UPDATE: Cloudflare outage! - Discourse update to "3.5.0.beta6" failed

Product Support Self-hosting

AI-generated summary

Users experienced failures during a Discourse update to 3.5.0.beta6 due to a widespread Cloudflare outage impacting services like npmjs.org, AWS, Google Cloud, and others haydenjames. Initial tests by haydenjames showed 503 errors when accessing npm registry endpoints via curl, coinciding with the Cloudflare issues. Several users reported that disabling the reverse proxy temporarily allowed them to rebuild eisammy, but this was ultimately ineffective as the npm registry itself was affected RGJ. Cloudflare services eventually recovered haydenjames, and users were able to successfully complete the update, with some ending up on version 3.5.0.beta7-dev. The situation highlighted the dependency on Cloudflare for package installations and the need to wait for their services to be restored.

1

Edit: I tried testing with curl. This works:

curl -I https://registry.npmjs.org/
HTTP/2 200 
date: Thu, 12 Jun 2025 18:26:56 GMT
content-type: application/json
cache-control: public, immutable, max-age=31557600
set-cookie: _cfuvid=...-0.0.1.1-604800000; path=/; domain=.npmjs.org; HttpOnly; Secure; SameSite=None
server: cloudflare
cf-ray: 94eb6640d930ac93-YYZ

But this and others fail with 503 error:

curl -I https://registry.npmjs.org/@swc/core/-/core-1.12.0.tgz
HTTP/2 503 
date: Thu, 12 Jun 2025 18:27:36 GMT
content-type: application/json
set-cookie: _cfuvid=...-0.0.1.1-604800000; path=/; domain=.npmjs.org; HttpOnly; Secure; SameSite=None
server: cloudflare
cf-ray: 94eb664ced5cabe8-YYZ

Edit 2: Is the issue related to npmjs.org having issues with package install, package publish, and login?: also see: https://status.npmjs.org/

2

ha! :smile: Talk about timing! Looks like, coincidentally, Cloudflare is having issues right now: https://www.cloudflarestatus.com/

The image shows the Cloudflare System Status page indicating broad service outages, with updates and a notice indicating an investigation into issues with certain services, particularly affecting Access authentication and Cloudflare Zero Trust WARPs. (Captioned by AI)
The image shows the Cloudflare System Status page indicating broad service outages, with updates and a notice indicating an investigation into issues with certain services, particularly affecting Access authentication and Cloudflare Zero Trust WARPs. (Captioned by AI)1053Ă—577 52.4 KB

My update request was about ~5 mins when this issue started. Argg.

Question, when this is resolved will ./launcher rebuild app be able to complete the upgrade? Anything additional I need to be aware of?

Update: This is a pretty big outage. AWS impacted, Google Cloud, Etsy, Google Drive, Mailchimp, etc. etc. etc. :popcorn:

I revert VM to a backup if this is extended.

Back online:

https://x.com/dok2001/status/1933270392656765397

3

Oh my gosh Really CF?

4

Disabling reverse proxy helped me to rebuild… I guess.

@edit For some reason I not getting rebuild without put to skip mac address

./launcher rebuild app --skip-mac-address

5

Nice. Well, actually everything was restored a while now:

**Update** - Cloudflare services are recovering quickly around the globe...
@Jun 12, 2025 - 20:32 UTC

In my case, the same curl test I ran earlier, started working again, so I just updated successfully also:

$ curl -I https://registry.npmjs.org/@swc/core/-/core-1.12.0.tgz
HTTP/2 200 
date: Thu, 12 Jun 2025 21:27:27 GMT
content-type: application/octet-stream
cf-ray: 94ec6eabdf84a21d-YYZ
cf-cache-status: HIT
...

Now I ended up on 3.5.0.beta7-dev :wink:

6

I guess not. That rebuild failed because registry.npmjs.org is fronted by CF. There was nothing you could do - except wait for it to recover.