If the performance of your VMware Integrated OpenStack deployment deteriorates, you can adjust the settings for various VMware Integrated OpenStack components.
Because VMware Integrated OpenStack is deployed in many different environments, recommended values for performance parameters are not given. Adjust these parameters based on your environment and the resources available to you.
The parameters in the following table are located in the custom.yml file. You must run the viocli deployment configure command before your changes can take effect.
| Name | Default Value | Description | Usage |
|---|---|---|---|
| nova_ram_allocation_ratio | 1.5 |
Allocation ratio of virtual memory to physical memory for CPU filters |
Increase the value to address the following error in nova-placement-api.log: InvalidAllocationCapacityExceeded: Unable to create allocation for 'MEMORY_MB' on resource provider |
| nova_cpu_allocation_ratio | 16 |
Allocation ratio of virtual CPUs to physical CPUs for CPU filters |
Increase the value to address the following error in nova-placement-api.log: InvalidAllocationCapacityExceeded: Unable to create allocation for 'VCPU' on resource provider |
| nova_disk_allocation_ratio | 0.0 |
Allocation ratio of virtual disk space to physical disk space for disk filters |
Increase the value to address the following error in nova-placement-api.log: InvalidAllocationCapacityExceeded: Unable to create allocation for 'DISK_GB' on resource provider |
| keystone_token_expiration_time | 7200 |
Time in seconds that a token remains valid |
Increase the value to address the following error in various log files: WARNING keystoneclient.middleware.auth_token [-] Authorization failed for token |
| haproxy_nova_compute_client_timeout | 1200s |
Time in seconds that the load balancer waits for a response from Nova acting as a client |
Increase these values to address the following error in nova-compute.log: Exception during message handling: Gateway Time-out (HTTP 504) |
| haproxy_nova_compute_server_timeout | 1200s |
Time in seconds that the load balancer waits for a response from Nova acting as a server |
|
| haproxy_cinder_client_lb_timeout | 300s |
Time in seconds that the load balancer waits for a response from Cinder acting as a client |
Increase these values to address the following error in cinder-volume.log: VolumeBackendAPIException: Bad or unexpected response from the storage volume backend API |
| haproxy_cinder_server_lb_timeout | 300s |
Time in seconds that the load balancer waits for a response from Cinder acting as a server |
