[DigitalPoint] App for Cloudflare®

[DigitalPoint] App for Cloudflare® 1.8.4

No permission to download
That wouldn’t have anything to do with this addon. 520 means your origin web server is returning something erroneous. Typically caused by a misconfiguration of your web server. You’ll want to contact Cloudflare support so they can help you track down what it is exactly.

Did Cloudflare resolve their HTTP/2 to origin bug that can cause 520 errors?
 
Did Cloudflare resolve their HTTP/2 to origin bug that can cause 520 errors?
No idea... my origins are all running HTTP/2 without errors, so not something I know anything about first-hand. If Cloudflare has an error with it, maybe it's specific to certain web server configs? 🤷🏻‍♂️

I just checked my web logs to be certain and the last 20,000 entries in the request logs (as far back as I felt like doing right now), have been 100% served by the origin via HTTP/2 to Cloudflare. With zero errors.

If Cloudflare had an issue on their end with HTTP/2 to the origin on a large scale, I'd think it would be fairly catastrophic for them and most sites using Cloudflare.
 
I haven't had it happen on my own servers, fortunately. Only a small number of my clients have had this. The first client... wow what a headache it was tracking it down to a Cloudflare issue!
 
I haven't had it happen on my own servers, fortunately. Only a small number of my clients have had this. The first client... wow what a headache it was tracking it down to a Cloudflare issue!
Ya, if it's limited to certain servers, I'd say it's probably a server config thing, not a Cloudflare thing. But who knows... like I said, I've never seen it myself so no idea if Cloudflare "fixed" anything on their end.

I haven't seen any issue firsthand with Cloudflare getting content from the origin via whatever protocol... even pre-HTTP/2 I was serving SPDY to Cloudflare without issue.

Maybe something with a server having old crypto libs or something that were generating unexpected output or handshakes or something? 🤷🏻‍♂️
 
Nope... no clue what page_cache is. Probably a third-party addon (no reference to it in XenForo's source). image_cache is the directory that the image proxy uses.

But ya, you can move it to R2 if you want... however, it's a pretty terrible idea to store cached things in the cloud if you ask me... But do whatever you want. 🤷🏻‍♂️
It's image_cache, not page cache. 🤦‍♂️

(My brain's in automotive repair mode today, not "coding like it's 1996" day...)

So since it's the image proxy cache, it should work the same as having attachments in R2, I'm thinking?

I wouldn't store anything else in R2 though.
 
It's image_cache, not page cache. 🤦‍♂️

(My brain's in automotive repair mode today, not "coding like it's 1996" day...)

So since it's the image proxy cache, it should work the same as having attachments in R2, I'm thinking?

I wouldn't store anything else in R2 though.
It's not something I would do... but like I said, there's nothing preventing you from doing it, so have at it.

To me a "cache" is something that doesn't need to persist (nothing catastrophic is going to happen if you lose the contents of the cache) and should also be very fast. Storing it in R2 makes it not super fast... request comes in, server handles it, first makes an API call to fetch the image from some remote data center where R2 data lives, then it needs to take that and pass it through to the user. So let's say it's a 10MB image... the server needs to download 10MB from R2 before it can then send that same 10MB to the user.

Again... it will work, but to me a cache should be local and as fast as possible. Like for me, I have a lot of things on my servers mapped to /dev/shm (a RAM volume). /tmp, /var/logs/nginx (web server logs), image_cache, oembed_cache, etc... these are all things that aren't going to be catastrophic if the server reboots and loses what's in memory... it saves a lot of disk writes (especially for things like /tmp and web logs) and accessing it is extremely fast.

Just depends on how your server resources are... if you are short on disk space, then ya... probably makes sense to free that up by moving it to R2.
 
It's not something I would do... but like I said, there's nothing preventing you from doing it, so have at it.

To me a "cache" is something that doesn't need to persist (nothing catastrophic is going to happen if you lose the contents of the cache) and should also be very fast. Storing it in R2 makes it not super fast... request comes in, server handles it, first makes an API call to fetch the image from some remote data center where R2 data lives, then it needs to take that and pass it through to the user. So let's say it's a 10MB image... the server needs to download 10MB from R2 before it can then send that same 10MB to the user.

Again... it will work, but to me a cache should be local and as fast as possible. Like for me, I have a lot of things on my servers mapped to /dev/shm (a RAM volume). /tmp, /var/logs/nginx (web server logs), image_cache, oembed_cache, etc... these are all things that aren't going to be catastrophic if the server reboots and loses what's in memory... it saves a lot of disk writes (especially for things like /tmp and web logs) and accessing it is extremely fast.

Just depends on how your server resources are... if you are short on disk space, then ya... probably makes sense to free that up by moving it to R2.
totally agree with you, that's why I have a feature request to have R2 attachments local cached...
 
Does anybody know a good way to debug cURL 6+7 errors? I can't catch the reason I have it a few times a week seen via ACP. I suspect it has to do with IPV6, but we can't find the real reason for that.
 
Does anybody know a good way to debug cURL 6+7 errors? I can't catch the reason I have it a few times a week seen via ACP. I suspect it has to do with IPV6, but we can't find the real reason for that.
Probably just the standard tools like traceroute and stuff to try and determine where in the route is having problems. The bigger issue is that just like with IPv4, if it’s a switch or router out of your control that is having problems with the underlying network traffic, is well… out of your control. Debugging IPv6 network connectivity doesn’t have anything to do with XenForo or Cloudflare, so the good news is Google is going to be a good resource for anyone on the Internet trying to debug similar issues.
 
I've been wondering this for a while. Does Cloudflare work on sites behind an htaccess password? I want to do some testing with an XF install I've been working on, but haven't activated App for Cloudflare on it.
 
I've been wondering this for a while. Does Cloudflare work on sites behind an htaccess password? I want to do some testing with an XF install I've been working on, but haven't activated App for Cloudflare on it.
It does, yes. However I've found that Cloudflare's Zero-Trust Network Access system is more convenient than HTTP AUTH. Same way the addon can lockdown access to the XF admin area, except do it for the whole domain, rather than just admin.php.

It allows you to do different things like have different users accessing it (without everyone sharing a login/password), doesn't require you to reauth all the time, allows you to bypass it if you are on certain IPs, etc. Some of that you can do with HTTP AUTH if you know how, but it's much easier with Cloudflare's Zero-Trust system.

It's what is being used to keep you out of this site: https://appforcf.com/
 
It does, yes. However I've found that Cloudflare's Zero-Trust Network Access system is more convenient than HTTP AUTH. Same way the addon can lockdown access to the XF admin area, except do it for the whole domain, rather than just admin.php.
I do like that step, and it also stays in spirit with our license (running a second copy for development, privately).
 
Probably just the standard tools like traceroute and stuff to try and determine where in the route is having problems. The bigger issue is that just like with IPv4, if it’s a switch or router out of your control that is having problems with the underlying network traffic, is well… out of your control. Debugging IPv6 network connectivity doesn’t have anything to do with XenForo or Cloudflare, so the good news is Google is going to be a good resource for anyone on the Internet trying to debug similar issues.
I tried to traceroute, but it doesn't allow to traceroute to R2 buckets. Pinging works fine by the way with any data loss for over hours of pinging.
 
I tried to traceroute, but it doesn't allow to traceroute to R2 buckets. Pinging works fine by the way with any data loss for over hours of pinging.
So I tried to enable IPV6 on the server level and so far so good.

In addition, how can I locate these files and delete them from the threads and/or make any action to eliminate the error?

League\Flysystem\FileNotFoundException: File not found at path: attachments/9/9609-f935b1f5e093cecfb3b4bd558ebfb35d.data
src/addons/DigitalPoint/Cloudflare/League/Flysystem/Adapter/R2.php:132
 
A better place for assistance on cleaning up your site's missing attachments would be in the Troubleshooting and problems area (doesn't have anything to do with this addon).

This thread might be helpful:

 
digitalpoint updated [DigitalPoint] App for Cloudflare® with a new update entry:

Rewrote JavaScript to be native (not use jQuery), utilize Sec-Fetch-Site HTTP header, etc.

IMPORTANT for existing users: New functionality requires 1 additional API permissions in order to use the new function. You can go to your Cloudflare API Tokens, edit the token you have and add the following permission:
  • Account.Billing: Read
At this point, you should have a total of 19 permissions for your API token.
  • Added sanity check to make...

Read the rest of this update entry...
 
Top Bottom