Tuesday·10·January·2012
Illegal attempt to re-initialise SSL for server (theoretically shouldn’t happen!) //at 02:52 //by abe
After dist-upgrading my main Hetzner server from Lenny to Squeeze, Apache failed to come up, barfing the following error message in the alphabetically last defined and enabled virtual host’s error log:
[error] Illegal attempt to re-initialise SSL for server (theoretically shouldn't happen!)
Well this is not theory but the real world and it did happen — and it took me a while to find out what was wrong with the configuration despite it worked with Lenny’s Apache version.
To avoid that others have to search as long as I had to, here’s the solution:
Look at all enabled sites, pick out those which have a VirtualHost on port 443 defined and verify that all these VirtualHost containers do have their own “SSLEngine On” statement. If at least one is missing, you’ll run into the above mentioned error message.
And it won’t necessarily show up in the error log of those VirtualHosts which are missing the statement but only in the last VirtualHost (or the last VirtualHost on port 443).
To find the relevant site files, I used the following one-liner:
grep -lE 'VirtualHost.*443' sites-enabled/*[^~] | \ xargs grep -ci "SSLEngine On" | \ grep :0
Should work for all sites which have defined just one VirtualHost on port 443 per file.
I suspect that the raise of SNI made Apache’s SSL implementation more picky with regards to VirtualHosts.
Oh, and kudos to this comment to an article on Debian-Administration.org because
it finally pointed me in the right direction. :-)
Tagged as: Apache, CLI, commandline, Debian, error, experience, grep, HTTPS, KMMR, Lenny, Squeeze, SSL, xargs
// show without comments // write a comment
