Cloudpods – Basic Resources

Docs: Hosts

Mon, 01 Jan 0001 00:00:00 +0000

A host is a physical server that provides underlying resources for VMs on the cloud management platform. Hosts are domain resources.

Host sources:

The cloud management platform can sync host information from VMware, ZStack, OpenStack, Zetta Cloud, HCSO, and other cloud platforms.
After a physical server installs the proprietary host service, it automatically reports host information to the default domain and is shared by default.
Convert a physical machine (bare metal) to a host.

Entry: In the cloud management platform, click the navigation menu in the upper left, then click “Resource Management / Compute / Basic Resources / Hosts” in the left menu to open the Hosts page.

The upper right of the list shows the total number of hosts and status counts (running, stopped, operation failed, unknown).

List fields

The host list supports many display fields, grouped as follows:

Basic information fields

Field	Description
Name	Host name; click to open details. Icons after the name may include: - Host type icon (bare metal badge): The host was converted from a physical machine - Passthrough device icon: The host has GPU or other passthrough devices - Huge page icon: Huge page memory is enabled (page_size_kb > 4)
Status	Running status, such as running, stopped, running (offline), unknown, maintaining, or maintain failed
Enabled status	Whether the host is enabled (enabled/disabled). Disabled hosts cannot create VMs
Service (host_status)	Host service status, such as online/offline; alarm/error icons may appear beside it—hover to view sys_warn/sys_error details
Tags	Host tags; add and edit supported (except BingoCloud)
IP	Multiple IP addresses on the host: - Management IP (access_ip) - IPMI IP (ipmi_ip; bare metal hosts only) - Public IP (EIP)
IPMI password	Shown only for bare metal hosts (is_baremetal); click to get/view the password
#VM	Number of non-system VMs on the host; sortable
CPU architecture	CPU architecture, such as x86_64 or aarch64
Platform / Brand	Cloud platform brand the host belongs to
Cloud account	Cloud account the host belongs to
Share scope	Share scope of a domain resource: private / partial share / global share
Domain / Project	Domain and project the host belongs to
Region	Region where the host is located
Alert status	Monitoring alert status (requires the corresponding scoped policy)
Created at	Creation time

Resource usage fields (all sortable)

Field	Description
CPU usage	Used CPU cores / actual CPU cores; progress bar; hover for values
CPU allocation rate	Allocated VCPU cores / (actual CPU cores × CPU overcommit ratio); progress bar; hover for allocation details and overcommit limit
Memory usage	Used memory / actual memory; progress bar; hover for memory details (GB/MB)
Memory allocation rate	Allocated memory / (actual memory × memory overcommit ratio); progress bar; hover for allocation details
Storage usage	Used storage / storage pool capacity; progress bar
Storage allocation rate	Allocated storage / (storage pool capacity × storage overcommit ratio); progress bar; hover for storage overcommit limit
Disk read/write rate	Real-time disk read rate and write rate (B/s)
Network throughput	Real-time network inbound and outbound traffic (B/s)

Hardware information fields

Field	Description
Brand (manufacture)	Server vendor; icons for Dell, HP, HPE, Inspur, Lenovo, Super Micro, Huawei, Red Hat, and more
Model	Server hardware model
SN	Server serial number (sn); can be copied
Host type	Virtualization type: KVM, container host (container), bare metal (baremetal), and more
Scheduling tags	Bound scheduling tags shown as tags; shows “Not configured” when empty. Used for VM scheduling policies

Note: Fields can be shown or hidden via column settings in the upper right; actual columns may differ by configuration.

Host service offline notes

Common cases when the OneCloud host service goes offline:

At startup, the host service checks whether qemu is installed, whether the nbd module exists, and whether the native openvswitch kernel module exists. If any check fails, the host service goes offline and shows a specific error.

When an administrator intentionally takes the host service offline (ocadm node disable-host-agent <node-name> on the control node), VM status on the host does not change; users can still connect and use VMs remotely, but the cloud management platform cannot manage those VMs.

If region detects an abnormal offline, VMs on the host are set to unknown and cannot be used normally.

On private cloud platforms, host service offline may be caused by the host being powered off or related services being abnormal.

Remote control

Connect to a host via SSH for management. Depending on configuration, you may need the initial account password after connecting.

On the Hosts page, click “Remote control” in the Actions column to expand the submenu:
- SSH management IP: Open a web SSH connection using the host management IP.
- SSH management IP: custom port: If port 22 is occupied, set a custom SSH port in the dialog, then click “OK”.
- Custom account/password connection: Click the edit icon next to the SSH option, enter port, username, and password, then connect with custom credentials.
If multi-factor authentication (MFA) is enabled, complete MFA before connecting.

Notes:

Only hosts in the running state support SSH; for other statuses the SSH menu is disabled.

Bare metal hosts also support SOL web console and JNLP console (IPMI remote management).

BingoCloud hosts do not support SSH.

Enable / Disable

Enable/disable controls whether new VMs can be created on the host.

Enable: Enable a disabled host. Enabled hosts can create VMs and be scheduled.
Disable: Disable an enabled host. Disabled hosts cannot create VMs, but existing VMs keep running normally.

Platform limits: BingoCloud hosts do not support enable/disable.

Permission: Only users in the resource owner domain (ownerDomain) can enable/disable.

Batch enable/disable: In the host list, select one or more hosts, click “Batch actions” above the list, then select “Enable” or “Disable”.

Set tags

Hosts support tags for classification and filtering. You can edit tags directly on the list page.

Limits: BingoCloud hosts do not support tag editing.

Batch set tags: In the host list, select one or more hosts, click “Batch actions” above the list, then select “Set tags”.

Adjust scheduling tags

Bind scheduling tags to a host. Hosts with scheduling tags are preferentially scheduled to create VMs. When multiple hosts need the same tag, use batch adjust scheduling tags; to bind different tags, adjust each host separately.

On the Hosts page, click “More” in the Actions column, then select “Adjust scheduling tags”. The dialog appears.
Select scheduling tags to bind, then submit.

Limits:

BingoCloud hosts do not support adjusting scheduling tags.

Only users in the resource owner domain can perform this operation.

Batch actions: In the host list, select one or more hosts, click “Batch actions” above the list, then select “Adjust scheduling tags”.

Adjust overcommit limits

Set CPU and memory overcommit ratio limits and system reserved memory. Overcommit means the host can allocate more than its actual resources (allocatable = actual × overcommit ratio).

CPU overcommit ratio can be much greater than 1 (depending on workload)
Memory overcommit ratio is recommended to be 1 (to avoid OOM from memory overcommit)

Important: After huge page memory is enabled, memory overcommit ratio and system reserved memory cannot be adjusted.

Limits: ZStack hosts do not support adjusting overcommit limits; BingoCloud does not support it.

Parameter description

Parameter	Description
CPU overcommit limit	Upper limit of the host CPU overcommit ratio. Allocatable VCPUs = physical CPU cores × overcommit ratio
Memory overcommit limit	Upper limit of the host memory overcommit ratio. Allocatable memory = (actual memory − system reserved) × overcommit ratio
System memory reserved (GB)	Memory reserved for the host system; minimum 1 GB. Reserved memory is not allocated to VMs

Batch actions: In the host list, select one or more hosts, click “Batch actions” above the list, then select “Adjust overcommit ratio”.

Change domain

Change the domain the host belongs to. After a server installs the proprietary host service, it registers to the OneCloud platform under the default domain. Use Change domain to move it.

Conditions for changing domain (all required):

The current user is in the admin backend (admin backend domain).

Three-level permissions are enabled on the platform.

The host share scope is private (not shared).

Not a BingoCloud host.

Only users in the resource owner domain can perform this operation.

Batch actions: In the host list, select one or more hosts, click “Batch actions” above the list, then select “Change domain”.

Set the share scope of a host as a domain resource.

Domain resource share scopes:

Not shared (private): Only users in this domain can view and use the host.
Domain share - partial (multi-domain share): Shared to specified domains (one or more). Only users in the host’s domain and shared domains can use it.
Domain share - all (global share): Shared to all domains; all users in the system can view and use it.

Conditions for setting sharing (all required):

The current user is in the admin backend.

Three-level permissions are enabled on the platform.

Not a BingoCloud host.

Only users in the resource owner domain can perform this operation.

Batch actions: In the host list, select one or more hosts, click “Batch actions” above the list, then select “Set sharing”.

Auto migrate on crash

Set whether OneCloud hosts enable auto migrate on crash. When enabled, VMs created with shared storage on the host can automatically migrate to other available hosts if the host is powered off or fails. VMs using local storage on the host do not trigger migration.

On the Hosts page, click “More” in the Actions column, then select “Auto migrate on crash”.
In the dialog, select or clear “Enable auto migrate”, then click “OK”.

Limits:

Only OneCloud (OneCloud/provider=onecloud) hosts support this; other platforms show it disabled with “Only OneCloud hosts support this”.

The host must have health checks enabled (allow_health_check=true).

Only users in the resource owner domain can operate.

BingoCloud does not support this.

Batch actions: You can set auto migrate on crash for multiple hosts via batch actions.

Recycle as physical machine

Recycle a host as a physical machine (bare metal). Physical machines create bare metal servers; hosts create KVM VMs.

Conditions for recycling as a physical machine (all required):

Host type must be KVM / hypervisor (not baremetal or container).

Must be a bare-metal-converted host with IPMI information (is_baremetal=true).

Number of virtualized machines (#VM) must be 0 (no VMs).

Host must be disabled (enabled hosts cannot be recycled).

ARM architecture (cpu_architecture=aarch64) hosts do not support recycling as physical machines yet.

Only users in the resource owner domain can operate.

BingoCloud does not support this.

When conditions are not met, the button is disabled with a reason (such as “Cannot recycle when virtualized machines > 0” or “Enabled hosts cannot be recycled”).

Batch actions: In the host list, select one or more hosts, click “Batch actions” above the list, then select “Recycle as physical machine”.

Enter maintenance mode

When a host needs temporary downtime (hardware repair, firmware upgrade, and so on), put it into maintenance mode. VMs on a host in maintenance mode automatically migrate to other idle hosts.

Conditions for entering maintenance mode:

Only OneCloud (hypervisor) hosts support this (not baremetal/container).

host_status must not be offline (offline cannot auto-migrate).

Host status must be running or maintain_fail.

Other idle available hosts must exist as migration targets.

VMs on the host must not have GPU devices or optical discs attached (if attached, VMs must be stopped).

VMs on the host must not have backup instances.

VM statuses on the host must be stopped, running, or unknown (other statuses may not migrate).

Only users in the resource owner domain can operate.

BingoCloud does not support this.

On the Hosts page, click “More” in the Actions column, then select “Enter maintenance mode”.
After confirmation, VMs on the host migrate one by one to other hosts. Host status becomes “Maintaining” (maintaining), and enabled status becomes disabled.
If some VMs fail to migrate, status becomes “Maintain failed” (maintain_fail). Handle failed VMs manually, then retry or exit maintenance mode.

Exit maintenance mode

When the host is healthy again, exit maintenance mode and rejoin the scheduling pool.

On the Hosts page, for a host in “Maintenance mode” or “Maintain failed”, click “More” in the Actions column, then select “Exit maintenance mode”.
After confirmation, host status becomes “Running”. Enabled status remains disabled; you must manually Enable before the host can create and schedule VMs.

Note: When entering maintenance mode fails (maintain_fail), Exit maintenance mode can return the host to normal running.

Limits: Only hypervisor (KVM) hosts support this; BingoCloud does not; only owner-domain users can operate.

Set NUMA affinity

Set NUMA (Non-Uniform Memory Access) affinity on a host to optimize memory access across NUMA nodes.

On the Hosts page, click “More” in the Actions column, then select “Set NUMA affinity”.
Configure the NUMA policy in the dialog and submit.

Set CPU reserved resources

Reserve some CPU cores on the host for the host system or other processes. Reserved cores are not allocated to VMs.

Limits:

Only OneCloud (OneCloud) hosts support this.

All VMs on the host must not be running.

Shared hosts cannot set CPU reservation.

Only users in the resource owner domain can operate.

Parameter: Reserved CPU cores can be 0–15 (0, 1, 2, …, 15).

Batch actions: Select multiple hosts in the list and set CPU reserved resources via batch actions.

Set passthrough device reserved resources

Reserve dedicated CPU, memory, and storage for each GPU card on the host. Reserved resources cannot be allocated to ordinary VMs and are dedicated to GPU passthrough.

Limits:

Only OneCloud (OneCloud) hosts with GPU passthrough devices support this.

Hosts without GPUs show the button disabled with “Hosts with GPUs are supported”.

When a host has multiple GPUs, the same amount is reserved for each GPU.

GPU reserved resources are deducted after host CPU/memory/storage overcommit: compute total allocatable by overcommit ratio, then deduct passthrough reservations.

Only users in the resource owner domain can operate.

BingoCloud does not support this.

Parameter description

Parameter	Description
CPU (cores)	CPU cores reserved per passthrough device (per GPU)
Memory (MB)	Memory reserved per passthrough device (MB)
Disk (MB)	Disk reserved per passthrough device (MB)

Batch actions: You can set passthrough device reserved resources for multiple hosts in batch.

Delete

Delete a host. After deletion, it is removed from the platform and no longer scheduled or managed.

Delete conditions:

Number of VMs on the host is 0 (no running or existing VMs).

Host is disabled.

Common delete checks pass ($getDeleteResult), such as no dependent resources.

Only users in the resource owner domain can operate.

BingoCloud does not support deleting hosts.

When delete conditions are not met, the button is disabled with the specific reason.

Batch delete: In the host list, select one or more hosts, click “Batch actions” above the list, then select “Delete”.

View host details

Click a host name in the list to open details. Top action buttons match list Actions (remote control, enable/disable, adjust scheduling tags, adjust overcommit ratio, recycle as physical machine, enter/exit maintenance mode, delete, and more).

The details page includes the following tabs, in order:

Host information

Shows all detailed host attributes:

Basic info: Cloud ID, ID, name, status, domain, hostname, tags, share scope, platform, enabled status, IP, MAC address, #VM, scheduling tags, host version, hardware virtualization, ISO boot, passthrough device count, region, availability zone, cloud account, cloud subscription, created at, updated at, and description.
Brand info: Brand/vendor name, model, and serial number.
CPU info: Physical CPU cores, sockets, overcommit ratio, virtual CPUs, current overcommit ratio, description, CPU architecture, passthrough device reserved CPU, and system reserved cores.
Memory info: Physical capacity (total/allocated), overcommit ratio, current overcommit ratio, system reserved (typically about 10% of actual host memory, capped at 10 GB), and passthrough device reserved memory.
Storage info: Root partition capacity (used/total), storage pool capacity (used/total), type, current overcommit ratio, invalid storage (total capacity of disks not in available status), passthrough device reserved storage, and adapter information.
Network interfaces: All NIC information (IP, MAC, subnet mask, type, speed).

Dashboard (resource stats and monitoring)

Shows host capacity stats, real-time monitoring, and TOP5 resource consumers:

Capacity stats: Ring charts for CPU, memory, local storage, and VM count (total, used, and usage percentage).
TOP5: Top 5 VMs by CPU usage, network inbound, network outbound, disk read rate, and disk write rate, with values.
Real-time monitoring: Gauges for system load, memory utilization, disk IO usage, disk space usage, NIC inbound bandwidth usage, and NIC outbound bandwidth usage.

VM list

Lists all VMs (non-system VMs) on the host and supports common VM management (create, delete, start, stop, restart, adjust configuration, and more). See VMs.

Note: For container hosts (host_type=container), this tab title becomes “Containers” and shows container instances.

Network (bare metal network information)

Shows host network interface details. Requires the baremetalnetworks_list permission.

Field	Description
Index	NIC index
MAC address	NIC MAC address
NIC type	NIC type
IP address	IP address bound to the NIC
IP subnet	Owning IP subnet
Layer 2 network	Owning Layer 2 network
Driver	NIC driver name

For NICs without an allocated IP, you can bind a Layer 2 network manually: click “Set Layer 2 network” in the Actions column, select a Layer 2 network, then submit.

Storage

Shows storage information and local storage pools on the host. Requires the hoststorages_list permission.

Field	Description
Name	Storage name
Capacity	Total storage capacity
Allocated	Capacity allocated to VMs
Wasted	Wasted capacity (allocated but not actually used)
Storage type	Storage type, such as local storage, NFS, or Ceph
Enabled	Whether storage is enabled
Mount point	Mount path of the storage on the host

Supports storage management (enable/disable, adjust configuration, and more).

Passthrough devices / GPU

Shows all GPU cards and other passthrough devices on the host:

Field	Description
Device type	GPU card type
Device model	Specific model
Associated host	Owning host
Attach status	Whether attached to a VM

Supports GPU management (attach, detach, and more).

Recycle bin

Lists VMs on this host that are in the recycle bin (pending_delete=true). Supports purge or restore; see Instance recycle bin.

Monitoring

Shows historical host monitoring metrics as line charts across time ranges (last 1 hour / 3 hours / 6 hours). You can set a monitoring alert for each metric.

Monitoring metrics:

Metric category	Included metrics
CPU	CPU usage (user/system/iowait/idle breakdown), CPU idle rate, and more
System load	1/5/15-minute average load, average load per CPU core
Memory	Memory usage, used, free, and total
Disk	Disk usage, used, free, IO usage, IOPS (read/write)
Network	Network inbound traffic, network outbound traffic

Set monitoring alert: Click “Set monitoring alert” to the right of a metric. In the create dialog, configure query period, comparison operator, threshold, alert level, alert method, and recipients. After submit, an alert rule for that metric is created.

Alert history

Lists historical monitoring alert events related to the host, including alert level, metric, trigger time, and recovery time.

Tasks (Task Drawer)

Shows async compute tasks related to the host, such as VM migration, host maintenance mode, and overcommit ratio adjustment. View progress, status, error information, and more.

Operation logs

Shows operation logs related to the host.

On the host details page, click the “Operation logs” tab.
- Load more: 20 logs by default; click “Load more” for more.
- View log details: Click “View” in the Actions column; you can copy the content.
- Time filter: Set start and end dates in the upper right to query a time range.
- Export logs: Click the export icon in the upper right, set export columns, then click “OK” to export logs on this page (CSV/Excel).

DMESG logs

Shows host kernel boot logs (dmesg output) for troubleshooting hardware and kernel-level boot issues.

Hardware logs (BMC Log, bare metal hosts only)

Shown only for bare metal hosts converted from physical machines (is_baremetal=true). Displays hardware event logs recorded by the BMC (Baseboard Management Controller).

Set an end time (cutoff) to query hardware logs before that time.
When there are many logs, click “Load more” below the list for more history.

Docs: Physical Machines

Mon, 01 Jan 0001 00:00:00 +0000

A physical machine is a physical server registered with IPMI information on the cloud management platform and fully managed and controlled by the platform. Physical machines are used to create bare metal devices; bare metal exclusively uses physical machine resources.

Supported server vendors and models for management include Dell, HP, Huawei, Inspur, Lenovo, Super Micro, and other mainstream x86 servers.

Entry: Left menu → Resource Management / Compute / Basic Resources / Physical Machines

The upper right of the list shows the total number of physical machines and counts for running, stopped, operation failed, and unknown statuses.

View the physical machine list

View information in the physical machine list.

In the cloud management platform, click the navigation menu in the upper left, then click “Resource Management / Compute / Basic Resources / Physical Machines” in the left menu to open the Physical Machines page.
View the following parameters:
- Enabled: Whether the physical machine can be used to create bare metal. Only enabled physical machines can create bare metal.
- Status: Running status of the physical machine.
- IP: Management IP and out-of-band IP. The out-of-band IP is the IPMI address.
- SN: Serial number and related information.
- Allocation: Whether the physical machine has been allocated to create bare metal; shows the bare metal name created on this physical machine.
- Initial account: Get the username and password used to connect to the physical machine.
- IPMI: IPMI username, password, and IP.
- Maintenance mode: Whether the physical machine has entered maintenance mode (the yunionos mini system).

Add a physical machine

The OneCloud platform currently supports four ways to add physical machines:

Hosted: Host an existing physical machine that already has an OS. After hosting succeeds, a bare metal record is also created.
Pre-register: Pre-rack servers that do not yet have BMC configured. Provide the server MAC address and BMC information to configure. After the server is set to PXE boot and powered on, the OneCloud Baremetal service responds to PXE requests matching the MAC, boots the server into the YunionOS mini system, collects server information, resets BMC information, and completes registration.
ISO boot registration: Immediately boot servers that already have BMC configured. Does not require a DHCP Relay network, but requires Redfish support. Verifies BMC configuration via the Redfish API, boots YunionOS, and collects more server configuration.
PXE boot registration: Immediately boot servers that already have BMC configured. Requires a DHCP Relay network and PXE boot. Verifies BMC via IPMI, boots YunionOS, and collects more server configuration.

Before adding physical machines, ensure the Baremetal service is enabled. If it is not, the Add button on the Physical Machines page is disabled. Follow Enable Baremetal to enable it, then refresh the browser.

YunionOS is a temporary OS booted via out-of-band control. In this OS you can configure RAID, deploy OS images, and more.

ISO boot registration

Immediately register and boot a server that already has BMC configured. Does not require DHCP Relay, but requires Redfish.

On the Physical Machines page, click “Add” to open the add page.
Select the domain and set the add method to “ISO boot registration”. Two modes are supported.
- Single entry: Configure the following parameters.
  - Specify domain: Select the domain the physical machine belongs to.
  - Physical machine name: Set the physical server name.
  - IPMI address: Enter the configured IPMI address.
  - IPMI username: Enter the configured IPMI username.
  - IPMI password: Enter the configured IPMI password. Incorrect BMC information prevents successful boot registration.
  - Management port IP: Select an IP subnet. To use a static IP, click “Configure IP manually”, set the address, then click “OK”. Because ISO boot registration fits environments without DHCP Relay, you must provide the IP subnet or IP for the management port.
  - Register only, do not boot: Do not reboot into YunionOS. When selected, only partial server information is collected; the server cannot create bare metal until you run Detect hardware configuration.
- Template import: Import an XLSX template that contains physical machine registration information.
  - Follow the preset format; download the template first.
  - Fill in physical machine information in the downloaded host_template.xlsx. Put all physical machine information on the same sheet.
  - Click the dashed box or drag the template file into it, then click “OK” to upload.
  - Register only, do not boot: Do not reboot into YunionOS. When selected, only partial server information is collected; the server cannot create bare metal until you run Detect hardware configuration.
Click “OK” to complete boot registration.

PXE boot registration

Immediately register and boot a server that already has BMC configured. Requires a DHCP Relay network.

On the Physical Machines page, click “Add” to open the add page.
Select the domain and set the add method to “PXE boot registration”. Two modes are supported.
- Single entry: Configure the following parameters.
  - Specify domain: Select the domain the physical machine belongs to.
  - Physical machine name: Set the physical server name.
  - IPMI address: Enter the configured IPMI address.
  - IPMI username: Enter the configured IPMI username.
  - IPMI password: Enter the configured IPMI password. Incorrect BMC information prevents successful boot registration.
  - Management port MAC: Set the management port MAC. Servers that support Redfish do not need this field.
  - Management port IP: Optional. Select an IP subnet; to use a static IP, click “Configure IP manually”, set the address, then click “OK”. Leave empty for DHCP allocation.
  - Register only, do not boot: Do not reboot into YunionOS. When selected, only partial server information is collected; the server cannot create bare metal until you run Detect hardware configuration.
- Template import: Import an XLSX template that contains physical machine registration information.
  - Follow the preset format; download the template first.
  - Fill in physical machine information in the downloaded host_template.xlsx. Put all physical machine information on the same sheet.
  - Click the dashed box or drag the template file into it, then click “OK” to upload.
  - Register only, do not boot: Do not reboot into YunionOS. When selected, only partial server information is collected; the server cannot create bare metal until you run Detect hardware configuration.
Click “OK” to complete boot registration.

Pre-register

Pre-rack servers that do not yet have BMC configured. Pre-register only configures related information; actual boot registration happens after the server is set to PXE boot and powered on. Requires a DHCP Relay environment.

If the server already has BMC configured and you choose pre-register, BMC information may be reconfigured during boot to the values set at pre-register time, which may change the server’s BMC settings.

On the Physical Machines page, click “Add” to open the add page.
Select the domain and set the add method to “Pre-register”. Three modes are supported.

Single entry: Configure the following parameters.
- Specify domain: Select the domain the physical machine belongs to.
- MAC address: MAC address of the physical machine.
- Physical machine name: Name of the physical machine.
- IPMI address: Set the IPMI address. If omitted and an IPMI IP subnet exists on the platform, the system allocates an IP automatically. For new machines, leave empty; for existing machines, set the previous IP.
- IPMI username: Set the IPMI management username. If omitted, the system defaults to root. For new machines, leave empty; for existing machines, set the previous username.
- IPMI password: Set the IPMI user password. If omitted, the default password is YunionDev@123. For new machines, leave empty; for existing machines, set the previous password.
- When finished, click “OK”.
Batch entry: Note the following.
- One line is one physical machine record; up to 100 records per import.
- Format: “MAC address,name,IPMI address,IPMI username,IPMI password”, separated by English commas “,”. If the password contains a Chinese comma “，”, use single entry instead.
- MAC address and name are required; other parameters can be omitted.
- Example: ee:b3:f4:48:1c:f5,gpuhost01,192.168.1.1,root,admin123
- Example: ee:b3:f4:48:1c:f5,gpuhost01,,, — omits IPMI address, username, and password
- When finished, click “OK”.
Template import: Import an XLSX template that contains physical machine registration information.
- Follow the preset format; download the template first.
- Fill in physical machine information in the downloaded host_template.xlsx. Put all physical machine information on the same sheet.
- Click the dashed box or drag the template file into it, then click “OK” to upload.

After pre-register configuration, an administrator must manually set the server to PXE boot and power it on; the server then completes registration automatically.

Hosted

Manage a server that already has an OS installed. Run the following command on the server so the cloud management platform can manage it; a bare metal server with the same OS configuration is also created in the bare metal list.

Before hosting a server, ensure:

IP subnets of IPMI type and physical machine type that cover the physical server have been created on the platform;
The physical server has BMC configured (IPMI address, username, password);
Port 2222 or a later port is open on the physical server; if 2222 is occupied, open a later port.

The domain of a hosted physical machine depends on the operating user.

On the Physical Machines page, click “Add” to open the add page.
Set the add method to “Hosted”, click “Click to copy”, and run the copied command on the physical machine to be managed.
During execution, enter the server IPMI username, password, IPMI IP address, and related information.
After completion, the managed physical machine appears in the list.

Remote control

Connect to a physical machine via SOL (Serial over LAN) or SSH. SOL is remote connection over a serial port. After connecting, log in with the username and password from the Initial account column.

On the Physical Machines page, click “Remote control” in the Actions column, then select “SOL remote control” to connect via SOL.
Click “Remote control” in the Actions column, then select “SSH IP address” to open a web SSH connection.
If port 22 is occupied and SSH uses another port, select “SSH IP address: custom port”, set the port in the dialog, then click “OK”.
Click “Remote control” in the Actions column, then select “Java console” to download and open the JNLP file and connect.

Enable

Enable a physical machine that is disabled. Enabled physical machines are used to create bare metal servers.

Enable one

On the physical machine page, click “More” in the row actions for the target physical machine, then choose “Enable”.
Confirm the action.

Enable in batch

In the physical machine list, select one or more physical machine (disabled), then click “Enable” above the list.
Confirm the action.

Disable

Disable a physical machine that is enabled. Disabled physical machines cannot create bare metal servers.

Disable one

On the physical machine page, click “More” in the row actions for the target physical machine, then choose “Disable”.
Confirm the action.

Disable in batch

In the physical machine list, select one or more physical machine (enabled), then click “Disable” above the list.
Confirm the action.

Adjust scheduling tags

Bind scheduling tags to a physical machine. Physical machines with scheduling tags are scheduled to create bare metal servers according to scheduling policies.

Adjust scheduling tags one

On the physical machine page, click “More” in the row actions for the target physical machine, then choose “Adjust scheduling tags”.
Confirm the action.

Adjust scheduling tags in batch

In the physical machine list, select one or more physical machine, then click “Adjust scheduling tags” above the list.
Confirm the action.

Sync status

Sync the running status of the physical machine OS. Stopped physical machines do not support sync status.

Sync status one

On the physical machine page, click “More” in the row actions for the target physical machine, then choose “Sync status”.
Confirm the action.

Sync status in batch

In the physical machine list, select one or more physical machine, then click “Sync status” above the list.
Confirm the action.

Install OS

Create a bare metal device.

Click “More” in the Actions column, then select “Install OS” to open the create bare metal page.
Set the following parameters:
- Name: Bare metal device name.
- Quantity and specification: Because bare metal exclusively uses the physical machine, quantity and specification cannot be changed.
- Operating system:
  - Public image: Linux public images in the system image list, maintained by administrators.
  - Custom image: Linux custom images in the system image list, usually uploaded by users.
  - Boot from ISO: Linux OS images in ISO format in the system image list. When selecting an ISO, the user must install the OS for the bare metal server.
- Disk configuration: Click “Add disk” to open the create disk configuration dialog.
  - Select the disk type on the physical machine (HDD:3.6TiB means a 3.6 TiB HDD), set RAID (No RAID, RAID0, RAID1, RAID5, RAID10)
  - Set the disk count, that is, how many disks of the same configuration are used for RAID. RAID0 uses 100% of disk capacity; RAID1 with 2 disks uses 50%; RAID5 with 3 or more disks uses (n-1)/n; RAID10 with 4 or more disks uses 50%.
    - RAID0 and No RAID have no disk count requirement;
    - RAID1 requires at least 2 identical disks;
    - RAID5 requires at least 3 identical disks;
    - RAID10 requires at least 4 identical disks;
  - After RAID configuration, partition names and sizes are shown. You can modify partitions on the UI: click the yellow “System partition” area or the “Pending partition” area. Partition parameters:
    - Mount point: Partition mount point; the system partition defaults to “/”.
    - Partition format: ext4, xfs, ntfs, or swap.
    - Partition size: Maximum capacity or manual input. Maximum capacity uses all disk space; manual input sets size in GB.
- Network: Set the bare metal IP address. When two NICs connected by cable need bonding, select “Enable bonding”.

When a physical machine has exactly two cabled NICs, they are bonded automatically. When there are more than two cabled NICs, bonding results are unpredictable; proceed with caution.

 - **Specify IP subnet**: Select an IP subnet. To use a static IP, click “Configure IP manually”, set the address, then click “OK”.
- **Specify scheduling tags**: Set scheduling tags and preferences. Preferences override the default strategy on the tag. The optimal IP subnet is chosen by tags and preferences.
- **Administrator password**: Set the administrator password (root on Linux, administrator on Windows).
- **Random generate**: Generate a random password; users can view and copy it in the password column of the physical machine list.
- **Associate key**: Linux only. Create an SSH key under “Network - Keys” first. After association, log in with the SSH key pair.
- **Keep image settings**: Use the administrator password already in the image.
- **Manual input**: Set the administrator password manually.
- **Description**: Set remarks for the bare metal.

Click “Create” to create the bare metal device.

Convert to host

Convert a physical machine to a host. Physical machines create bare metal servers; hosts create VMs.

Preparation

A host is a compute node with the host service installed. When converting a physical machine to a host on the cloud management platform, use the dedicated host image provided by our company.
Upload that image to the management platform and record its image ID.
On the control node, use climc to set the default image for converting physical machines to hosts.

# Set the default image ID
$ climc service-config --config convert_hypervisor_default_template=a9b67435-8c08-4063-8ea6-d885ea26aa79 region2

When converting, you can select disk RAID settings directly, or use custom settings to choose the host image.

Steps

On the Physical Machines page, click “More” in the Actions column, then select “Convert to host”. The convert to host dialog appears.
Configure the following parameters:
- Host name: Defaults to the physical machine name; can be changed.
- Host type: Currently only KVM hosts are supported.
- Disk RAID configuration: Default (highest redundancy), RAID-1/RAID-10 (2× redundancy), RAID-5 (1.x× redundancy), RAID-0 (no redundancy), or custom.

When the physical machine has only one RAID controller and that controller has 4 or more identical disks, all options above are supported;

When disk RAID configuration is not “Custom”, the image used for conversion is the one set in region.conf on the control node, and NIC bonding is not supported.

When disks of different sizes exist on the physical machine, only custom configuration is supported;

- When selecting Default (highest redundancy), RAID-1/RAID-10 (2× redundancy), RAID-5 (1.x× redundancy), or RAID-0 (no redundancy), configure network information and whether to enable bonding.
- **Network**: Set the host IP address. When two cabled NICs need bonding, select “Enable bonding”.

When a physical machine has exactly two cabled NICs, they are bonded automatically. When there are more than two cabled NICs, bonding results are unpredictable; proceed with caution.

 - **Specify IP subnet**: Select an IP subnet. To use a static IP, click “Configure IP manually”, set the address, then click “OK”.
- **Specify scheduling tags**: Set scheduling tags and preferences. Preferences override the default strategy on the tag. The optimal IP subnet is chosen by tags and preferences.
- When selecting custom configuration, choose a disk image, add disks in custom disk configuration, select disks and RAID, set network information, and choose whether to bond.
- **Image**: For custom settings, set the image to the dedicated host image.
- **Custom disk configuration**: Click “Add disk” to open the create disk configuration dialog.
- Select the disk type on the physical machine (HDD:3.6TiB means a 3.6 TiB HDD), set RAID (No RAID, RAID0, RAID1, RAID5, RAID10)
- Set the disk count for RAID. RAID0 uses 100% of disk capacity; RAID1 with 2 disks uses 50%; RAID5 with 3 or more disks uses (n-1)/n; RAID10 with 4 or more disks uses 50%.
- RAID0 and No RAID have no disk count requirement;
- RAID1 requires at least 2 identical disks;
- RAID5 requires at least 3 identical disks;
- RAID10 requires at least 4 identical disks;
![RAID configuration](/docs/en/docs/web_ui/images/computing/raid.png)
![Partition configuration](/docs/en/docs/web_ui/images/computing/partition.png)
- Click “OK” to show partition names and sizes. You can modify partitions on the UI: click the yellow “System partition” area or the “Pending partition” area. Partition parameters:
- **Mount point**: Partition mount point; the storage pool default is “/opt/cloud/workspace”.
- **Partition format**: ext4, xfs, ntfs, or swap.
- **Partition size**: Maximum capacity or manual input. Maximum capacity uses all disk space; manual input sets size in GB.
![Create partition](/docs/en/docs/web_ui/images/computing/createpartition.png)
- **Network**: Set the host IP address. When two cabled NICs need bonding, select “Enable bonding”.

When a physical machine has exactly two cabled NICs, they are bonded automatically. When there are more than two cabled NICs, bonding results are unpredictable; proceed with caution.

 - **Specify IP subnet**: Select an IP subnet. To use a static IP, click “Configure IP manually”, set the address, then click “OK”.
- **Specify scheduling tags**: Set scheduling tags and preferences. Preferences override the default strategy on the tag. The optimal IP subnet is chosen by tags and preferences.

Click “OK” to convert the physical machine to a host.

Verify IPMI information

Verify whether IPMI information is configured correctly and whether IPMI over LAN is enabled. When management fails due to IPMI configuration issues, after you fix IPMI settings you can use Verify IPMI to check them; if correct, the physical machine is managed again.

Verify IPMI information one

On the physical machine page, click “More” in the row actions for the target physical machine, then choose “Verify IPMI information”.
Confirm the action.

Verify IPMI information in batch

In the physical machine list, select one or more physical machine, then click “Verify IPMI information” above the list.
Confirm the action.

Detect hardware configuration

Sync and obtain hardware configuration of the physical machine. When you selected “Register only, do not boot” while adding, use Detect hardware configuration so the physical machine can be used normally.

Detect hardware configuration one

On the physical machine page, click “More” in the row actions for the target physical machine, then choose “Detect hardware configuration”.
Confirm the action.

Detect hardware configuration in batch

In the physical machine list, select one or more physical machine, then click “Detect hardware configuration” above the list.
Confirm the action.

Power on

Power on a physical machine that is stopped.

Power on one

On the physical machine page, click “More” in the row actions for the target physical machine, then choose “Power on”.
Confirm the action.

Power on in batch

In the physical machine list, select one or more physical machine (stopped), then click “Power on” above the list.
Confirm the action.

Power off

Power off a physical machine that is running. Not supported when the physical machine is already allocated to bare metal.

Power off one

On the physical machine page, click “More” in the row actions for the target physical machine, then choose “Power off”.
Confirm the action.

Power off in batch

In the physical machine list, select one or more physical machine (running), then click “Power off” above the list.
Confirm the action.

Enter maintenance mode

Put a physical machine that has been allocated to bare metal into the offline system (yunionos network-boot mini system). In the offline system, administrators can change partitions, rebuild RAID, change passwords, and more. Unallocated physical machines are already in the offline system by default.

Enter maintenance mode one

On the physical machine page, click “More” in the row actions for the target physical machine, then choose “Enter maintenance mode”.
Confirm the action.

Enter maintenance mode in batch

In the physical machine list, select one or more physical machine, then click “Enter maintenance mode” above the list.
Confirm the action.

Exit maintenance mode

Exit maintenance mode.

Exit maintenance mode one

On the physical machine page, click “More” in the row actions for the target physical machine, then choose “Exit maintenance mode”.
Confirm the action.

Exit maintenance mode in batch

In the physical machine list, select one or more physical machine, then click “Exit maintenance mode” above the list.
Confirm the action.

Delete

Delete a physical machine. Only unallocated and disabled physical machines can be deleted.

Delete one

On the physical machine page, click “More” in the row actions for the target physical machine, then choose “Delete”.
Confirm the action.

Delete in batch

In the physical machine list, select one or more physical machine, then click “Delete” above the list.
Confirm the action.

Tag management

See Multi-cloud Management - Tag Management.

View physical machine details

View detailed information about a physical machine.

On the Physical Machines page, click a physical machine name to open details.
The top menu supports remote control, enable, disable, adjust tags, sync status, install OS, convert to host, detect hardware configuration, power on, power off, enter maintenance mode, exit maintenance mode, delete, and more.
View physical machine details.
- Basic info: Cloud ID, ID, name, status, domain, project, share scope, enabled status, maintenance mode, IP, allocation, MAC address, scheduling tags, host version, hardware virtualization, ISO boot, region, availability zone, cloud account, created at, updated at, and description.
- Brand info: Brand name, model, and serial number.
- CPU: CPU cores, sockets, overcommit ratio, current overcommit ratio, and description.
- Memory: Memory capacity, system reserved, overcommit ratio, and current overcommit ratio.
- Storage: Storage capacity, type, current overcommit ratio, invalid storage, and adapter information.
- Network interfaces: All NIC information: IP address, MAC address, subnet mask, type, and speed.

View bare metal server information on the physical machine

View bare metal servers on the physical machine and manage them.

On the physical machine details page, click the “Bare metal servers” tab.
View bare metal server information and manage them. For details, see Bare metal.

View physical machine network information

View network information of the physical machine.

On the physical machine details page, click the “Network” tab.
View network information: index, MAC address, NIC type, IP address, IP subnet, and Layer 2 network. A green MAC indicates the cable is plugged in.
When a NIC has no allocated IP, you can set a Layer 2 network. Click “Set Layer 2 network” in the Actions column, select a Layer 2 network in the dialog, then click “OK”.

View physical machine storage information

View storage information on the physical machine.

On the physical machine details page, click the “Storage” tab.
View storage information: name, capacity, allocated, wasted, storage type, enabled, and mount point.

View physical machine GPUs

View GPU devices on the physical machine.

On the physical machine details page, click the “GPU” tab.
View GPU information: device type, device model, and physical machine.

View physical machine monitoring

View monitoring information for the physical machine.

On the physical machine details page, click the “Monitoring” tab.
View the following monitoring information.
- Line charts for the last 1 hour, 3 hours, and 6 hours: CPU usage, system load, memory usage, memory usage details, disk usage, disk usage details, disk IO usage, disk IOPS, network inbound traffic, and network outbound traffic.
  - CPU usage: CPU usage, CPU idle, user-space CPU, kernel-space CPU, and I/O wait share.
  - System load: 1/5/15-minute average load, and average load per CPU core for 1/5/15 minutes.
  - Memory usage: Memory used, free, and total.
  - Disk usage: Disk used, free, and total.
  - Disk IOPS: Average read IOPS and write IOPS per second.
You can set monitoring alerts for the metrics above.
- Click “Set monitoring alert” to the right of a metric. The create monitoring alert dialog appears.
- The metric is fixed. Set query period, comparison operator, threshold, alert level, alert method, and recipients, then click “OK” to create the alert.

View physical machine hardware logs

View hardware logs of the physical machine.

On the physical machine details page, click the “Hardware logs” tab.
Set an end time to query hardware logs before that time.
When there are many logs, click “Load more” below the list.

View physical machine operation logs

On the physical machine page, click the physical machine name to open details.
Click the “Operation logs” tab.
- Load more: 20 logs by default; click “Load more” for more
- View details: click “View” in the row actions
- Time filter: set start/end dates to query a time range
- Export logs: click the download icon to export logs on this page