We're using some Atlassian products: Jira, Confluence and Crowd. Crowd provides SSO for comfortable switching between this applications. Also we're using subversion, which was successfully integrated with crowd (and SSO as well) using this instructions:
https://confluence.atlassian.com/display/CROWD/Integrating+Crowd+with+Apache
https://confluence.atlassian.com/display/CROWD/Integrating+Crowd+with+Subversion
So, subversion can create or accept SSO

We want the same thing for Zabbix, so I've configured Zabbix for using http-authentication, install Crowd apache connector 2.0.2 (see the first link from above).
Apache configuration file looks like this:
<VirtualHost *:80>
DocumentRoot /var/www/zabbix
<Directory />
Options FollowSymLinks
AllowOverride None
AuthName "Atlassian Crowd"
AuthType Basic
AuthBasicProvider crowd
CrowdAppName <crowd_app_name>
CrowdAppPassword <crowd_app_password>
CrowdURL <crowd_url>
CrowdAcceptSSO on
CrowdCreateSSO on
Require group <restriction_group>
Require valid-user
</Directory>
</VirtualHost>

What we expect:
1. After log in to Zabbix it creates SSO cookie, so, if we go to Jira we must be already logged in.
2. After log in to Jira it creates SSO, so when we go to Zabbix it must accept SSO cookie and login automaticly.

What we get:
1. Zabbix creates SSO succesfully, so when we go to Jira evrything is fine, we are already logged in.
2. When we log in to Jira it creates SSO (we're automaticly logged in into confluence and svn), but when we try to go to Zabbix it does not accept SSO cookie, and it consider us as guest (see attachment)
So, to access Zabbix we have to user browser's anonymous mode.