Host

What is Host?

A host provides compute, network, and storage resources for VM instances. Hosts are core resources on the Cloud. VM instances are running on hosts.

图 1. Host


Add a Host

On the main menu of ZStack Cube Ultimate, choose Resource Center > Hardware > Computing Facility > Host. On the Host page, click Add Host. Then, the Add Host page is displayed.

ZStack Cube Ultimate allows you to add a host by one of the following methods:
  • Scan-based Addition
  • Manual Addition
  • Template Import

Scan-based Addition

You can use this method to add hosts when the distributed storage is of the ZCE-X or ZStone type. To use this method, follow these steps:
  1. Detect Host

    After you select Scan-ban-based Addition, the platform automatically detects hosts and displays their hardware information. You can select at least one detected host to add it to the platform. If no desired host is found, click Rescan in the upper-right corner to initiate another detection.

    After all desired hosts are selected, click Next to proceed to the Configure host page.

  2. Configure Host
    On the Configure Host page, set the following parameters:
    • Name: Enter a name for the host.

      The name must be 1 to 128 characters in length and can contain Chinese characters, letters, digits, spaces, hyphens (-), underscores (_), periods (.), parenthesis (), colons (:), and plus signs (+) and cannot begin or end with spaces.

    • Root Password: Set a password for the root user of the host.
    • SSH Port: Enter the SSH port of the host. Default: 22.
    • Node Type: Select a node type for the host. You can add the host as a hyper-converged node or a compute node.
      • If you select Hyper-converged Node, the host functions as bot a compute and storage node, providing computing and storage services. The platform automatically formats all disks, except the system disk, on the host. All existing data on these disks are permanently erased and cannot be recovered. Proceed with caution.
        • Cluster: Select the cluster where the host is located.
          Note:
          • Hosts that apply the KVM virtualization technology can be added only to KVM-based clusters.
          • Hosts of the X-Dragon architecture can be added only to X-Dragon-based clusters.
        • Primary Storage: Displays primary storage deployed through Bootstrap. Primary storage deployed through other methods does not support expansion.
        • Storage Pool: Select a storage pool that the host belong to as a storage node.
          • You may fail to specify a storage pool if the distribute storage version is too low. To ensure the node can properly provide storage service, we recommend that you
            • Upgrade the distributed storage to the latest version.
            • If a upgrade is not feasible due to business requirements, you can enter the distributed storage page to add the storage node.
      • If you select Compute Node, the host functions only as a compute node and provide computing services.
        • Cluster: Select the cluster where the host is located.
    • Scan Host IOMMU Setting: Optional. Choose whether to scan the IOMMU setting of the host. The IOMMU setting is used in scenarios such as peripheral device passthrough and virtualization. By default, the host IOMMU setting is not scanned.
      Note: If you choose to scan the host IOMMU setting, note the following:
      • Scanning the host IOMMU setting will traverse available GPU devices on the host and physical NICs that can be virtualized. Before you scan the setting, make sure that Intel VT-d or AMD IOMMU is enabled in the host BIOS.
      • Before you can use GPU passthrough, vGPU virtualization, and SR-IOV, scan the host IOMMU setting.
      • If you are scanning the host IOMMU setting for the first time, restart the host to make the setting take effect.
    • Intel EPT Hardware Assist: Optional. Choose whether to enable Intel EPT hardware assist for Intel CPUs to improve the CPU performance. By default, this feature is enabled.
      Note:
      • If the CPU model of your Intel server is too old, you might fail to create VM instances or cannot open the console properly of the newly created VM instances. In this case, disable the Intel EPT hardware assist feature.
      • Disabling Intel EPT Hardware Assist will lower the VM performance.
  3. Preview

    Preview host configurations and click OK to confirm to add hosts.

Manual Addition

With this method, you can specify an IP address or an IP range to add hosts. You can add up to 500 hosts at a time. To use this method, set the following parameters:
  • Name: Enter a name for the host.

    The name must be 1 to 128 characters in length and can contain Chinese characters, letters, digits, spaces, hyphens (-), underscores (_), periods (.), parenthesis (), colons (:), and plus signs (+) and cannot begin or end with spaces.

  • Description: Optional. Enter a description for the host.
  • Virtualization Technology: Select the hypervisor type of the server. You can select KVM or X-Dragon.
    • If the server applies the KVM virtualization technology, select KVM.
    • If the server applies the X-Dragon architecture, select X-Dragon.
  • Tag: Optional. Attach one or more tags to the host.
  • Cluster: Select the cluster where the host is located.
    Note:
    • Hosts that apply the KVM virtualization technology can be added only to KVM-based clusters.
    • Hosts of the X-Dragon architecture can be added only to X-Dragon-based clusters.
  • Host Info: Configure the host information such as the host IP and SSH port according to your actual needs. You can configure the host information by entering an IP address or an IP range.
    • If you select the IP Address method, set the following parameters:
      • IP Address: Enter the IP address of the host.
      • SSH Port: Enter the SSH port of the host. Default: 22.
      • SSH Username: Enter the SSH user name of the host. Default: root. You can enter a user name according to your actual needs.
      • SSH Password: Enter an SSH password for the user name.
    • If you select the IP Range method, set the following parameters:
      • IP Range: Enter the planed IP range for the host.
      • SSH Port: Enter the SSH port for the host. Default: 22.
        Note: Make sure that all hosts in the specified IP range share the same SSH port.
      • SSH Username: Enter the user name of the host. Default: root. You can enter a user name according to your actual needs.
        Note: Make sure that all hosts in the specified IP range share the same user name.
      • SSH Password: Enter a password for the user name.
        Note: Make sure that all user name in the specified IP range share the same password.
  • Scan Host IOMMU Setting: Optional. Choose whether to scan the IOMMU setting of the host. The IOMMU setting is used in scenarios such as peripheral device passthrough and virtualization. By default, this feature is disabled.
    Note: If you choose to scan the host IOMMU setting, note the following:
    • Scanning the host IOMMU setting will traverse available GPU devices on the host and physical NICs that can be virtualized. Before you scan the setting, make sure that Intel VT-d or AMD IOMMU is enabled in the host BIOS.
    • Before you can use GPU passthrough, vGPU virtualization, and SR-IOV, scan the host IOMMU setting.
    • If you are scanning the host IOMMU setting for the first time, restart the host to make the setting take effect.
  • Intel EPT Hardware Assist: Optional. Choose whether to enable Intel EPT hardware assist for Intel CPUs to improve the CPU performance. By default, this feature is enabled.
    Note:
    • If the CPU model of your Intel server is too old, you might fail to create VM instances or cannot open the console properly of the newly created VM instances. In this case, disable the Intel EPT hardware assist feature.
    • Disabling Intel EPT Hardware Assist will lower the VM performance.
图 1. Manual Addition


Import Template

With this method, download the CSV template file provided by the platform, fill in the host information according to the specified format, and upload the file. You can add hosts in bulk with this method. The procedures are as follows:
  1. Download the configuration template.

    Click Download Template to download the configuration template file in CSV format.

    图 2. Configuration Template File


  2. Fill in the host information according to the specified format.

    The configuration template contains a header and a row of sample. You can delete or overwrite the sample as needed.

    Set the following parameters in the template:
    • Name: Enter a name for the host. If null, the Cloud allocates a name (HOST-host IP address) to the host by default.
    • Description: Optional. Enter a description for the host.
    • Cluster: Enter the UUID of the cluster where the host is located.
      Note: You can add only KVM-based clusters in the template.
    • Host IP: Enter an IP address or an IP range for the host.
      • IP Address: Enter the specified IP address of the host.
      • IP Range: Enter the specified IP range to bulk add hosts.
        Note:
        If you enter multiple IP ranges for the host, use commas (,) to separate the ranges. ^ means that the IP range is not included. For example,
        192.168.0.1-192.168.0.100,^192.168.0.3-192.168.0.5
    • Scan Host IOMMU Setting: Optional. Choose whether to scan the IOMMU setting of the host. The IOMMU setting is used in scenarios such as peripheral device passthrough and virtualization. By default, this feature is disabled.
      Note: If you choose to scan the host IOMMU setting, note the following:
      • Before you can use GPU passthrough, vGPU virtualization, and SR-IOV, scan the host IOMMU setting.
      • Scanning the host IOMMU setting will traverse available GPU devices on the host and physical NICs that can be virtualized. Before you scan the setting, make sure that Intel VT-d or AMD IOMMU is enabled in the host BIOS.
      • If you are scanning the host IOMMU setting for the first time, restart the host to make the setting take effect.
    • Intel EPT Hardware Assist: Optional. Specify whether to enable Intel EPT hardware assist for Intel CPUs to improve the CPU performance. This feature is enabled by default.
      Note:
      • If the CPU model of your Intel server is too old, you might fail to create VM instances or cannot open the console properly of the newly created VM instances. In this case, disable the Intel EPT hardware assist feature.
      • Disabling Intel EPT Hardware Assist will lower the VM performance.
    • SSH Port: Enter the SSH port for the host. If null, the Cloud will use port 22 by default.
    • SSH User Name: Enter the SSH user name of the host.
    • SSH Password: Enter an SSH password for the user name.
  3. Upload the configuration file.

    After you fill in the host information, verify that the grammar is correct, and then upload the file to the Cloud.

    图 3. Import Template


  4. Add hosts by using the template.

    Click OK. Then, the Cloud add hosts according to the configuration file.

Manage a Host

On the main menu of ZStack Cube Ultimate, choose Resource Center > Hardware > Computing Facility > Host. Then, the Host page appears.

The following table lists the actions that you can perform on a host.
Action Description Preconditioned Host Status
Add Host Add one or more hosts. /
Edit Host Edit the name and description of a host. /
Enable Host Enable a host. Disabled
Disable Host Disable a host.
Note: Disabling a host does not affect the resources already running on the host. However, the host is no longer available as a candidate when you apply for new resources.
Enabled
Reconnect Host Reconnect a host.
Note: In most cases, host reconnection can be implemented when the host configurations are updated. For example, if the memory or volume capacity of a hots is updated, you can reconnect the host to update the database.
ALL
Enter Maintenance Mode Enter the maintenance mode. If a host is in maintenance mode, you can power off the host or troubleshoot failures.
  • If the primary storage is of the LocalStorage type, VM instances on a host are stopped when the host enters maintenance mode.
  • If the primary storage is of a shared storage type, VM instances on a host are automatically migrated when the host enters maintenance mode.
Note: If the primary storage is of a shared storage type, you can configure the policy on failures of VM migration triggered when the host enters maintenance mode.
On the main menu of ZStack Cube Ultimate, choose Settings > Platform Setting > Global Setting > Advanced > Host, and configure the Policy for VM Migration Failure in Entering Host Maintenance Mode setting:
  • StopVmOnMigrationFailure: When the host enters maintenance mode, VM instances are automatically migrated to another host. If a VM instance fails to be migrated, the VM instance is forcibly stopped.
  • JustMigrate: When the host enters maintenance mode, VM instances are automatically migrated to another host. If a VM instance still runs on the host, the host fails to enter the maintenance mode.
Connected
Add Bond Aggregates the physical NIC ports on the host, so as to realize a high availability or a load balance. Supports two bond mode: active-backup and LACP.
  • The active-backup mode allows you to bind 1-2 physical NIC ports, and we recommend that you bind 2 ports, one acting as the primary port and the other as the secondary port. By default, all network flows are handled by the primary port. Once the primary port fails, the secondary port replaces it to handle network flows.
  • The LACP mode allows you to bind 1-8 physical NIC ports, and we recommend that you bind at least 2 ports. Bound ports share the same speed and duplex setting. The network flows are evenly sent to these ports to handle, thus realizing a load balance. A bond of this mode determines its network export based on a hash computation. Three hash policies are supported: layer2+3, layer 3+4, and layer2.
    • layer2+3: Picks out a port to send data packets based on the hash computation on the source MAC address, destination MAC address, and IP address.
    • layer3+4: Picks out a port to send data packets based on the hash computation on the IP address and port. TCP/IP stacks are supported.
    • layer2: Picks out a port to send data packets based on the hash computation on the source MAC address and destination MAC address.
Note: Make sure that the NIC ports added to a same bond have a same speed rate.
ALL
Power on Power on a host. Powered off
Power off Power off a host.
Note:
  • Powering off a host may cause VM instances, VPC vRouters, SLB instances, and other resources running on the host powered off. To ensure your business stability and security, we recommend that you make the host enter the maintenance mode before you perform the operation.
  • If the host is used also as a Monitor node of Ceph primary storage, powering it off will stop the monitor node service of the primary storage, affect the stability of the storage cluster, and thus may cause data loss.
  • If the host is designated as a migration server, powering it off will make migration jobs running on the host end up failure.
Powered on/Unknown
Restart Restart a host.
Note:
  • Restarting a host may affect normal running of VM instances, VPC vRouters, SLB instances, and other resources on the host. To ensure your business stability and security, we recommend that you make the host enter the maintenance mode before you perform the operation.
  • If the host is used also as a Monitor node of Ceph primary storage, restarting it will stop the monitor node service of the primary storage, affect the stability of the storage cluster, and thus may cause data loss.
  • If the host is designated as a migration server, restarting it will make migration jobs running on the host end up failure.
Powered on/Unknown
Enter Web Terminal Enter a web terminal of a host to perform operations on the host. Powered on/Unknown
Attach Tag Attach a tag to one or more hosts.
Note:
  • You can attach a maximum of 50 tags to a resource. The number of resources to which a tag can be attached is not limited.
  • You can attach one or more tags to one or more resources.
  • A tag created by a tenant can be attached only to the resources owned by the tenant. The admin can attach a tag to all resources.
  • The admin can detach or delete tags created by a tenant.
  • You can sort tags attached to resources by attachment time or tag name. Tag names are sorted based on symbols, digits, Chinese characters, and English letters in descending order. To configure the sorting method, go to the main menu of ZStack Cube Ultimate, choose Settings > Platform Setting > Global Setting > Basic, and then configure the Tag Sort Method setting. By default, resources are sorted by tagName.
ALL
Detach Tag Detach a tag from a host.
Note:
  • The admin tags are created and owned by the admin. The tenant tags are created and owned by tenants.
  • A tag created by a tenant can be attached only to the resourced owned by the tenant. The admin can attach a tag to all resources.
  • The admin can detach or delete tags created by a tenant.
  • Tags created in a project belongs to the project. All project members, including the project admin, project managers, and normal project members, can perform actions on the tags.
  • You cannot change the owner of a tag.
  • If you change the owner of a resource, all tenant tags attached to the resource are detached. The admin tags are not affected.
  • After a seamless upgrading of the cloud platform, tags that were created are automatically updated and displayed in the latest format. If an error or exception occurs, refresh the browser or create a new tag.
  • You can detach a batch of tags from a resource or detach a tag from a batch of resources.
  • A tenant can only detach tags from resources owned by the tenant. The admin can detach tags from all resources.
ALL
Modify IPMI Info Modify the IPMI username and password of the host.
Note: You can only modify the IPMI info of a host that is in IPMI Unmanaged or Unknown power status.
Unknown
Update SSH Password Update the SSH password of a host. After the password is updated, the host is automatically reconnected. /
Delete Host Delete a host.
Note: If you delete a host, note that:
  • Deleting a host stops all VM instances running on the host.
  • If a LocalStorage primary storage is attached to the cluster to which the host belongs:
    • Deleting a host also deletes all VM instances and volumes on the host.
    • If you add the deleted host to the cloud platform, the system redeploys the host. If you did not backup the database, the related business data cannot be recovered.
  • If a shared primary storage is attached to the cluster to which the host belongs:
    • If you disable HA for VM instances:

      Deleting a host stops the VM instances running on the host.

    • If you enable HA for VM instances:

      Deleting a host stops the VM instances for which HA is enabled and then migrates the VM instances to other hosts where resources are available. Then the VM instances are automatically started. During the process, data security is not affected. If host resources are insufficient, the VM instances are stopped upon the host deletion.

    • Deleting a host does not affect data volumes stored on shared storage.
/

Host Details

pNUMA Topology

pNUMA Topology: A pNUMA topology (physical NUMA topology) is the topology of the host NUMA nodes predefined by the CPU vendor based on the host NUMA architecture.

Concepts

  • NUMA: Non-uniform memory access (NUMA) is a computer memory design where the memory access time depends on the memory location relative to the CPU. Under NUMA, a processor can access its own local memory faster than non-local memory and thus improves VM performance.
  • pNUMA node: A pNUMA node (physical NUMA node) is a host NUMA node predefined based on the host NUMA architecture. It is used to manage the CPUs and memory of the host. A pNUMA node mainly consists of pCPUs and local memory.
  • vNUMA node: A vNUMA node (virtual NUMA node) is generated by passing-through associated pNUMA nodes via CPU pinning. It is used to manage the CPUs and memory of a VM instance. A vNUMA node mainly consists of vCPUs and local memory.
  • vNUMA topology: A vNUMA topology (virtual NUMA topology) is the topology of VM NUMA nodes generated by passing-through associated pNUMA nodes via CPU pinning.
  • Local memory: Local memory is the memory that a CPU (pCPU or vCPU) accesses through the Uncore iMC (Integrated Memory Controller) of the same NUMA (pNUMA or vNUMA) node. Compared with accessing non-local memory, accessing local memory has lower latencies.

Fundamentals

After you add a host on ZStack Cube Ultimate, you can view the pNUMA topology of the host. You can also configure vNUMA for VM instances running on the host according to the pNUMA topology.

vNUMA configurations are achieved through CPU pinning, which assigns the vCPUs of a VM instance to specific pCPUs of the host. In vNUMA configurations, all vCPUs of the VM instance are pinned to pCPUs of the host. In addition, the pCPUs pinned by a vCPU belong to the same pNUMA node.

After the vNUMA configurations are completed, a vNUMA topology with one or more vNUMA nodes is generated by passing-through the topology of associated host pNUMA nodes. Then the vCPUs of the VM instance primarily access the local memory of the vNUMA node where the vCPUs reside.

图 1. vNUMA Configuration Basics


pNUMA Topology

The following describes the pNUMA topology of a host on ZStack Cube Ultimate:
  • All pNUMA nodes of the host and the VM instances associated with each pNUMA node are displayed.
  • Total Memory: The total amount of local memory that a pCPU can access from the pNUMA node where it resides.
  • Free Memory: The amount of free local memory that a pCPU can access from the pNUMA node where it resides.
  • Both the total memory and free memory are measured based on physical hardware resources.
图 2. pNUMA Topology


You can click the Redirect icon to view the vNUMA topologies of the associated VM instances. For more information about vNUMA topologies, see vNUMA Topology.

Host QEMU Version

Version Requirement

  • All hosts in a same cluster should be installed with QEMU of the same version. If you need to add a host with a QEMU version different from those of other hosts in the cluster, update the QEMU of the hosts in this cluster to the Cloud recommended version or create a new cluster to add this new host.
  • To avoid function or security patch losses, we recommend that all hosts use QEMU of the recommended version.

View Host QEMU Version

ZStack Cube Ultimate monitors the QEMU versions of hosts in a cluster. When a part of the host use QEMU that is not of the Cloud recommended version, a prompt appears on the cluster details page to inform you to update the host QEMU version. You can view the QEMU version state of each host by clicking Check in the prompt. The QEMU state is displayed as To be Updated if it is not of the recommended version.
图 1. View Host QEMU Version




Update Host QEMU Version

ZStack Cube Ultimate allows you to update the QEMU version for all hosts in a cluster. Note that the QEMU update is a risky action. Before you execute the QEMU update, contact our technical support to evaluate the risk level.

Updating host QEMU version mainly contains following steps:
  1. Check the connection of Cloud services and hardware.
  2. Set VM HA mode as None in Global Setting.
  3. Query the host QEMU version.
  4. Back up the database of the two management nodes.
  5. Upload the latest ISO image to the two management nodes and update the local repo source.
  6. Use CLI commands to updates QEMU version for hosts in the cluster.
  7. Use CLI commands to query the update process and confirm that the update is successful.
Note: For detailed update commands, contact our technical supports.

Considerations

  • After the update, restart or migrate the VM instances, VPC vRouters, and LB instances on the hosts to make the update take effect.
  • When the QEMU version of a VM instance, VPC vRouter, or LB instance is not consistent with that of the host it resides on, its QEMU state is displayed as To be Updated. You can restart or migrate the VM/VPC vRouter/LB Instance to update its QEMU version.
    图 2. VM Instance QEMU State

    图 3. VPC vRouter QEMU State


    图 4. LB Instance QEMU State