cVM Deployment Failure

Symptom

Error log: cVM deployment failed. The error message "No valid host found" is displayed.

Root Cause Analysis

The compute node fails to detect secure memory, or the secure memory information on the compute node has not synchronized to the controller node.

Solution

Check the OpenStack resource usage by following instructions in Querying VM Information. Check whether the storage and CPU resources are sufficient and whether the secure memory information is obtained. If the secure memory information is missing, check the following items in sequence:

1. Check whether the TMM driver is installed.

   lsmod | grep tmm_driver

2. Check whether virtCCA is enabled. The value 1 indicates that virtCCA is enabled.

   cat /sys/kernel/tmm/virtcca_enabled

   

3. Check whether libvirt identifies virtCCA capabilities.

   virsh domcapabilities | grep virtcca

   

If any of the preceding items fails, perform the following operations.

1. Install the TMM driver.

   insmod tmm_driver.ko

2. Clear the libvirt node information cache. Restart the libvirt process and check that libvirt can identify virtCCA capabilities.

   rm -rf /var/cache/libvirt/qemu/capabilities/*
   systemctl restart libvirtd
   virsh domcapabilities | grep virtcca

3. Restart httpd and nova-compute.

   systemctl restart httpd openstack-nova-compute