
	TOC
	Prerequisites
	Step 1: Check application node setup
	Step 2: Create a manual allocation image
	Step 3: Create a tier
	Step 4: Add nodes to the tier
	Managing manually allocated tiers and their images
	Troubleshooting
	Sidebars
	About networks and manually allocated tiers
	Using copSSH to set up OpenSSH on nodes running Windows

know-how:

Manually Allocated Tiers: Configuration and Management

Intended for use with Cassatt Active Response V5.0.

Manual allocation is a way to put a set of nodes—and the application they are running—under minimal control. When you allocate resources manually, you determine exactly which nodes are grouped together in a tier (including the applications they are already running).

This article describes how to set up, configure, and manage manually allocated tiers. Tasks include:

Verify application node monitoring, SSH connectivity, and power control – ensure that Cassatt Active Response can manage your nodes once you put them under Cassatt Active Response control
Create and configure a manual allocation "image" – define for Cassatt Active Response how to determine whether nodes are running
Create the tier – define for Cassatt Active Response how to manage your nodes
Add nodes to the tier – put each node under Cassatt Active Response control

Prerequisites

Cassatt Active Response environment:

At least one control node connected to your network switch(es) with Cassatt Active Response 5.0 or later installed.
In the Cassatt Active Response Controller, make sure that automatic discovery is turned "Off": Discovered Pool > Properties > Automatic Discovery.

Automatic discovery should be off to prevent Cassatt Active Response from taking control over servers inadvertently; automatic discovery is counter to manual allocation, and should be turned on only when your intention is to implement full automation for utility computing. For more information about automatic discovery, read Understanding Automation: Node Discovery, Inventory, and Quarantine.

Hardware requirements:

Power control: Application nodes must have supported integrated power controllers (includes IBM BladeCenter Management Modules; see Recommended Hardware and Firmware) or must use a supported external power controller from a third party (see External Power Controllers: Configuration and Management).

As a best practice, add third party external power controllers to Cassatt Active Response prior to adding manually allocated nodes to Cassatt Active Response. When you add an external power controller to Cassatt Active Response, Cassatt Active Response validates connectivity to the device.

If you do not add external power controllers in advance of adding the manually allocated nodes that use them, Cassatt Active Response creates the device during node addition, but does not check connectivity.

Do not add BCMMs to Cassatt Active Response manually when your manual allocation nodes are IBM BladeCenter blades. Manually adding a BCMM to Cassatt Active Response causes Cassatt Active Response to discover and inventory the blades, which interrupts node operations. As an extra precaution, you can turn off automatic inventory and automatic move to free pool in the Controller.

Monitoring: Application nodes must be set up for monitoring so Cassatt Active Response can detect failures. Ping is the minimum. For more information on monitoring applications and application nodes with Cassatt Active Response, see Understanding Monitoring Collectors. Make sure you set up every node for monitoring using at least one supported monitoring collector.
SSH and shutdown setup: Whenever Cassatt Active Response powers down a node for any reason, it first attempts a graceful shutdown using an operating system shutdown command. Next, whether or not the shutdown was successful, Cassatt Active Response powers off the node.

As a best practice, you should set up all nodes with an appropriate SSH configuration when SSH is supported by the OS. If your nodes are running an OS that is not currently supported by Cassatt Active Response, or that does not support SSH, contact Services to either update Cassatt Active Response with the appropriate shutdown command/user or to set up Cassatt Active Response to bypass graceful shutdown.

When you add each node to Cassatt Active Response, Cassatt Active Response copies the SSH keys to the node to allow the control node to issue shutdown commands.

Pointers:
- If your nodes are running Windows, and you do not already have an SSH implementation, you can use copSSH to set up OpenSSH.
- Set the SSH configuration for "root" if using Linux or Solaris and "Administrator" if using Windows.
- Make sure that PermitRootLogin is set to "yes" in /etc/ssh/sshd_config. Restart the service after making any changes.
- Make sure the SSH service is started.
Make sure that all nodes are up and running.
Create a properties file for each node you will add to the tier. The properties file must contain all of the required values listed in the next table, plus any of the optional values you need for your application nodes. Omit any properties that are not used. For external power controllers: if you add the power controllers to Cassatt Active Response in advance as recommended below, only the type and IP address are used when adding nodes. See a sample properties file.

Property	Required?	Valid Values	Description
node.name		Use only letters, numbers, spaces, underscores, dashes, and periods; maximum length 24	Display ID for the node
node.mac		Valid MAC address in hexadecimal format: 01:23:45:AB:CD:EF
node.ip		Valid IP address in format: 10.10.10.10
node.os		linux windows solaris	For any other OS, contact support@cassatt.com
node.description		Freeform	Display description for the node
node.rack		Alphanumeric; maximum length 32	Node rack location
node.slot		Alphanumeric; maximum length 32	Node slot number
node.powerdevid	*	Integer	The slot in a BCMM, or the outlet in an external power controller
node.sshuser		Must match the value in Cassatt Active Response exactly ("root" or "Administrator")	Existing username for the node; used for shutdown and startup operations When your OS does not support SSH, and Cassatt Services has set up Cassatt Active Response to bypass graceful shutdown for those nodes, set this value to null in the properties file.
power.name		Use only letters, numbers, spaces, underscores, dashes, and periods; maximum length 32	Display name for the power device
power.ip		Valid IP address in format: 10.10.10.10
power.type		APC	External power controller
		Baytech	External power controller
		Dell DRAC/ERA	Dell PowerEdge Servers 1750, 1850, 1950, etc.
		HP iLO	HP server with "integrated Lights Out" power controller
		IBM BCMM	IBM BladeCenter
		IBM RSA 2	IBM servers with the RSA 2 power controller
		Cyber Switching Dualcom	External power controller
power.description		Freeform	Display description for the power device
power.location1		Alphanumeric; maximum length 32	External power controller or BCMM rack location
power.location2		Alphanumeric; maximum length 32	External power controller or BCMM rack location
power.username	**		Configured user login for the power controller; note that, for Baytech and CyberSwitching, this property is not used
power.password	**		Configured user password for the power controller; note that, for Baytech and CyberSwitching, this field designates the write community and not the power controller password

*Required only for nodes connected to multi-node power devices.

** If not specified in the properties file, must be specified for the default for the power controller type in the Cassatt Active Response Controller. Except for Baytech and CyberSwitching, you should always specify the username and password in pairs.

Network Requirements:

Before determining your network strategy, read About networks and manually allocated tiers.

In the Controller, add any networks the nodes are using, including networks for power control (if any): Networks > New Network.

On the first page of the network wizard, be careful to enter the network configuration correctly to match your existing switch setup.

On the second page, reserve any IP addresses that are in use by devices such as power controllers, switches, and load balancers.

Do not add network switches to Cassatt Active Response. Networking for manually allocated tiers is predetermined, so Cassatt Active Response control over switch configuration is not needed (or desirable). You should add switches to Cassatt Active Response only when you want Cassatt Active Response to programmatically control switch port mappings using the Network Manager in a dynamic, utility computing environment. For more information about Network Manager, read Understanding Cassatt Active Response Network Manager.

Tier Settings:

SLAs: Determine the capacity settings for the tier. To understand capacity levels in general, see Understanding Tier Configuration and Personalization.

For manually allocated tiers, set the operational target and the maximum to the total number of nodes you have available (or set the operational target to that number, and the maximum to the total number you expect to have if you intend to add nodes for the application later). If your actual operational requirements are less than total nodes available, you can reduce the operational target after verifying that nodes are correctly allocated to the tier.

Be aware that Cassatt Active Response powers off any nodes beyond the operational target.

Step 1: Check application node setup

Verify monitoring setup

Test each node as follows:

Log into the control node as root.
Run the ccmonitor command, as follows:

/opt/cassatt/bin/ccmonitor -f -p protocol -u IPaddress

Where:
protocol is the monitoring collector protocol; valid values are ping, SNMP, script, HTTP, and socket (JMX does not return an error, but is not supported for this command)
IPaddress is the IP address of the application node to validate

Verify SSH

Test the SSH installation by connecting to the node via SSH from a Linux machine (you can use the control node).

Verify power control

Make sure your nodes are running, then test the power setup using the Cassatt Active Response power command:

/opt/cassatt/bin/ccpower -t power_type -a IPaddress -u username -p password -d deviceID status

Where:
power_type is the power controller; valid values are dell_rac, dell_drac, ibm_bcmm, ibm_rsa2, hp_ilo, vmm, baytech, apc, and dualcom (note that valid values for the ccpower command do not match valid values for the ccaddnode properties file)
IPaddress is the IP address of the power controller for the application node to validate
username is the username of the power controller for the application node to validate
password is the password of the power controller for the application node to validate
deviceID only used with BCMM (identifies the slot) and external power controller (identifies the outlet)

In the return text, look for this line:

Node/blade "server" is On

Where:
server represents the slot in a BCMM, or the outlet in an external power controller, or "server-1" for an integrated power controller

Step 2: Create a manual allocation image

In this step, you will duplicate the manual allocation template, configure the copy for your application, and activate the copy for use with a tier.

About the manual allocation template

The manual allocation template is an image stub that you copy and then configure to give Cassatt Active Response the information it needs to monitor and manage your business application.

Cassatt Active Response treats the template—and the duplicates you make—as though they were regular images, even though they contain no business application software. Consequently, I'll refer to the copies you'll make as manual allocation images.

A manual allocation image contains most of the same properties as an image (aside from the software, which is already deployed to your servers, and hardware requirements, which are predetermined by the servers you manually allocate to run the software).

If you already understand image properties, then proceed full-speed ahead. If you have never configured an image, you'll need to know about two things: monitoring (read Understanding Monitoring Collectors) and the Service Startup Time Limit.

The Service Startup Time Limit is the amount of time Cassatt Active Response should wait during initial startup to replace an application node if a monitoring collector indicates that the node is not starting. When Cassatt Active Response starts an application node, it waits for every monitoring collector defined for the OS and application in the image to report in. If any monitoring collector fails to report within the service startup time limit, Cassatt Active Response considers the startup to have failed and replaces the application node.

Roughly, the Service Startup Time Limit for each image should be:

The time it takes for your hardware to start, plus
The time it takes for the operating system to start and the monitoring collectors you have defined for your operating system to respond, plus
The time it takes for your applications to start and the monitoring collectors you have defined for each application to respond (optional)

Note that, for manually allocated tiers, nodes should be running when you add them to the tier: the Service Startup Time Limit comes into play only when Cassatt Active Response powers up a node, either to increase tier capacity or when an active node fails.

Follow these steps:

In the Cassatt Active Response Controller, select Images in the left navigation pane.

Cassatt Active Response displays the Images list.
In the Images list, select the checkbox for the manual allocation template, then select Duplicate in the Actions dropdown box.

Cassatt Active Response displays Duplication dialog box.
Give the duplicate a descriptive name, select a version number, specify the Cassatt base filesystem (the default), and click Next.

The image name is used for image selection from a dropdown list when creating tiers. Make sure to give the image a name that indicates it is a manual allocation image, and consequently can be used only with manually allocated tiers.

Cassatt Active Response creates the duplicate and adds it to the Images list.

Close the image duplication Results box.
Select the new image name in the Images list.

Cassatt Active Response displays the Image Properties page.
Set the Service Startup Time Limit and OS Monitoring for the image.
Optional: select the Applications tab. Add applications and set the monitoring values according to the monitoring setup on your nodes.
Click Save Changes.
In the Images list, select the checkbox for your new image, then select Activate in the Actions dropdown box.

Your manual allocation image is now ready for use in a manually allocated tier.

Step 3: Create a tier

In the Cassatt Active Response Controller, define the tier for your application. Cassatt Active Response prepopulates tier values with the parameters you set for the image. If necessary, you can change them during tier creation.

On the left navigation pane, click Tiers to display the Tier List.
On the Tier List page, click the New Tier button (on the right).
On the Properties page, set the properties for your tier.

Be sure to give the tier a name and description that indicate that the tier is a manually allocated tier. The Cassatt Active Response Controller does not differentiate between manually allocated tiers and dynamically allocated tiers.

The next table lists the suggested values for a manually allocated tier; unless you have site-specific reasons, you can accept the defaults in other fields.

At this prompt...

Enter...

Notes

Image Select your image Select the image you created in Step 2: Copy the manual allocation template and configure for your application.

Automatic move to maintenance pool

Off

Setting this property to "On" results in an error when adding nodes.

Moving manually allocated nodes to the maintenance pool is disallowed because Cassatt Active Response would attempt to reboot the nodes with the Cassatt Active Response diagnostic image. Unless your nodes are configured for PXE, this reboot would fail.

On the Requirements page, set the hardware requirements for the tier.

At this prompt...

Enter...

Notes

Operational Target

TargetNodes

Set the operational target and the maximum to the total number of nodes you have available (or set the operational target to that number, and the maximum to the total number you expect to have if you intend to add nodes for the application later). If your actual operational requirements are less than total nodes available, you can reduce the operational target after verifying that nodes are correctly allocated to the tier.

Set the minimum nodes to 0 or 1. If your actual minimum requirement is more than 1, raise the minimum after you have added at least that number of nodes to the tier. The ccaddnode command will fail if the minimum is not met when you add a node to a tier.

Minimum Nodes

0 or 1

Maximum Nodes

MaxNodes

On the networks page, select the network for the tier, and select the Primary Network radio button.
On the IPs and Hostnames page, select the Specify Manually radio button, then reserve the IP addresses of the nodes that will run in this tier.

Cassatt Active Response creates the tier.

When the tier is created, do not allocate or activate: Cassatt Active Response does that for you when you add nodes (in the next step). Allocation and activation fail when attempted prior to adding nodes (in the next step).

top

Step 4: Add nodes to the tier

In this step, you will add nodes to the tier.

Copy each properties files for each node to the control node.

Don't forget to make sure your properties files get backed up in case you need to remove and readd the nodes. For more information, see Backing up and Restoring Cassatt Active Response.

Make sure all nodes are running and power controllers connected.
On the control node, log in as root run the ccaddnode command, for example:

/opt/cassatt/bin/ccaddnode -f /path/filename -u username -p password

Where:
/path/filename is the absolute path to the properties file for the node
username is a Cassatt Active Response login username with administrative privileges
password is the Cassatt Active Response login password associated with the username

Cassatt Active Response does several things:
1. Unless you used the -s option (--skipkeypropagation), propagates the ssh keys to the node and adds the node to the known hosts file on the control node. You should use the -s option for nodes with OSs that do not support SSH when Cassatt Active Response has been set up to bypass graceful shutdown for those nodes.
2. Validates the properties file for required values and for values that must match values in Cassatt Active Response, as described in the prerequisites.
3. Validates that the tier containing the IP of the node has Automatic Move to Maintenance Pool set to Off.
4. If this is the first node for the tier, activates the tier so that Cassatt Active Response does not reboot nodes as you add them.
5. Allocates the node to your tier.
Repeat these steps until you have added all the nodes to your tier.

What happens when you add more nodes than the tier needs?

If a tier is at its operational target when you add a node, Cassatt Active Response puts the new node into the free pool and powers it off.

Cassatt Active Response does not allocate the node to any other tier: nodes added via the ccaddnode command can be allocated only to the tier for which the IP address of the node has been reserved.

Special note about multi-node power controllers

When you add a node to Cassatt Active Response using the ccaddnode command, and it is connected to an external power controller that does not already exist within Cassatt Active Response, the command creates the power controller in Cassatt Active Response with slots up to the power device ID number of the node you added.

For example, if you add an IBM BladeCenter blade with a power device ID of 5 and a power controller of type BCMM, Cassatt Active Response creates a BCMM with empty slots 1–4 and slot 5 populated with your node. If you then add a blade with a power device ID of 9, Cassatt Active Response adds empty slots 6–8 and slot 9 populated with your node to the BCMM.

Cassatt Active Response does not create any slots above the power device ID number of a node you add, regardless of how many slots the power controller actually has. Why not? Because Cassatt Active Response doesn't know how many slots there might be beyond the highest slot number you provide for a node.

Managing manually allocated tiers and their images and nodes

Many activities surrounding management of manually allocated tiers are identical to management activities for standard images and tiers, as described in Tier and Image Management. Some actions, however, are irrelevant with manual allocation: any action that affects your business applications is unnecessary because Cassatt Active Response manual allocation images don't contain any software (which instead is installed directly on your application nodes). And some actions have implications you might need to understand in advance.

The next several tables provide additional information about each action you can take on manual allocation images, manually allocated tiers, and manually allocated nodes.

Image actions

Action	Notes
New Version	Works as usual.
Duplicate	Works as usual.
Configure Image/Configuration Complete	Not necessary. This action is irrelevant to manual allocation images and will fail because the image remains a stub, and does not contain any software.
Verify Monitors/Verification Complete	Not necessary. This action is irrelevant to manual allocation images and will fail because the image remains a stub, and does not contain the monitoring components.
Activate/Deactivate	Works as usual.
Delete	Works as usual.

Tier actions

Action	Notes
Allocate/Deallocate	Not necessary during initial tier creation, when manually allocated tiers are allocated via the ccaddnode command, not via the Cassatt Active Response Controller. If you do attempt to allocate a tier using this action prior to manually adding nodes, the action will fail as Cassatt Active Response attempts to locate nodes with the IP addresses reserved for the tier. Works as usual for deallocation/allocation once the nodes are added to the tier using ccaddnode.
Activate/Deactivate	Works as usual.
Increase Capacity/Decrease Capacity	Works as usual. Increasing capacity is limited to nodes that have been manually allocated to the tier, but which were moved to the free pool when the operational target was less than the number of nodes available.
Delete	Works as usual.
Update Image	Not necessary. This action is irrelevant to manual allocation images because the image remains a stub, and does not contain any software.
Personalize/Personalization Complete	Not necessary. This action is irrelevant to manual allocation images because the image remains a stub, and does not contain any software to personalize. Personalizing a tier fails if you do not have maximum nodes available for the tier.

Node actions

Action	Notes
Power On	Works as usual.
Power Off	Works as usual.
Reboot	Works as usual.
Shut Down	Works as usual.
Move to Free Pool	Works as usual. Be aware that the node will be powered off.
Move to Maintenance Pool	Do not move manually allocated nodes to the maintenance pool. If you do, Cassatt Active Response attempts to boot the nodes with the diagnostics image, which fails unless the node is configured to PXE boot and is on the Cassatt Active Response network. If this happens, delete the nodes and readd them using the ccaddnode command.
Enable Node	Works as usual.
Disable Node	Works as usual.
Enable Image Instance	Works as usual.
Disable Image Instance	Works as usual.
Clear Failure	Works as usual.
Delete Node	Works as usual. Be aware that the node must be powered off before you can delete it.
Move to Discovered Pool	Do not move manually allocated nodes to the discovered pool. If you do, Cassatt Active Response attempts to boot the nodes with the inventory image, which fails, unless the node is configured to PXE boot and is on the Cassatt Active Response network. If this happens, delete the nodes and readd them using the ccaddnode command.

Troubleshooting

Description

The ccaddnode command fails with the following error:

return code of (cccopykey root IPaddress) is: 1

Resolution

Make sure the node is online and you can ping the node from the control node.

top

Was this article useful? Tell us what you think.
Email infocentral@cassatt.com.

Privacy | Legal | Feedback