Hey guys
My forum (https://forum.billysbilling.dk) just went down. As far as I can see it’s due to the CDN. I hope someone can help me - maybe @sam or @codinghorror?
Hey guys
My forum (https://forum.billysbilling.dk) just went down. As far as I can see it’s due to the CDN. I hope someone can help me - maybe @sam or @codinghorror?
It’s up for me. Is it still down for you?
Yes, unfortunately it is.
Can you try clearing your browser’s cache and refreshing?
@zogstrip it isn’t a caching issue. We’re located in Denmark, and right now I’m getting a 504 Gateway Time-out across different devices on different networks (while trying to entering the page - not on loading resources).
Sometimes the page will load, but at most times with 404 and 50* errors on resource loading.
Our customer-base (entering this forum) is based in Denmark also, but I don’t see it only as a CDN issue anymore.
Confirmed … investigating … will have it sorted shortly
@christianjohnsen / @nicklasmoeller is it still down now?
Seems to work now! Thanks.
Will get back to you, if the problem occurs again
Looks OK, apologies for the difficulty – we are having some issues with our glusterfs network file system:
We’ve moved more essential resources (stylesheet, etc) to an internal site path so they won’t be stored in the network file system and vulnerable to issues like this. I believe @sam also moved the default letter avatars to an internal site path.
(It’s OK for a picture of a lolcat in a topic, or even some avatars, to be temporarily missing, but it is much more traumatic if the CSS file for the site is missing!)
Ahh okay. I was in the process of making a topic about this. It’s happening here on Meta and the Imgur community too. Images appear broken from Discourse’s fastly CDN.
OK, so that’s not just me. Sometimes when my phone is running out of space some images won’t load, and since I was getting the low space warnings I initially chalked it up to that.
Yeah we are still working on fixing the missing images but stylesheets should be very solid now, no longer served from a shared fs.
Note this was sitting in my bookmarks list, this particular failure mode is no longer likely at all:
https://github.com/discourse/discourse/commit/f58d85edea0f85a9da52798aa45bf6a202c10a61
(this is already deployed to all our customers)
Our filesystem migration is over, if any of our customers reading this see anything amiss PM me.