I suspected a VM to cause the server crash. I stopped that VM and tried to move it to the second ESX/storage two times. Every time, the main server crashed at 34-35% of the job.
There is a Veeam replication job on that VM, so I did a permanent failover on the original VM and the replicated VM is now running on the other server without any issue since more than a week.
Maybe a corruption in the filesystem or VMDK file.