Manage Racks and Physical Machines

Go to the Infrastructure View by pressing the button at the top of the main screen, select a datacenter in the list, and press the '>' button to see the datacenter information. Here you can manage racks and physical machines and enable high availability of virtual machines.

Management Actions

The actions for managing racks and physical machines are:

The buttons for performing these actions are found at the bottom of the machine list.

If there is no rack, the button will open Creating Racks. Otherwise, if a rack is selected, the same button will open Creating Physical Machines on the selected rack.

What is a Rack

In Abiquo, a rack is a logical entity used for grouping physical machines and providing VLAN isolation to the virtual machines deployed there. The Administrator's Guide describes the hypervisor network configuration. Abiquo supports two types of racks: standard racks and managed racks.

High Availability of Virtual Machines

Abiquo High Availability (HA) is a vendor-neutral automatic VM restart capability that allows enterprises deploying multi-hypervisor datacenters to have a cost-effective high availability mechanism. Abiquo High Availability is based on Abiquo's virtual machine management features and is designed to complement hypervisor HA. In Abiquo, High Availability is implemented at rack level.

To configure HA, first see High Availability Prerequisites and Configuration in the Administrator's Guide.

Introduction to Abiquo High Availability

When High Availability is enabled on a rack, Abiquo periodically checks the hypervisors on the physical machines of the rack. If a hypervisor does not respond after two checks, Abiquo will start the HA process. It will restart the virtual machines on another hypervisor of the same type in the same rack from the shared datastore. Then, Abiquo shuts down the failed physical machine. If there are not enough resources on suitable hypervisors, the virtual machines will remain on the failed hypervisor and when the administrator restarts the machine, they will be unchanged. In Abiquo, the administrator must manually re-enable the physical machine that failed.

See and Configuring High Availability of Virtual Machines and The High Availability Process in the Administrator's Guide.

High Availability Configuration in the Abiquo GUI

• Enable HA when Creating Racks.
• Configure IPMI when Creating a Physical Machine.
• Create and enable a shared datastore on all hypervisors on the HA rack when Creating Physical Machines.
• Manually re-enable a physical machine after HA by Modifying a Physical Machine.

Creating Racks

Before you add physical machines to your Infrastructure, you must create a Rack. On a standard rack, you manually create physical machines, but on a managed rack Abiquo will communicate with the rack and automatically detect physical machines, so you cannot create them. If you have set up your systems for high availability of virtual machines, mark the High Availability checkbox to enable it for the rack.

Creating Standard Racks

General Information Tab for Standard Racks

Field Description Name The name of the rack Short description A short description of the contents of the rack Enable High Availability Mark to enable High Availability of Virtual Machines Rack Type Standard Rack or Managed Rack (Cisco UCS)

Creating Managed Racks and Blades

Abiquo communicates directly with managed racks and blades to automatically create the blades in an Abiquo rack. So you cannot manually create physical machines in an Abiquo managed rack but you can work with blades as with other physical machines.

General Information Tab for Managed Racks

Field Description Name The name of the rack Short description A short description of the contents of the rack Enable High Availability Mark to enable High Availability of Virtual Machines Rack Type Standard Rack or Cisco UCS Management IP The management IP required for the managed rack User The username required for the managed rack Password The password required for the managed rack Port The port required for management of the managed rack

After you create a managed rack, it will appear in the PHYSICAL SERVERS list. Click on the rack to open it and display the list of blades.

Network Parameters Tab for All Racks

Field________________ Description VLAN ID Min Min VLAN tag used by the switch VLAN ID Max Max VLAN tag used by the switch Excluded VLANs ID Exclude some VLAN tags with comma separated values ',' and range value '-' Estimated VLANs per VDC Estimated number of VLANs that will usually be deployed in VDCs. These VLANs are not reserved for VDCs - if the maximum VLANs or allocation limits can exceed this setting, the VLANs may run out before this setting is reached.* VLAN pool size for VDCs that exceed estimation VLANs kept for use in deployed VDCs after the total number of VLANs created exceeds the total expected number. New VDCs will be deployed in a rack while there are enough VLANs for a new VDC to deploy without using VLANs from this pool.*

*See VLAN Allocation Tuning and VLAN Scheduler Algorithm

VLANs

By definition there is VLAN isolation inside a rack. This is an excellent security feature to prevent external access to a client's internal private networks and offer scalable VLAN networks. Thus a whole VDC (Virtual Datacenter) will be deployed in the same rack so the virtual machines can network locally. The Abiquo Scheduler will determine which rack will be the most suitable (in terms of VLAN availability) when deploying the first virtual machine of a virtual datacenter.

VLAN Allocation Tuning

The 'Estimated VLANs per VDC' and the 'VLAN pool size for VDCs that exceed estimation' as well as the virtual datacenter allocation limits and the 'abiquo.server.networking.vlanPerVdc' configuration property give cloud administrators some control over VLAN allocation. The 'Estimated VLANs per VDC' is only used to decide how many VDCs can be deployed in a rack. It does not reserve VLANs for VDCs. To prevent a VDC from using more than a certain number of VLANs, set this number as an allocation limit. The default balanced scenario is shown here.

Default Balanced Scenario

Estimated VLANs per VDC = 1 (or other low value)
VLAN pool size for VDCs that exceed estimation = 10%

• balances VLAN allowance with number of VDCs per rack
• allows low average number of VLANS for each VDC
• small pool of VLANs for VDCs that require more than estimated VLANs
  • to control the maximum number of VLANs per VDC, set allocation limits

These parameters can only be modified on a rack with no VDCs deployed. For detailed information about VLAN allocation, please see VLAN Scheduler Algorithm.

Editing Racks

To edit a rack, select it in the rack list and press the Configuration Button in the lower right corner of the PHYSICAL SERVERS pane.

You can change the same rack values that are used for rack creation creation.

Deleting Racks

To delete a rack, select it in the rack list and press the Delete Button in the lower left corner of the PHYSICAL SERVERS pane. The rack will be deleted.

Creating Physical Machines on Standard Racks

Automatically Discovering Physical Machines

Enterprise Edition uses the Discovery Manager remote service to automatically discover physical machines and their data with two options: basic or advanced discovery.

Basic Physical Machine Discovery

In the drop down menu of the Infrastructure Screen, select the 'Physical Machine' option.

Leave the Basic radio button selected and enter the IP address of the physical machine to add.

After you enter the IP address, the Discovery Manager will look for information about the Hypervisor running on the physical machine. The client has a configuration file with the default user and password (see Default Hypervisor User, Default Hypervisor Password and Default Hypervisor Port in the Configuration View to use when logging in to the remote hypervisor. If the default user and password match those of the physical machine, the Discovery Manager will show the form described in the previous section, with the discovered values completed. Otherwise it will open the Advanced Physical Machine Discovering with the 'Invalid Credentials' error.

If the user and password do not match those of the remote hypervisor, you will see a screen like the advanced screen. If no hypervisor responds to the call, the blank form is shown.

Advanced Physical Machine Discovery

Press the 'Advanced' button, to see the advanced screen for retrieving the physical machine information. You can use this option when you know all the connection data of the remote hypervisor.

Depending on the hypervisor you select, the next step will be to complete the username and password (XenServer, Hyper-V and ESXi) or complete the port field in the same screen (Xen, KVM and VirtualBox).

If the hypervisor does not respond, the empty physical machine form will also open to allow you to complete the physical machine definition.

Create a Physical Machine on a Standard Rack

To create a new physical machine, you will need to complete the following form:

This table describes each field in the form:

Configuring IPMI for Abiquo High Availability

IPMI is used by Abiquo in the high availability (HA) process to shut down a physical machine that is not responding. Click Show on the Create Physical Machine form to open the IPMI parameters section.

Field Description IPMI Show or Hide the IPMI parameters section for High Availability IPMI IP The IP address for the IPMI connection IPMI Port An optional port for the IP connection IPMI User User with privileges to shut down the machine IPMI Password Password for the user with privileges to shut down the machine

For further information about Abiquo high availability, see Configuring High Availability of Virtual Machines in the Administrator's Guide.

Datastore Management

In Abiquo, datastores include the disks used by the hypervisors to store information related to the virtual machines and the NFS repositories.

When you create a physical machine, you will see a list of all the datastores that are installed on that machine. For some hypervisors this list includes the NFS repository (usually /opt/vm_repository), which should not be used as a hypervisor datastore.

Mark the Enabled checkbox on the create physical machine form to enable a datastore. Abiquo will deploy a VM on the enabled datastore with the most available space.

Do Not Enable the NFS Repository

Wherever possible, the datastores list does not include the NFS repository because it is an internal datastore that is used to share images between the NFS server and the hypervisor. If the NFS Repository is visible in the datastores list, you should not enable it as a hypervisor datastore because deployment over a network to the same NFS shared directory from which the virtual image templates are being copied will severely degrade performance.

Available Space

Abiquo runs a periodic infrastructure check process to update infrastructure values in the Abiquo database, including the real disk space available on a datastore. However, at deployment time, Abiquo only checks the database value.

High Availability with a Shared Datastore

Abiquo High Availability of Virtual Machines requires all the hypervisor hosts involved to share the same datastore, which must be enabled for deployment of virtual machines. This will allow Abiquo to restart machines in other hypervisors from the shared datastore. High availability does not copy virtual machine system disks. Thus, if none of the other high availability hypervisors can see the datastore used by a hypervisor, the virtual machines using this datastore will not be restarted. Therefore, Abiquo recommends that you should enable only shared datastores (and disable other datastores) on high availability hosts and ensure that this datastore is available to all other high availability hosts on the same rack.

Field Description Name Datastore Name Root Path Absolute path on the hypervisor and mount point for the hypervisor Directory Relative directory path, which is added on the root path link to form the datastore path, if this is blank, the datastore path will be just the root path. The Directory and the Root Path must be consistent Size Max Size in the datastore Used Used size in the datastore Enabled If the datastore can be used by the scheduler to deploy Abiquo VMs or not

Creating Multiple Physical Machines

You can use this feature to enter a defined range of IPs and register multiple existing physical machines with these IPs into Abiquo. For each IP, the Discovery Manager will check if the physical machine has the requirements established in the Multiple Physical Machines screen. The main difference from adding a single Physical Machine is that you will not see the confirmation screen with the physical machine parameters.

In the drop down menu of the Infrastructure Screen

Select the 'Multiple Physical Machines' option to open the following screen:

Field Description Rack The rack where machines will be added IP from First IP to search IP to Last IP to search Hypervisor Type Select the virtualization technology User-password/Port The login credentials or the remote AIM listening port Virtual Switch The hypervisor host must have a virtual switch with this exact name

Discovering Blades on Managed Racks

The Blades on UCS Racks are automatically discovered to create the list of blades on the UCS Rack when the rack is created. In the rack pane, click Discover Blades to discover the hypervisors running on these blades. Click Refresh Blades to rescan the blades.

Field Description Rack The rack where the blades will be discovered IP from First IP to search IP to Last IP to search Hypervisor Type Select the virtualization technology User-password/Port The login credentials or the remote AIM listening port Virtual Switch The hypervisor host must have a virtual switch with this exact name

When you discover blades, you will not see the confirmation screen with the physical machine parameters. All blades found by the discovery process will be automatically added or updated in the Rack. You can then work with these blades as physical machines.

Managing Physical Machines

Disabling a Physical Machine

The provisioning process always finds the best physical machine for deploying a virtual machine, according to the Virtual Machine Allocation. You can disable a physical machine in order to avoid deploying virtual machines on it. To do this, click the disable button:

While the machine is disabled, the disable button will change to an enable button. To re-enable the physical machine, click the enable button.

Enabling a Physical Machine Disabled by HA

After a physical machine has been disabled by HA, the system administrator should restart the machine and check the hypervisor.

A machine disabled by HA will have the state HA DISABLED in the Abiquo GUI.

Manually enable the physical machine in Abiquo by clicking the enable button on the right-hand side of the screen.

Reserving a Physical Machine for an Enterprise

Physical machines can be reserved for a specific enterprise and the enterprise's deployments can be restricted to these reserved machines.

This is done in Users View by selecting the editing the Enterprise and selecting the Reservations tab (see Reservations.)

Powering On/Off a Blade

Click the Power button in the top right corner of the blade details pane to power on/off the blade.

Removing a Physical Machine

To delete a physical machine, select it and press the Remove Button at the bottom of the physical servers pane; the machine will be deleted.

Physical and Virtual Machine Details

A summary of the properties of the machine appears in the lower virtual machines pane when the user selects a virtual machine or a physical machine. This summary contains the information on the selected object and the user with the appropriate privileges can edit its properties.

Each time a user deploys or undeploys a virtual appliance, the used resources (CPU, RAM, Hard Disk) in the physical infrastructure are updated. These resource allocation values can be viewed in the State tab of the Physical Machine details.

View Virtual Machines on Physical Machines

You can view the virtual machines created in different physical machines (or grouped by rack). The virtual machines can be viewed as a list of names or icons with each entry representing a virtual machine object.

In each virtual machine object you can:

• Power On/Power Off the virtual machine.
• Pause the virtual machine.
• Restart the virtual machine.
• Open VDRP/VNC to access the virtual machine (if supported by the hypervisor).

Working with Imported Virtual Machines

Abiquo can discover if there are virtual machines running in the registered physical machines. You can select the virtual machines you wish to import into Abiquo. After importing these virtual machines you will be able to see them in Abiquo and power them on and off.

Retrieve Virtual Machines from a Physical Machine

At physical machine level, you can get information about the virtual machines running in the hypervisor but not yet managed by the Abiquo Platform.
Click the Retrieve Button to retrieve the details of a physical machine. All the virtual machines found will appear in the Virtual Machine list with the following icon: . The Abiquo health check will refresh the state of these virtual machines. Remember that Abiquo high availability does not support imported virtual machines.

Import a Retrieved Virtual Machine

You can import the retrieved virtual machines into Abiquo, so that you can see the machines and power them on and off. You cannot use them in High Availability.

To import machines so that they can be seen in Abiquo and powered on or off, you must have a Virtual Datacenter already set up. Select a virtual machine and press the Import Button which is on the left of the Virtual Machine detail pane.

The Abiquo GUI will then show you the Capture Virtual Machine screen listing all the virtual appliances into which you can import the virtual machine.

You can select the virtual appliance or create a new one by pressing the Create Button. The form to create a new virtual appliance is the same as the Create Virtual Appliance form.

When you have decided where to import the virtual machine, click Accept and the machine will be imported there and it will be listed and you can power it on and off.

Removing Unmanaged Virtual Machines

You can remove all unmanaged virtual machines from the platform. Unmanaged machines are ones that were retrieved but not imported. (The unmanaged machines are identified by the icon). Select the physical machine, and press the button.

The unmanaged virtual machines will no longer be displayed on the screen. Remember you can retrieve them again at any time.

Sending Email Notifications

Click the email icon to send email notifications about a physical machine or its virtual machines. Notifications can be sent to either the owner of the machine (the user who created it) or to the administrator or enterprise manager.

The managers of an enterprise are all the users of that enterprise with the "Define Enterprise Manager" privilege. When an e-mail notification about a physical machine is sent, all users who have virtual machines deployed in that physical machine will receive the notification.