# nova hypervisor-show 57
+---------------------------+------------------------------------------+
| Property | Value |
+---------------------------+------------------------------------------+
| cpu_info_pcpus | 4 |
| cpu_info_topology_cores | 24 |
| cpu_info_topology_threads | 48 |
| current_workload | 0 |
| disk_available_least | None |
| free_disk_gb | 382 |
| free_ram_mb | 59779 |
| host_ip | 172.28.0.3 |
| hypervisor_hostname | domain-c7(cluster-01).1722fc2b-fb14-4fc3 |
| | -878c-e57cd2a942d4 |
| hypervisor_type | VMware vCenter Server |
| hypervisor_version | 6005000 |
| id | 57 |
| local_gb | 537 |
| local_gb_used | 155 |
| memory_mb | 87939 |
| memory_mb_used | 28160 |
| running_vms | 7 |
| service_disabled_reason | None |
| service_host | vmware-nova-compute002 |
| service_id | 79 |
| state | up |
| status | enabled |
| vcpus | 48 |
| vcpus_used | 15 |
+---------------------------+------------------------------------------+
The hypervisor memory from openstack command line:
1. memory_mb: 87939 (85.88GB)
2. memory_mb_used: 28160 (27.5GB)
3. free_ram_mb: 59779 (58.38GB)
The cluster in vCenter (hypervisor corresponding) memory:
1. Total: 95.98GB
2. Used: 13.61GB
3. Free: 82.38GB
Why OpenStack's hypervisor memory is inconsistent with vCenter's cluster, who know the reason for this problem?
Thanks!
I suppose the point 3 can help us to identify the issue:
- did your colleague delete VMs directly from vSphere? or he deleted instances from VIO?
Hi Jeff,
Please provide the following:
1) Which version of VMware VIO are you running ?
2) Which kind of hardware nodes are the cluster composed of?
3) Is it running a virtual storage appliance on each node?
4) Is there any VM running into the cluster, not part of VMware VIO?
Please, let me know.
Cheers,
Domenico
Dear Domenico, thanks for your reply.
1) I am not using VMware VIO, but the community Mitaka version of the nova driver.
2) There is only one x86 server in the cluster "cluster-01" and the ESXi version is 6.5
3) The VSA or vSAN program is not running on this node.
4) In addition to OpenStack managed virtual machines, there is only one vCenter VM (8CPU, 8G Memory) .
so... just to clarify the design, before starting with math:
1) Cluster01 contains only one host: tot MB of the host?
2) vCenter Server is it installed in the above single host? no management cluster?
3) "nova hypervisor-show 57" stats "running_vms | 7 ": could you tell us, total ram used by these VMs? It should be fine a screenshot of host summary
Cheers,
Domenico
1) Sorry, I was wrong, this cluster has two x86 servers, each 48G memory, the total is 96G.
2) This is a test environment, no management cluster. The vCenter VM memory is configured as 8192M (8G) and 2211MB is used.
3) This test environment belongs to one of my colleagues, he has deleted the virtual machine to do other things, so now there is no virtual machine memory usage data.
I think the memory_mb that get from OpenStack cli should be equal to 88G (96G-8G), but the actual value is 85.88G. Is it necessary to subtract the memory from the ESXi program in each ESXi host?
Domenico, thank you very much!
I suppose the point 3 can help us to identify the issue:
- did your colleague delete VMs directly from vSphere? or he deleted instances from VIO?
You are right. If the virtual machines are deleted in vsphere, will cause the memory data to be used inaccurate.
We should remove the virtual machines through the OpenStack cli (or calling API) .
But now I want to know why the total amount of memory is inconsistent.
Thanks!
honestly, it is difficult to understand this difference, without any live investigation...
Does the cluster based on the two Esxi, currently, host only vCenter Server? Are there any other VMs (like Openstack management vMs)
Can you post the ESXTOP command result, for memory only (Once in the esxtop screen press ‘m’ to display memory statistics.), for each ESXi?
Cheers,
Domenico
Sorry, our test environment has been removed, so there is no way to provide the command result.
Thanks again for your help, Domenico!