Just wanted to tell you are not the only one with this problem.
I installed a fresh version of 5.2 yesterday and have the same problem with the Facebook page.
I tried running games in Facebook directly via apps.facebook.com and the games work. But whenever I try using anything else which is native Facebook stuff and uses www.facebook.com as the URL, the white page with status "done". I have tried several web browsers and several OS'es and still the same problem.

Reporting the same problem with clarkconnect 4.1 and 4.3
:huh:

Same thing here. When proxy is used (or in transparent mode with my setup) everything but the root domain works. Just upgraded to 5.2 today so I assume it's a recent change.

Interestingly, if you look at the page source once it has finished 'loading' and is showing a blank screen, you only get:

</body></html>

Oh and also, I don't have the content filter active.

*EDIT* I've had a quick look into this and halfway through my testing the behaviour (and facebooks links) changed slightly so I think this is something on their end.

I have same problem :/

Now i tried http://www.facebook.com/login.php and everything works...

Hi Mario...

I still use ClearOS 51, and this problem still happened right now.If i tried like this : https:// www.facebook.com, then I can enter to the facebook wall but still cannot access to my profile ( and other my friends profile), but like tommy says, i still can play facebook games like zynga poker and other else.

I can only access facebook home and thats all...
(I'm using IpCop)

http://www.outsidaz.org/blog/2009/10/06/facebook-squid-and-internet-standards/

Here's a discussion that may shed some light on the issue. In the comments, people have recommended adding server_http11 to the squid.conf but I believe that only works in squid2.7 whereas ClearOS is using squid2.6. This is starting to get over my head though. :blush:

I just want to say I am confirming this issue with ClearOS 5.2, as I can not access facebook either. I told the proxy to bypass for facebook.com and www.facebook.com, still no luck.

EDIT: Running this update WILL screw up Zarafa webmail. I'm not sure how or why, but updating squid to 3.1 results in "Can't connect to the zarafa server." at the webmail interface. using the commands

/etc/init.d/zarafa-server restart 

/etc/init.d/zarafa-spooler restart

/etc/init.d/zarafa-dagent restart

does appear to resolve the issue... IF you use this process, please be warned it will have unforeseen consequences, and should do it at your own risk.

UPDATE: Facebook works... I updated SQUID. This works for me, but no promises for anyone else...

PLEASE NOTE: You will have to reset any custom changes you made to squid configuration, namely your cache size and proxy mode.

The Process:

mkdir /tmp/squid

cd /tmp/squid

wget http://download.clearfoundation.com/community/timb80/squid-3.1/squid-3.1.3-2.clearos.i686.rpm

rpm -U squid-3.1.3-2.clearos.i686.rpm

cp /etc/squid/squid.conf /etc/squid/squid.conf.backup

cp /etc/squid/squid.conf.rpmnew /etc/squid/squid.conf

/etc/init.d/squid restart

Hope this helps someone.

having strange FB issues myself. Not able to test off my network at the moment. It works for a bit then poof getting the white screen with the body html tag noted above.

edit: i read that mailing list link and it's cryptic. I'm assuming they mean to add the "server_http11" to the squid.conf file? Does the version of squid in ClearOS even support that?

I don't think it does. All the reading I did indicated it only works with squid2.7. I did upgrade to 3.1 with the instructions from the previous post. Seems to be working okay and hopefully I didn't make too much of a mess of the conf. I'll have to have a look through it more thoroughly tomorrow morning after some rest.

If anyone can help point me in the right direction, I'd appreciate it. I've not got much experience with manual updating, but I had to do it once before with CC 4.x. If someone can give me a few pointers, or maybe the term that I should search for, I'd be grateful!
Thanks for helping the newb!

Cody, please refer to my post on the first page of this thread. It has the steps to update to squid 3.1 from squid 2.6, which is currently packaged with ClearOS 5.2.

Issue: Following my steps works UNTIL you restart the ClearOS Box... It blows away the config file and replaces it with a config file which does not allow the squid proxy to start. I figured a "work around" for this:

cd /etc/squid

cp -squid.conf.default squid.conf

/etc/init.d/squid restart

More investigation is needed to figure out how to make changes to the config without it going boom upon reboot...

Facebook works just fine here with 5.2 final release and transparent proxy.

Can someone who is having the problem post the section from /var/log/squid/access.log or /var/log/squid/cache.log?

I would be careful before you all go upgrading to Squid 3.1 as a solution...it was provided as a testing build! ClearOS webconfig will not play nice with it, nor will it necessarily respect the changes in the config file

The link Jared posted suggests is a DNS problem...have you tried emptying the Squid cache yet (from the webconfig)?

Tim,

Facebook didn't work with a fresh install of ClearOS 5.2, as I just installed it from scratch last night. As far as squid and DNS issues, I would have assumed the "Bypass proxy" feature would have resolved that. And honestly, for about 40 minutes, it did. Then poof.

I do see what you mean about Squid 3.1 not playing nicely... In fact, it doesn't play nicely at all.... Alas, it works though. Would you have any thoughts on how it could be "nicely" integrated into the web interface?

Does adding the following the to bypass list help? (rather than just the domain name *.facebook.com)

69.63.176.0/20
66.220.144.0/20
204.15.20.0/21

That seems to cover the IP addresses on the net for facebook.com and fbcdn.net. There maybe more?

Integrating 3.1 into the webconfig isn't an easy task (at least for me) so it's provided as is, but you have to configure it manually from the config file.

In addition do any of the following links work?

http://69.63.189.16
http://69.63.181.12
http://69.63.189.11
http://69.63.181.11

I tried adding 69.63.0.0/24 earlier today as a bypass as well as facebook.com. Neither worked then, but having not done anything more on the issue (just watching this thread) it now works on my install (might be worth noting im still running CC 4.3 community).

Whatever happened, something changed and it now works for me without my applying any fixes or configuration changes.

(Also to your last post Tim, the issue appeared to affect mainly the straight facebook.com domain, i was able to load facebook.com/home.php but not facebook.com/profile.php.... etc - so i presume at a guess it was an issue with some form of HTTP header or a piece of HTML code that wasn't surviving the proxy process. All of those links work for me, but so does the main site now.)

The main problem is with the version used by facebook is http 1.1. If using squid 2.7 stable just by adding the code server_http on. (already tested with other linux distro)

How to implement it in squid 2.6 that used in ClearOS?
Can we do patching http1.1 to ClearOS squid ?

Why is still using squid 2.6 while others have used squid 2.7? With the support of better features and support to http 1.1, it would be better to use squid 2.7 than squid 2.6 .I think it's time for ClearOS to upgrade their version of official squid

Hm, I'm confused. I installed google-chrome on client and everything works fine now, Opera FF Chromium... :blink:

Blaine Bouvier wrote:

More investigation is needed to figure out how to make changes to the config without it going boom upon reboot...

I assume this is because when the box restarts webconfig does some sort of 'reset' to the squid.conf. Does your workaround just remove the default file that it resets to? The "cp -source dest" line is confusing me.

From what I was looking at last night, the conf for squid 3.1 isn't that different from 2.6. I think the only thing I needed to drop was "broken_vary_encoding allow apache" and the associated acl which I believe was a work around for sites using gzip compression.

It would be better if ClearOS make patches for their squid 2.6 to support http 1.1 or provide the latest version of squid update. I think this solution than upgrading to unofficial squid 3.1 that after I tried it many problems there, especially related to webconfig. it always reset my squid.conf to its original ClearOS squid.conf.How did it happened?I'm so confused... :huh:

Jared Smith wrote:

From what I was looking at last night, the conf for squid 3.1 isn't that different from 2.6. I think the only thing I needed to drop was "broken_vary_encoding allow apache" and the associated acl which I believe was a work around for sites using gzip compression.

I don't know honestly, I can look into that and see if that fixes the issue... The cp command was to rewrite what the reboot got rid of, the original new config file.

In /usr/share/system/modules/squid there is an upgrade function and a squid.conf. I haven't rebooted yet but I have replaced the original squid.conf there with one valid for 3.1. I think that should do the trick but I'm just fiddling at the moment (and backing up everything as I go!).

andi micro wrote:

it always reset my squid.conf to its original ClearOS squid.conf.How did it happened?I'm so confused... :huh:

It appears as though the web interface has precedence to the command line config file... As of now, I don't know a solution.

This morning my network can access facebook normally without any change to my clacrconnect 4.1

In the meantime , I am doing is bypassing the following ip address in web proxy :
- 69.63.189.0/24
- 69.63.181.0/24

Hopefully soon, ClearOS Developer will update their squid to the latest version, especially that support http 1.1

From everything i've read around the topic, this problem is caused by the packet headers returned by facebook and a symptom of broken web server. Albeit it rather inconveniently messes up facebook for some users using the squid proxy.

It appears this is due toexcess data / content length defined in the returned headers, which can be alleviated in some cases for Squid v2.7 and v3.x by turning on http1.1 support. (As they have http1.1 support...2.6 [strike]doesn't[/strike] is experimental)
http://www.squid-cache.org/Doc/config/server_http11/

Squid versions took a bit of a developer split at 2.7 versus 3.0 and by comparison these two versions are actually different. Squid 3.1 is now more or less everything that 2.6, 3.0 and now 2.7 was.. and 3.1 has only recently released as 'stable'.

As posted above this problem should correct itself eventually, but I can appreciate the IT admin pain when users start complaining that their favourite social networking site doesn't work anymore

When ClearOS6.0 is released it will be based on RHEL6...which you might like to know is already using Squid 3.1 so it will arrive eventually! (PHP5.3.2 also features...)
ftp://ftp.redhat.com/pub/redhat/rhel/beta/6/i386/os/Packages/squid-3.1.0.15-2.el6.i686.rpm

had my wife complaining about this yesterday but today it seems to work fine so im freewheeling on....

maybe was facebook problem yesterday... :silly:

Well, it's not just Facebook. I use an online family calendar program, called Cozi. I could not access that unless I set my computer to DMZ. Once I did that, FB AND Cozi started working on all computers. I removed the DMZ, and things are still working. Did that just cause a cache reset of some sort?
I don't really NEED faster internet, I already have RR Turbo, so speed isn't the issue. If there is something I can do to fix this as it is, I'm willing to do it. Whoever posted about adjusting the amount of time Squid holds on to the DNS cache or something (I'd have to go back and look it up) does that work? If the system reboots, would it still work?
My box generally is up for months without interruption, but I just upgraded to ClearOS less than week ago, from ClarkConnect 5.1.
Thanks all! I appreciate the help!
Cody