but if I click the link to “Module Index” it goes to /etc/php.ini
which reflects the same behavior as a top level php page and sessions under 'http" (but https) works as expected.
? FYI this is a new server installation and we upgraded php along the way after installing VirtualMin. I have run the reload modules command already a couple of times in the process, but this behavior does not change.
server settings appear OK except we have this report:
Default IP address is set to 64.151.72.192, but the detected external address is actually 64.151.71.252. This is typically the result of being behind a NAT firewall, and should be corrected on the module configuration page.
I’m not exactly where to change that or even if it is wise: eth0 is 64.151.72.192 and 64.151.71.252 is what ServePath refers to as the VLAN/Router address for the machine and “should not be used by you for anything” though I do certainly need to use it for the SPF text records at SOA because that is the IP from which mail comes…
Good news is we solved this, bad news is my support team (which includes me) are not good with documentation on solutions. I think it was as simple as
setting php to execution to FCGid (as virtual domain owner “bigfish”)
set the sessions directory to /home/bigfish/tmp (so that PHP, acting as “bigfish” will definitely have permisisons to read and write anywhere in /home/bigfish
I also set the errors log to /home/bigfish/etc/php5/php-error.log
for all domains onthe box because previously all php errors, from all domains were being aggregated into a single /etc/php/errors log.
#3 has nothing to do with your problem, but may help sorting things out. because you can tail that log and see only errors pertaining to that one domain.
Disclaimer: my wizard guys may have done some thing else too and simply told me “I fixed it.” But not how.