You are here: Supporting Documents > Stratus Cloud Installation Guide > OpenStack Installation and Configuration > Configuring OpenStack for Evacuation and Migration

Configuring OpenStack for Evacuation and Migration

Stratus strongly recommends the following OpenStack compute node configuration to ensure the correct operation for evacuations and migrations. If this configuration is not implemented, evacuations and migrations may fail.

To configure OpenStack for proper evacuation and migration:

  1. On each compute node where NFS shared storage is used, add the following options to the NFS mount entry in /etc/fstab:
  2. Complete the following steps to ensure that your SSH keys are properly configured. For additional information, refer to the following websites:
  3. Use an existing SSH key for /root, or create a new keypair using the following command:
  4. Enter the file in which to save the key:

  5. Enter the passphrase; leave the field blank for no passphrase.
  6. Enter the same passphrase again.
  7. The key fingerprint is 55:45:fc:1f:2d:9b:f5:69:6d:03:5d:ef:2b:50:e8:11 root@<server_name>.<domain>.com

  8. The key's randomart image is:

  9. A key pair is created, both public and private keys:
  10. Enable the Nova user for login using the command: usermod -s /bin/bash nova

  11. Create the folder required by SSH, and move the private key from step 1 into the folder using the following commands:

  12. Repeat steps 2 and 3 on each compute node.

  13. All nodes share the same key pair; do not generate a new one for the other compute nodes. Instead, copy the key for the compute node on which it was created in step 1. For example:

  14. Verify that the key is working properly, using the following commands:

  15. Make sure that all libvirt user IDs and group IDs match across all nodes:

  16. Only on compute nodes that will run the KVM hypervisor, edit /etc/libvirt/qemu.conf; uncomment and change these values:

    Caution: Do not uncomment or change these values on a compute node that will run the KVM-FT hypervisor.
  17. Reboot the compute node.