SSL Fails Often; Keep having to restart httpd

Hey guys,

I’ve just installed virtualmin a few days ago, and I’m having problems keeping httpd running on SSL. I have 2 IPs - one shared between several websites, and the other for just one website.

For some reason, httpd repeatedly stops listening on 443, and I have to issue ‘service httpd graceful’ to get it back up again. Once it is up, it works for maybe half a day. Normal http:// works always, but the https:// will go down without warning. Once I issue a restart, everything is fine for awhile again.

None of the shared-IP websites have SSL enabled, although I did TRY to be cheap and let them all SSL for the shared IP (ignoring the warning about older browsers not understanding shared ip ssl), but that is disabled now.

There is also a slight chance I borked the apache config file while adding a few directives for fcgi, but I’m not really sure because normal http:// does still work, and https:// works for a time before crashing.

Is there some usual place I can look for logs, or some diagnostic setting I can do with virtualmin, to help me better track down the cause of httpd suddenly killing its SSL? I’ve tried poking around in /var/logs , but I’m not seeing anything that jumps out at me, and I’m still a n00b anyway.

This is Centos6 BTW.

Howdy,

Hmm, that’s an odd issue!

Whenever that occurs, do you see any errors in the Apache error logs in /var/log/httpd/error_log?

Also, is this a dedicated server, or a VPS? If a VPS, do you know what kind it is? Xen, KVM, OpenVZ?

-Eric

Hey Eric,

Thanks for commenting. I’m using a KVM vps, as I’ve finally realized the potential for instability of OpenVZ. I have looked into /var/log/httpd/error_log, but didn’t see anything, although I was having trouble tracking down when exactly SSL stopped working.

I have it up right now with a program called “monit” waiting for SSL to go down so it can send me an alert, and I can match the alert time with the log entries and hopefully get one more step closer to what the logs might tell me.

I’ll report back if I figure out the solution or just can’t understand the logs, or get lost.

Just one note. The error log showed that I was using # instead of ; for comments when I manually installed mcrypt for php (in the dot-d php directory)… I don’t think that was the problem but it was worth noting maybe.