If you’re using Debian and you want to get rid of Hearbleed bug then it’s not sufficient to upgrade openssl package.
The really important package is libssl1.0.0:
apt-get install libssl1.0.0
It will restart all services which depends on this package.
There is very annoying bug in Open SSL 1.0 which affects curl. When you try to access Tomcat 7 with https with curl you’ll get fancy error:
curl: (35) error:14077438:SSL routines:SSL23_GET_SERVER_HELLO:tlsv1 alert internal error
-k parameter is not working at all
You’re not able to invoke any request against Tomcat 7 with https in default configuration.
The solution is to restrict available ciphers in Tomcat’s https connector:
Restart Tomcat and curl will work.
By adding more features you'll achieve less.
By adding less features you'll achieve more.
Why is this paradox so common in the Sea of Software development?
It’s easier to add more than to add less.
It’s easier to rush than run for a long distance.
It’s easier to promise castle in the air than build it.
It’s easier to blame old code in the system than find understanding.
Is it harder to write test before a code than fix a bug within a bug within a bug?