Platform Deployment

Installing from an ISO File

ZStack Zaku Container Service uses customized ZStack ISO images. Except for Kylin V10SP3-GFB, the installation procedure is the same on x86 and ARM servers. Follow these steps:

1. Configure the server.

   • Back up the data on the server’s disks. Installation will overwrite the existing data.

   • Configure an appropriate RAID level on the RAID controller to provide data redundancy. Skip this step for VM instances.

   • Set the system boot disk as the first boot device.

2. Select the boot option.

   On the ISO boot screen, select the default option to start installing the operating system.

   • Graphical, VNC, and text installation modes are available. Select the appropriate mode for your environment; graphical installation is recommended.

   • If the server has no VGA port and supports only a serial connection, use VNC or text installation.

   

3. Review the installation summary.

   This page displays the installation settings, which you can change as needed. The following defaults are configured:

   • DATE&TIME: Defaults to UTC+8 (Asia). The administrator should check the host clock in advance and set the current time and time zone.

   • LANGUAGE: English (United States) by default.

   • KEYBOARD: English (US) by default.

   

   Operating system installation screen

4. Select the installation mode.

   On the installation summary screen, click SOFTWARE SELECTION and select ZStack Compute Node.

   

5. Configure disk partitions.

   On the installation summary screen, click INSTALLATION DESTINATION to open the disk partitioning screen.

   Configure only the system disk at this stage; configure other disks after installation. Select Custom. to configure disk partitions manually:

   

   Use the following configuration for the server’s BIOS boot mode:

   • UEFI mode

   Partition	Capacity	Description
   /boot	1G	Core files required to boot Linux
   /boot/efi	500M	UEFI boot files
   /	All remaining space	Linux root directory
   swap	Delete	Swap partition

   • Legacy mode

   Partition	Capacity	Description
   /boot	1G	Core files required to boot Linux
   /	All remaining space	Linux root directory
   swap	Delete	Swap partition

   

6. The server restarts automatically after operating system installation.

   Details omitted.

Installing from a qcow2 File

1. Create VM instances.

   Create VM instances according to the node sizing, network planning, and performance requirements. The VM instance creation procedure is omitted here.

   Configure VM instance anti-affinity policies so that the instances are scheduled on different physical servers. This prevents a single physical server failure from affecting the high availability of ZStack Zaku Container Service and its business clusters. Create separate anti-affinity groups for management cluster nodes, business cluster control planes, and business cluster compute nodes.

2. Expand the VM instance root volume.

   The qcow2 image filesystem is approximately 20 to 30 GB. After creating a VM instance, expand its system disk. First increase the system disk capacity on the IaaS virtualization platform, then log in to the operating system and perform the following steps:

   # Expand the disk partition; this example uses vda as the system disk
   fdisk /dev/vda

   Enter n, p, and w in sequence.

   # Refresh the partition table and create a physical volume; /dev/vda3 is the newly created partition
   partprobe /dev/vda
   pvcreate /dev/vda3
   
   # Add the physical volume to the volume group
   vgextend zstack /dev/vda3
   
   # Extend the root logical volume using all free space in the volume group
   lvextend -l +100%FREE /dev/mapper/zstack-root
   
   # Grow the root filesystem
   xfs_growfs  /dev/mapper/zstack-root    # XFS filesystem
   resize2fs /dev/mapper/zstack-root    # ext4 filesystem
   
   # Verify that the root disk has been expanded
   df -h

All Nodes Are VM Instances

In this example, most NIC names are already consistent (enp3s0, ens3, and ens4). Only the management network NIC on one business cluster compute node needs to be renamed.

Standard configuration for the ens3 NIC (skip this step if the node has already obtained an IP address through DHCP):

# Configure an IP address for ens3
# zs-network-setting -i <NIC_NAME> <IP_ADDRESS> <NETMASK> <GATEWAY_ADDRESS>
zs-network-setting -i ens3 172.27.210.111 255.255.255.0 172.27.210.1

The ens9 NIC name differs from the other nodes and must be standardized:

# Create a bond
# zs-bond-ab -c <BOND_NAME>
zs-bond-ab -c ens3

# Add ens9 to the bond named ens3
# zs-nic-to-bond -a <BOND_NAME> <NIC_NAME>
zs-nic-to-bond -a ens3 ens9

# Configure an IP address for the bond named ens3
# zs-network-setting -i <BOND_NAME> <IP_ADDRESS> <NETMASK> <GATEWAY_ADDRESS>
zs-network-setting -i ens3 172.27.210.112 255.255.255.0 172.27.210.1

All Nodes Are Physical Servers

In production environments, configure NIC bonding to provide sufficient bandwidth and high reliability. Ensure that bond interface names are consistent after configuration.

Recommendations for physical NIC configuration:

• Physically isolate the management network, business network, and storage network. Use two links in a bond for each network.

• Prefer link aggregation mode for bonds to increase network bandwidth.

• When NICs are of the same model, bond ports across separate physical NIC cards to provide redundancy across cards.

The following example uses the business cluster management network:

• Physical NIC names may differ between servers, but the resulting bond interface names must be consistent. This example uses bond0.33 on all nodes.

• Apply the same principle to the business network and storage network. Use bond0 for the management network, bond1 for the business network, and bond2 for the storage network.

For a NIC without a VLAN subinterface, use the following procedure:

# Create a bond
# zs-bond-ab -c <BOND_NAME>; use zs-bond-lacp instead for link aggregation mode
zs-bond-ab -c bond0   # Or zs-bond-lacp -c bond0

# Add two physical NICs to bond0
# zs-nic-to-bond -a <BOND_NAME> <NIC_NAME>
zs-nic-to-bond -a bond0 p1p1
zs-nic-to-bond -a bond0 p2p1

# Configure an IP address for bond0
# zs-network-setting -i <BOND_NAME> <IP_ADDRESS> <NETMASK> <GATEWAY_ADDRESS>
zs-network-setting -i bond0 172.27.210.111 255.255.255.0 172.27.210.1

For a NIC that requires a VLAN subinterface, use the following procedure:

# Create a bond
# zs-bond-ab -c <BOND_NAME>; use zs-bond-lacp instead for link aggregation mode
zs-bond-ab -c bond0   # Or zs-bond-lacp -c bond0

# Add two physical NICs to bond0
# zs-nic-to-bond -a <BOND_NAME> <NIC_NAME>
zs-nic-to-bond -a bond0 p1p1
zs-nic-to-bond -a bond0 p2p1

# Create a VLAN subinterface
# zs-vlan -c <BOND_NAME> <VLAN_SUBINTERFACE_ID>
zs-vlan -c bond0 33

# Configure an IP address for bond0.33
# zs-network-setting -i <VLAN_SUBINTERFACE_NAME> <IP_ADDRESS> <NETMASK> <GATEWAY_ADDRESS>
zs-network-setting -i bond0.33 172.27.210.111 255.255.255.0 172.27.210.1

Mixed VM Instance and Physical Server Deployment

The platform supports mixed deployments of physical servers and VM instances. Because their NIC names can differ substantially, NIC names must also be standardized.

In this example, all business cluster control planes are VM instances whose management network NIC is ens3. Compute nodes are physical servers. Configure the management network bond interface on each physical server with the name ens3.

Configure bonds on the physical servers as follows:

1. If the NIC already has a configuration, remove it as described in Rolling Back NIC Configuration.

2. Create a bond and configure an IP address.

   For a NIC without a VLAN subinterface, use the following procedure:

   # Create a bond
   # zs-bond-ab -c <BOND_NAME>; use zs-bond-lacp instead for link aggregation mode
   zs-bond-ab -c ens3   # Or zs-bond-lacp -c ens3
   
   # Add two physical NICs to ens3
   # zs-nic-to-bond -a <BOND_NAME> <NIC_NAME>
   zs-nic-to-bond -a ens3 p1p1
   zs-nic-to-bond -a ens3 p2p1
   
   # Configure an IP address for ens3
   # zs-network-setting -i <BOND_NAME> <IP_ADDRESS> <NETMASK> <GATEWAY_ADDRESS>
   zs-network-setting -i ens3 172.27.210.111 255.255.255.0 172.27.210.1

   For a NIC that requires a VLAN subinterface, use the following procedure:

   Prepare the zaku-vlan tool on each physical server in advance.

   # Download the zs-vlan tool variant for your environment: H84r or Kylin
   # H84r operating system
   mv zaku-vlan-h84r /usr/local/bin/zaku-vlan
   chmod a+x /usr/local/bin/zaku-vlan
   
   # Kylin operating system
   mv zaku-vlan-kylin /usr/local/bin/zaku-vlan
   chmod a+x /usr/local/bin/zaku-vlan

   Create the VLAN subinterface using the following commands:

   # Create a bond
   # zs-bond-ab -c <BOND_NAME>; use zs-bond-lacp instead for link aggregation mode
   zs-bond-ab -c bond0   # Or zs-bond-lacp -c bond0
   
   # Add two physical NICs to bond0
   # zs-nic-to-bond -a <BOND_NAME> <NIC_NAME>
   zs-nic-to-bond -a bond0 p1p1
   zs-nic-to-bond -a bond0 p2p1
   
   # Create the VLAN subinterface ens3
   # zaku-vlan -c <VLAN_SUBINTERFACE_NAME> <BOND_NAME> <VLAN_SUBINTERFACE_ID>
   zaku-vlan -c ens3 bond0 33
   
   # Configure an IP address for ens3 (a subinterface of bond0)
   # zs-network-setting -i <VLAN_SUBINTERFACE_NAME> <IP_ADDRESS> <NETMASK> <GATEWAY_ADDRESS>
   zs-network-setting -i ens3 172.27.210.111 255.255.255.0 172.27.210.1

Rolling Back NIC Configuration

# Remove the IP address, netmask, gateway, and bridge from the NIC interface
zs-restore-network-setting -i bond0     # Without a VLAN configuration
zs-restore-network-setting -i bond0.33  # With a VLAN configuration

# Remove physical NICs p1p1 and p1p2 from the bond
zs-nic-to-bond -d bond0 p1p1
zs-nic-to-bond -d bond0 p2p1

# Delete ens3, the subinterface created on bond0 with VLAN 33
zaku-vlan -d ens3 bond0 33

# Delete the VLAN subinterface bond0.33
zs-vlan -d bond0 33

# Delete the bond
zs-bond-ab -d bond0

Installing the Deployment Bootstrap Program

ZStack Zaku deployment demonstration: Zaku Professional Edition management platform installation video

Download the installation package to the /root directory on the third management cluster node, then run:

bash zaku-1.24-3.X.0-all-in-one.bin

After installation, open http://SERVER_IP:12000 in Chrome to deploy ZStack Zaku Container Service through the graphical interface. Click Start Deployment under form-based deployment to open the deployment wizard.

Deployment Wizard

1. Platform settings

   Configure ZStack Cloud license integration, console embedding, and other platform settings.

   Parameter	Example	Description
   External NTP server	ntp1.aliyun.com	Synchronizes the clocks of all nodes
   ZStack Cloud license service	Enabled	When enabled, Zaku runs as a ZStack Cloud submodule.
   ZStack Cloud API address	172.21.1.50	/
   ZStack Cloud admin password	password	/
   Embed in ZStack Cloud	Enabled	Allows container resources to be managed directly in ZStack Cloud.
   ZStack Cloud address	http://172.21.1.50:5000	ZStack Cloud web console address

   

   To use a unified interface with ZStack Cloud, configure the nginx file on the ZStack Cloud MNs as prompted.

2. Node settings

   Configure basic information such as node IP addresses.

   Parameter	Example	Description
   Reuse the Business Network as the Management Network	Enabled
   cluster HA	Enabled	Enables high availability for the management cluster
   Management Network IPv4 Address	172.31.12.183; 172.31.12.145; 172.31.12.236	IP address of each node
   Business Network IPv4 Address	Automatically reuses the management network IP address
   Node Name	zaku-1; zaku-2; zaku-3
   node root password	password
   node ssh port	22

   

3. Basic settings

   Configure basic information such as node IP addresses.

   Parameter	Example	Description
   management network IPv4 netmask	255.255.255.0
   management network IPv4 gateway	172.31.12.1
   basic service management network VIP	172.31.1.21	Used for platform HA management
   basic service business network VIP	Automatically reuses the management network IP address	Used for platform HA management
   console management network VIP	172.31.12.160	Used to access Zaku Container Service
   system data disk	/dev/vdb	Stores platform metadata, container images, monitoring data, log data, and other data
   pod ip range	10.233.64.0/18	Pod IP addresses are allocated from this range. Ensure that it does not overlap with internal network ranges.
   service CIDR	10.233.0.0/18	Service IP addresses are allocated from this range. Ensure that it does not overlap with internal network ranges.
   DNS Server	223.5.5.5
   container images disk	Empty	Storage for containerd data files

   

4. Confirm settings

   Verify that all parameters are correct.

   

5. Generate YAML configuration

   For special requirements, you can modify the YAML file before deployment. Contact a ZStack engineer to evaluate the changes before modifying the file.

   The final deployment script uses the YAML configuration file as the authoritative configuration.

   

6. Install and deploy

   Platform deployment takes approximately 40 minutes. The duration varies with server performance and network conditions.

   Use the Actions button to the right of each step to view deployment logs or retry that step. If an error occurs, troubleshoot it using the logs for the corresponding step.

   

7. Deployment complete

   After deployment succeeds, click Enter Console to open the Zaku console. The default username and password are admin / admin@123.

Requesting a License

After you log in, the platform automatically opens License Management. Download the request key and provide it to a ZStack engineer to request a license.

After the license is issued, click Upload License in the upper-right corner to upload the license file.

Configuring ZStack Cloud Integration

This optional configuration integrates the ZStack Cloud tenant management module with ZStack Zaku Container Service.

Prerequisites:

• ZStack Cloud version 5.5.16 through 5.5.24.

• The Tenant Management module is enabled in ZStack Cloud.

Integration procedure:

1. Obtain the following four items.

   Description	Example
   ZStack Cloud password	password
   ZStack Zaku console VIP	172.10.207.201
   AccessKeyId	QQxxxxxxxxxxxxBz
   AccessKeySecret	56lxxxxxxxxxxxxxxvc

   To obtain AccessKeyId and AccessKeySecret, log in to ZStack Cloud and go to Container Management > service settings > AccessKey Management.

   

2. Configure AddContainerManagementEndpoint using zstack-cli.

   Log in through SSH to any ZStack Cloud management node and configure the endpoint using the following commands.

   Note: Replace the values highlighted in red with those for your environment.

   # Log in to the ZStack Cloud command-line interface
   # zstack-cli LogInByAccount accountName=admin password=<ZSTACK_CLOUD_PASSWORD>
   zstack-cli LogInByAccount accountName=admin password=password
   
   # Add a ContainerManagementEndpoint resource
   # zstack-cli AddContainerManagementEndpoint \
   # name=zaku-1 \
   # description=zaku-1 \
   # managementIp=<ZAKU_CONSOLE_VIP> managementPort=80 \
   # containerAccessKeyId=<AccessKeyId> \
   # containerAccessKeySecret=<AccessKeySecret> \
   # vendor=zaku
   zstack-cli AddContainerManagementEndpoint \
   name=zaku-1 \
   description=zaku-1 \
   managementIp=172.10.207.201 managementPort=80 \
   containerAccessKeyId=QQxxxxxxxxxxxxBz \
   containerAccessKeySecret=56lxxxxxxxxxxxxxxvc \
   vendor=zaku

3. Verify that the configuration has been added.

   zstack-cli QueryContainerManagementEndpoint
   # Returns the result in JSON format

   {
       "inventories": [
           {
               "createDate": "Sep 4, 2024 2:36:06 PM",
               "description": "Zaku trial environment",
               "lastOpDate": "Sep 4, 2024 2:36:06 PM",
               "managementIp": "172.31.1.39",
               "managementPort": 80,
               "name": "zaku",
               "uuid": "5dxxxxx0c",
               "vendor": "zaku"
           }
       ],
       "success": true
   }

4. Roll back the configuration.

   Delete the configuration using the following command:

   zstack-cli DeleteContainerManagementEndpoint uuid=5dxxxxx0c

Opening the Creation Wizard

Log in to ZStack Zaku Container Service, open Platform Management > cluster, and click Create Cluster to open the wizard.

Node Settings

Basic Settings

Confirming Settings

Verify that all parameters are correct, then click OK to start installing the Kubernetes business cluster.

Viewing Installation Logs

During cluster creation, click the log icon to view cluster installation logs.

Business Cluster Installation Complete

After installation, the cluster is in the Running state. Use One-Click Inspection to inspect its status.

Business cluster in the Running state