I have a lot of "GET /" 400 entries in my log for only one ssl virtual host. On my server there are many http virtualhosts and some ssl. All ssl virtualhost are ip-based, and some of the http virtual hosts are name virtualhost. I notice that only one ssl virtual host (the highest address) logs the entries in access_log (not in error_log !). Also the server_status page reports the connections (a lot): ip171.fmal.com test.fmal.com GET / ip171.fmal.com has address 151.1.219.171 test.fmal.com is an alias for ip171.fmal.com. Here are the listen.conf entries: Listen 151.1.219.161:80 Listen 151.1.219.164:80 Listen 151.1.219.165:80 Listen 151.1.219.166:80 Listen 151.1.219.167:80 Listen 151.1.219.169:80 Listen 151.1.219.170:80 <IfDefine SSL> <IfDefine !NOSSL> <IfModule mod_ssl.c> Listen 151.1.219.161:443 Listen 151.1.219.162:443 Listen 151.1.219.163:443 Listen 151.1.219.171:443 </IfModule> </IfDefine> </IfDefine> If I delete the entry for ip171, the 400 error goes to log for virtual host on ip 163.
Partially resolved: if I took away all ssl virtualhost definitions, I can see this message in error_log: ip171.fmal.com - - [20/Dec/2006:12:10:55 +0100] "GET / HTTP/1.0" 200 44 "-" "Apache/2.2.3 (Linux/SUSE) (internal dummy connection)" Reading manuals I understand this is a way to kill process exceeding MaxSpareServers. So I tried to higher this value from 12 to 48 and the error messages decreased greatly. The minor bug is that they are sent to ssl virtual host and so are not understood.
*** This bug has been marked as a duplicate of 39653 ***