Hello, In my experience with Kolla-Ansible, I’ve previously reserved specific resources on compute nodes. I hope this experience helps with your issue. It’s possible to run instances on designated CPU threads by configuring the Nova service on each compute host. This allows you to reserve processing resources per host. Here's an example of the configuration I used: ```ini # /etc/kolla/config/nova/<host01>/nova-compute.conf [DEFAULT] cpu_allocation_ratio = 2.0 [compute] cpu_dedicated_set = 10-17 cpu_shared_set = 18-47 ``` This setup reserves threads 10–17 for instances requiring dedicated CPUs, while threads 18–47 are available for shared workloads. The `cpu_allocation_ratio` is set to 2.0 to allow controlled overcommitment. This configuration ensures that a portion of the host’s CPU remains reliably available, even under high load or during evacuation scenarios. To reserve RAM and disk resources, you can use the following options: ```ini reserved_host_memory_mb = <value> reserved_host_disk_mb = <value> disk_allocation_ratio = <value> ``` After updating the configuration, apply the changes with: ```bash kolla-ansible -i multinode reconfigure --tags nova ``` I think this approach allows you to reserve CPU and RAM resources and return them to the OpenStack cloud capacity when needed. Of course, using these options requires careful planning and a deeper understanding. Here are a couple of useful references: - [Red Hat Guide: Tuning Nova Reserved Memory and CPU Allocation](https://docs.redhat.com/en/documentation/red_hat_hyperconverged_infrastructu...) - [OpenStack Documentation: CPU Topologies](https://docs.openstack.org/nova/latest/admin/cpu-topologies.html) Best, Sara