Jump to content
TrinityCore

Cloudflare is wigging things out.


Recommended Posts

Over the years, I noticed that Cloudflare is one of those things that either works really well -- or causes lots of headaches.  Unfortunately for your forums, Cloudflare is being a real <expletive word here> and I recommend disabling it completely.  The forum and community currently are both small enough it really wouldn't benefit much from the so-called caching which they provide.


ISSUE #1
Following image depicts the error I am getting whenever I try to post a reply, or a new topic.

http://i.imgur.com/a5N9E2r.png

 

Additionally, it makes it appear like I am spamming the forum with multiple responses when I have only clicked submit/reply/etc ONCE.


ISSUE #2
There is another issue as well with the documentation portal that has been giving a 502 bad gateway response for several days now.

http://i.imgur.com/Hm8VDtx.png   is what everyone is getting right now when attempting to go to http://collab.kpsn.org/display/tc/Databases+Installation  


ISSUE #3
Links seem to work on this site without any real problem, taking me where they appear they should be taking me -- except for pagination.
When clicking on a different page # in threads having multiple pages. The page does not change. However if I right click the page #, and open in a new tab, the page displays.  This could also be linked to Cloudflare as the source of this problem as I am sure your forums are working correctly.

ISSUE #4
Certain Ajax commands will spin forever, such as the Create dropdown at the top. While the drop-down action works, clicking either Topic or File Download result in a process that does not terminate or give the expected result such as opening the new Topic window/page never happens.  I would also blame Cloudflare for this.


Cloudflare DNS is an utter pile of trash, and they always blame the website using their service as being 'down' when in fact it is their <expletive> service that is buggering everything up.  It is also slower.  In my tests the TTFB with Cloudflare hosted stuff was around 200 times higher than without using their DNS. I was able to trace this to major DNS issues on their end which result in the typical 'community.trinitycore.org' is down messages, when in fact your server is NOT down.  (kpsn.org is, but this site is up).

Link to comment
Share on other sites

working on stuff so short answers atm, can expand later if you want.

 

#1 : not cloudflare, having issues with php-fpm timeouts.

#2 : wiki was being moved to new hosting, and to new url, www.trinitycore.info. after the move there was an upgrade, some of our plugins were not licensed for new version. all should be working well now.

#3 : pagination was part of #1, php-fpm + nginx work perfectly for the two dozen other sites but not this one 

#4 : exact same reason as #1+#3

 

im literally working on this as much as I can around my work to have it all fixed ASAP.

Link to comment
Share on other sites

Sounds good brother.   If you guys need an extra on the crew that's been in the scene for a while (since WAD), let me know.

Currently, just working to figure where to even begin to start patching up this mess.

Also, could you please let me know why this has gone the direction of patching the exe again, as if memory serves, that stopped being necessary some time ago.  The patching (as usual) works great.  I recall there being something about blizz allowing emu's and adding support for them by not changing the server in the wtf file (used to be realmlist.wtf, now WTF/Config.wtf ), and honoring whatever that setting was.

Thanks again --- and also nice to see this project continues even though MangOS aborted.

  • Upvote 1
Link to comment
Share on other sites

So regarding the patching:

1. Parts of the server adress is hardcoded into the client. This has to be removed, so we can set it freely.
2. Some packets are encrypted with RSA. Since we can't break blizz' key, we had to patch our own into the client.
3. Some parts of the authentication process are in external modules. During the loading process some verification takes place, which had to be removed too.

I guess that should sum it up.

Regards,
Carbenium
 

  • Upvote 1
Link to comment
Share on other sites

Fair enough.  I know this might be a bit much to ask, but is there a way to apply the patch (similar to the old patching system) where the modification kept the hash of the file the same ?

Way back around the era of WAD's wowemu which started this whole thing,the wow executable needed to be patched as well, however the file signature remained in tact.

Later on a few years into development, connection to certain emu servers no longer required any client side patching.  I thought this was due to blizz finally allowing emu's unofficially by respecting alternate server addresses placed inside realmlist.wtf .  It would first try blizz official, then failing that use the one in the realmlist.wtf file.  Prior to that (in client patch required days with wowemu and such), any attempt to connect to an alternative server would result in wow overwriting the realmlist.wtf with the 'official' server.

What is the current problem in the exe that prevents this ?  Does it no longer roll-over to respect the specified server ?   I did notice that if I attempt to use the wow.exe unmolested, i get an expediant disconnect which is slightly different than a failed credential check with blizz, so it appears that it does still respect the wtf file (new location  WTF/Config.wtf).

Thanks in advance :)

Link to comment
Share on other sites

 Share

  • Recently Browsing   0 members

    No registered users viewing this page.

×
×
  • Create New...