiting? The Toolforge bastions have their own public IP addresses while the Kubernetes cluster nodes would end up behind the 185.15.56.1 (nat.cloudgw.eqiad1.wikimediacloud.org) SNAT address.
- I just tested 1) from my local laptop, 2) from the dev.toolforge.org bastion, and 3) from inside a
webservice python3.11 shell
shell on the Toolforge Kubernetes cluster and had identical success with all three attempts. My test command wascurl -v -o foo.jpg https://live.staticflickr.com/7281/16727651966_6951d76d15_o_d.jpg
. I was hoping the-v
header dump would show some rate limit counter, but it did not obviously include any. I did see anx-cache
header stating "Miss from cloudfront" in the laptop and bastion responses and "Hit from cloudfront" on the Kubernetes response. All 3 responses had differentvia
headers pointing to cloudfront servers. I guess also potentially relevant is that all 3 tests negotiated an HTTP/2 session. -- BDavis (WMF) (talk) 23:30, 14 January 2025 (UTC) - I don't see a User-Agent set in https://yifeibot.toolforge.org/gitweb/?p=botscripts.git;a=blob;f=o/toolserver/bryan/flickr/bots/flickreviewr.py;h=99b13230fb69806edda810e0a25310d63db3e59b;hb=refs/heads/master#l342 . I'll fix that and see if it helps. AntiCompositeNumber (talk) 02:13, 15 January 2025 (UTC)
- I have added a User-Agent to the image requests. It doesn't seem to have changed anything. AntiCompositeNumber (talk) 03:14, 15 January 2025 (UTC)
- I've got the same problem with my bot, Flickr now intentionnally limits the number of direct accesses to live.* URLs, see here. It seems they don't want to say what is the enforced rate limit, so probably the WMF has to contact Flickr directly. vip (talk) 00:13, 17 January 2025 (UTC)
- I wonder if the Flickr Foundation, who wrote Flickypedia, could help. They at least have an email address. AntiCompositeNumber (talk) 00:51, 17 January 2025 (UTC)
- I've got the same problem with my bot, Flickr now intentionnally limits the number of direct accesses to live.* URLs, see here. It seems they don't want to say what is the enforced rate limit, so probably the WMF has to contact Flickr directly. vip (talk) 00:13, 17 January 2025 (UTC)
- I have added a User-Agent to the image requests. It doesn't seem to have changed anything. AntiCompositeNumber (talk) 03:14, 15 January 2025 (UTC)
- Maybe this might be of interest to @Alexwlchan. RoyZuo (talk) 10:27, 15 January 2025 (UTC)
- Pinging @Sannita (WMF), as WMF support/resources might be needed here. All the Best -- Chuck Talk 01:27, 17 January 2025 (UTC)
- The polite question could be "what administrative rules?". — 🇺🇦Jeff G. ツ please ping or talk to me🇺🇦 01:38, 17 January 2025 (UTC)
- Based on the forum thread, anything more than 1 request per second will 403 out due to flickr not wanting to be used just as a "hosting" site the you embed, rather they want you to download the file and host it yourself. Still unsure why our usecase would be flagged, but I don't know how the bot gets licensing data. All the Best -- Chuck Talk 04:03, 17 January 2025 (UTC)
- The person I know is keeping contact with the Flickypedia people is currently out of office, because she's at a regional conference (or is coming back from it). I'll flag this for Monday to her, and hopefully will let you know soon. Sannita (WMF) (talk) 14:02, 17 January 2025 (UTC)
- Based on the forum thread, anything more than 1 request per second will 403 out due to flickr not wanting to be used just as a "hosting" site the you embed, rather they want you to download the file and host it yourself. Still unsure why our usecase would be flagged, but I don't know how the bot gets licensing data. All the Best -- Chuck Talk 04:03, 17 January 2025 (UTC)
- The polite question could be "what administrative rules?". — 🇺🇦Jeff G. ツ please ping or talk to me🇺🇦 01:38, 17 January 2025 (UTC)
- Pinging @Sannita (WMF), as WMF support/resources might be needed here. All the Best -- Chuck Talk 01:27, 17 January 2025 (UTC)
📣 Join the Wiki Loves Folklore 2025 Office Hour!
[edit]Hello, Wikimedia contributors!
We are excited to kick off Wiki Loves Folklore 2025], an annual global campaign celebrating the world’s rich and diverse cultural heritage. 🌍✨
To help organizers and contributors prepare for this year’s campaign, we’re hosting a Wiki Loves Folklore 2025 Office Hour, and we’d love for you to join us!
🗓 Event Details
[edit]- Date: January 18th, 2025
- Time: 4:00 PM UTC (Check your local time)
- Venue: Online via Google Meet
Join here
[edit]- Dial-in: +1 513-480-5188
- PIN: 876 566 200#
What’s in it for you?
[edit]Whether you’re an organizer, a contributor, or someone passionate about documenting cultural traditions, this session is for YOU!
- Learn About the 2025 Campaign: Get insights into this year’s theme, objectives, and exciting new elements.
- Ask Questions: Have questions or need clarifications? We’ve got you covered!
- Connect with Fellow Contributors: Hear stories and experiences from around the globe.
- Collaborate Globally: Share your ideas and contribute to making this campaign even more impactful.
Did You Know?
[edit]Wiki Loves Folklore has already documented over 121,000+ media files from more than 165 countries. Your contributions help preserve endangered traditions and unique cultural expressions for future generations.
How to Prepare
[edit]- Sign up for the Office Hour: Event Page
- Read the Blog: Explore what folklore is, what we look for, and how you can organize or contribute: Dance of Traditions: Join Wiki Loves Folklore 2025’s Global Celebration
- Sign Up to Organize: If you’re planning to organize a local Wiki Loves Folklore campaign, sign up here: Organizers Page
Let’s join hands to preserve and celebrate the beauty of folklore together. We look forward to seeing you at the Office Hour!
Warm regards,
On behalf of the Wiki Loves Folklore International Team
MediaWiki message delivery (talk) 06:01, 16 January 2025 (UTC)