Hello,

The following strange thing happened.
On a HA Cluster, the 2nd node noticed that the 1st node was down, but.. the first node wasn't down. This resulted in that the 2nd node tried to take over the resources but failed, because the resource was stil in use by the first node. This caused that the first node was left behind in a fuzzy state. I had no other choice to kill the heartbeat service, and reboot the server to solve the issue.

There where no network issues, all hardware is ok.

Are there any bugs know?
Is there a way to avoid it from happening again?

I beleive what you are looking for is STONITH. "Shoot The Other Node In The Head".

Generally, this is set up with a dedicated line, serial or crossover, so that if one
node wants to take over services, it sends a kill signal to ensure that only one
HA master is up at a time. This is designed to avoid contention where two or more
servers think that they should be the master.

Google "Heartbeat and Stonith". There are tons of resources available to guide you
through setting this up, and you can find the guide that most closely follows your current
configuration.

Test the config in a lab environment, and when you are satisfied that it works correctly
you can just migrate the configs to your live environment.