VM can go zombie and cause network issues

A VM is deleted from Coin and opennebula, but remains on the host as if the call to destroy it never got to it from opennebula. The biggest issue in this is that the DNS/MAAS? is also "unaware" of it and gives the same IP to new VM resulting in 2 VMs sharing the same IP address. 

This has happened twice now. It renders the IP fully unusable causing network resets every few seconds on the VMs. The IP pool seems to be recycled which leads to this causing quite a number of failures like:
https://testresults.qt.io/coin/logview/qt/qtdeclarative/869e92bf30c40ba5d83208d7fced3f3dae0f2bec/LinuxRHEL_8_2x86_64LinuxRHEL_8_2x86_64GCCqtci-linux-RHEL-8.2-x86_64-50-948d91Sccache/9e4bb9b501036a496fdde5eefd1cb64db95f501a/build_1627388472 

Ssh to the IP goes randomly to either one and complains host key verification, giving clue that there are multiple VMs behind that and allowing debugging to find the VM. The cause is yet unknown.