Adding hosts

The IBM® Cloud Infrastructure Center installation is first on the management node and adds the compute node (namely host) later. When the management node is installed correctly, you can log on to the IBM® Cloud Infrastructure Center with your username and password. As the UI shows, select Add Host on the Hosts page to add a host. However, before you add host, it is necessary to complete some requirements.

Note: The IBM Cloud Infrastructure Center supports z/VM® and KVM host management type.

Follow these steps to finish adding host:

On the Hosts page, select Add Host. A pop-up window is shown, on which you should type the information of host that you expect to add.

1) The item information in the following table is common for z/VM and KVM.

Name	Required	Type	Description
Hostname or IP address	Yes	Common	Type the hostname or ip address of the compute node you want to add, in which the hostname should be resolved by IBM® Cloud Infrastructure Center (You could check hostname by referring to Verify settings for the network).
Display name	No	Common	It's a unique name that is displayed in the IBM® Cloud Infrastructure Center to identify respective host. If you type nothing into this item, the host's static hostname is displayed in the IBM® Cloud Infrastructure Center instead. Note: This name can include only letters, numbers, dashes(-), and underscores(_).
User ID	Yes	Common	Type the Linux user of the host. Both root user and non-root user are supported. Add host with root/non-root user reference link to User setting for add Host with root or non-root
Authentication type	Yes	Common	The IBM® Cloud Infrastructure Center supports two types of authentication: Password: Select Password and specify the password for the User ID. SSH key: Select SSH key and then upload the ssh key that is generated in advance from the local directory of host. Note: As for non-root user ID, it should be created and granted with root authority in advance.

2) z/VM

Note for z/VM® host management

Fill the following fields to finish adding a host for z/VM:

Name Required Type Description

Vswitch Name

Yes

z/VM only

Name	Required	Type	Description
Vswitch Name	Yes	z/VM only	The name of vswitch on the compute node you plan to use. Note: the colon character (:) is not allowed.
FCP List	No	z/VM only	This value is required only if you want to use persistent storage as the root or data disks for the virtual machines deployed on z/VM. Refer to Boot From Volume and Attach volumes to virtual machines. Separate the FCP devices from different FCP channels with a semicolon, and comma-separate both continuous (A-B) or single ( C) FCP device lists on each channel. For example, for a two FCP channels, `1FB1-1FB4,1FB6,1FB8-1FBF;1FC1-1FC4,1FC6,1FC8-1FCF`. Each virtual machine that will use persistent storage as root or data disks consumes one FCP device from each FCP channel. Make sure you plan the number of FCP devices according to your requirements. You can also add list later using Edit Host Connection button in Hosts pane. Manage this FCP List carefully and make sure that the new FCP list also includes the old FCPs. The list must include all FCP devices, including FCP devices already attached to VMs, so that you get a full overview of all FCP devices that IBM Cloud Infrastructure Center is using. The FCP devices in the specified range should be Free meaning that they should neither be dedicated to other existing vms nor added into any hosts on the storage backend. IBM Cloud Infrastructure Center manages these FCP devices and allocates them to the guest virtual machines.
Disk Pool	No	z/VM only	This value is required to create root or data disks from the z/VM® dasd group for the virtual machines. It has 2 parts separated by a colon(:). The first part is the type of disks in the dasd group. The disks in one dasd group must in the same type (ECKD or FBA). Possible values of the dasd group type is either ECKD or FBA. The second part is the volume group name defined in your directory manager on your z/VM® system, that is used for allocating disks for new virtual machines. Sample Disk pool values: ECKD:diskpo1 or FBA:testpool. Note: A dollar sign ($) is not allowed in the volume group name. It is not supported to edit `Disk pool` with the Edit Host Connection button in the Hosts pane. Refer to Edit host. At least one of the disk_pool and the fcp_list is needed, it depends on which type of storage you use for the virtual machine. Refer to Planning for Local Storage on z/VM for more information. If the dasd group is used up, contact the z/VM® administrator to increase the dasd group size at z/VM® level and IBM® Cloud Infrastructure Center can automatically use the added disks.

The name of vswitch on the compute node you plan to use.

Note: the colon character (:) is not allowed.

FCP List

z/VM only

This value is required only if you want to use persistent storage as the root or data disks for the virtual machines deployed on z/VM. Refer to Boot From Volume and Attach volumes to virtual machines. Separate the FCP devices from different FCP channels with a semicolon, and comma-separate both continuous (A-B) or single ( C) FCP device lists on each channel. For example, for a two FCP channels, 1FB1-1FB4,1FB6,1FB8-1FBF;1FC1-1FC4,1FC6,1FC8-1FCF.

Each virtual machine that will use persistent storage as root or data disks consumes one FCP device from each FCP channel. Make sure you plan the number of FCP devices according to your requirements.

You can also add list later using Edit Host Connection button in Hosts pane. Manage this FCP List carefully and make sure that the new FCP list also includes the old FCPs. The list must include all FCP devices, including FCP devices already attached to VMs, so that you get a full overview of all FCP devices that IBM Cloud Infrastructure Center is using.

The FCP devices in the specified range should be Free meaning that they should neither be dedicated to other existing vms nor added into any hosts on the storage backend. IBM Cloud Infrastructure Center manages these FCP devices and allocates them to the guest virtual machines.

Disk Pool

z/VM only

This value is required to create root or data disks from the z/VM® dasd group for the virtual machines. It has 2 parts separated by a colon(:). The first part is the type of disks in the dasd group. The disks in one dasd group must in the same type (ECKD or FBA). Possible values of the dasd group type is either ECKD or FBA. The second part is the volume group name defined in your directory manager on your z/VM® system, that is used for allocating disks for new virtual machines. Sample Disk pool values: ECKD:diskpo1 or FBA:testpool.

Note:

A dollar sign ($) is not allowed in the volume group name.
It is not supported to edit Disk pool with the Edit Host Connection button in the Hosts pane. Refer to Edit host.
At least one of the disk_pool and the fcp_list is needed, it depends on which type of storage you use for the virtual machine. Refer to Planning for Local Storage on z/VM for more information.
If the dasd group is used up, contact the z/VM® administrator to increase the dasd group size at z/VM® level and IBM® Cloud Infrastructure Center can automatically use the added disks.

3) KVM

Note for KVM host management

Fill the following fields to finish adding a host for KVM:

Name Required Type Description

Network interface name

Yes

KVM only

Name	Required	Type	Description
Network interface name	Yes	KVM only	Enter a network interface on the host that can be assigned to Open vSwitch bridge, e.g. eth0. This interface is used to provide connectivity for KVM guests. Notes: Once the network interface is used, the original assigned IP to this interface is not accessible anymore. The input network interface can be an interface connecting to a single OSA card, a bonded interface or a team interface. For more information about these 2 types of interfaces, refer to Terminology When you add multiple KVM hosts (one host is one LPAR) and if the Network interfaces name you input for different KVM hosts are associated to the same OSA adapter, then only the 1st host can be added successfully. The others fail due to missing to set Bridgeport mode for this OSA adapter (the 1st host already set). This happens if an OSA adapter is shared across multiple LPARs on the same system, only a single LPAR can be configured for the Bridgeport mode at any point in time. So, separate KVM hosts for the Bridgeport mode concurrently require separate OSA adapters. For more information, refer to: Planning for KVM networks and Hardware and software requirements for KVM system.
Shared storage path	No	KVM only	It's a local path on the host that mounts to a remote `share storage (such as Spectrum Scale/NFS storage)`. If you do not enter anything to this item, the `local storage` on the host is used for the virtual machines as root and data disks. Refer to Planning share storge for more details. Notes: It's not supported to change `Shared storage path` once the host is managed. Make sure the Spectrum Scale/NFS storage server is always active, or else the host fails to be added for timeout exception and the virtual machine deployment may fail for permission denial. Make sure the `Shared storage path` is always mounted to the share storage server, or else the local storage of the host is used and the virtual machine deployment may fail for permission denial. Make sure the selinux context type of the `Shared_storage_path` is `nfs_t`, you can use `ls -alZ` to check it. If it's not `nfs_t`, you can use `chcon -t nfs_t Shared_storage_path` to update it.

Enter a network interface on the host that can be assigned to Open vSwitch bridge, e.g. eth0. This interface is used to provide connectivity for KVM guests.

Notes:

Once the network interface is used, the original assigned IP to this interface is not accessible anymore.
The input network interface can be an interface connecting to a single OSA card, a bonded interface or a team interface. For more information about these 2 types of interfaces, refer to Terminology
When you add multiple KVM hosts (one host is one LPAR) and if the Network interfaces name you input for different KVM hosts are associated to the same OSA adapter, then only the 1st host can be added successfully. The others fail due to missing to set Bridgeport mode for this OSA adapter (the 1st host already set). This happens if an OSA adapter is shared across multiple LPARs on the same system, only a single LPAR can be configured for the Bridgeport mode at any point in time. So, separate KVM hosts for the Bridgeport mode concurrently require separate OSA adapters.
For more information, refer to: Planning for KVM networks and Hardware and software requirements for KVM system.

Shared storage path

KVM only

It's a local path on the host that mounts to a remote share storage (such as Spectrum Scale/NFS storage). If you do not enter anything to this item, the local storage on the host is used for the virtual machines as root and data disks. Refer to Planning share storge for more details.

Notes:

It's not supported to change Shared storage path once the host is managed.
Make sure the Spectrum Scale/NFS storage server is always active, or else the host fails to be added for timeout exception and the virtual machine deployment may fail for permission denial.
Make sure the Shared storage path is always mounted to the share storage server, or else the local storage of the host is used and the virtual machine deployment may fail for permission denial.
Make sure the selinux context type of the Shared_storage_path is nfs_t, you can use ls -alZ to check it. If it's not nfs_t, you can use chcon -t nfs_t Shared_storage_path to update it.

Click Add Host, the add host process begins. If the host is added for the first time, the authenticity of host cannot be verified and click Connect to continue. After Add Host is submitted, another dialogue with Add another or Dismiss window is shown.

a). Only one management node can manage the host at a given time. If the host to be added has already been managed by the IBM Cloud Infrastructure Center, the host needs to be removed from the other management node first.

b). If the authentication type is SSH key, more detailed steps are at: add host with SSH key.

c). z/VM only The host can be used to do both, deploy the new virtual machines and manage the existing virtual machines.

d). KVM only The UI shows a warning information "Existing access key files for user 'username' on compute node 'hostname' may lead to live migration failure." If the existing key file 'id_rsa' or 'id_rsa.pub' in the folder '/root/.ssh/' on the host added from the UI. To avoid this problem, remove the key files 'id_rsa' and 'id_rsa.pub' under '/root/.ssh/' before adding the host. 'hostname': is the host name added to the management node. 'username': user name to manage the compute node by the IBM® Cloud Infrastructure Center.

e). KVM only The migration may not work without any notification if the key files 'id_rsa', 'id_rsa.pub' or 'authorized_keys' are changed during the compute node's management.

f). KVM only If the shared storage of the IBM® Spectrum Scale or NFS is used for migration, complete the configuration of the share storage mount point before adding the host operation and fill the mounted path on the UI. And ensure the shared storage server of the IBM® Spectrum Scale or NFS be available.
Adding a host might take a few minutes. If the host is discovered and added into IBM Cloud Infrastructure Center successfully, the host's information is shown on the Hosts page, which includes the host's display name, processors, memory, the number of virtual machines on the host, and the host connection item.
After the host is added as compute node, one new config file nova-<nodename>.conf is created under /etc/nova on the management node.

As for z/VM, the nodename is the hypervisor hostname of the host.
As for KVM, the nodename is the hypervisor FQDN(Fully Qualified Domain Name) hostname of the host.

Note: Once the host is added to the IBM Cloud Infrastructure Center, it is not allowed to change its hostname (you could use hostname command to check it). Otherwise there would be some unexpected results, such as instances deploy and management failed. If the "Existing access key files for user root on compute node hostname may lead to live migration not work." it is shown on the UI, the migration operation fails.

Note: Once the host is added to the IBM Cloud Infrastructure Center, the allow_lun_scan on the host is turned off. If you have any volumes mapped to the host, with the allow_lun_scan disabled, a system reboot causes volume lost as the kernel no longer performs a LUN scan, automatically. Refer to FCP LUNs that are part of the root file system or FCP LUNs that are not part of the root file system to add and persist the volume.