Presentation is loading. Please wait.

Presentation is loading. Please wait.

Module 5: Cluster Administration

Similar presentations


Presentation on theme: "Module 5: Cluster Administration"— Presentation transcript:

1 Module 5: Cluster Administration

2 Overview Administration Tools Configuring the Cluster Properties
Creating a Group Creating a Cluster Resource Failover and Failback

3 After you have installed Cluster service, you will use administration tools to change configurations of cluster components and to create new groups and resources. You will also need to configure failover and failback parameters for groups and resources so that Cluster service can manage them in the event of node failures.

4 After completing this module, you will be able to:
Use administration tools. Configure cluster properties. Create a new group. Create a new cluster resource. Set failover and failback parameters for groups and resources.

5 Administration Tools Using Cluster.exe Using Cluadmin.exe

6 After you have installed Cluster service, you will need to perform the maintenance, monitoring, and failover administration of the cluster. There are two cluster administration tools, the Cluster.exe (a command line tool), and the Cluadmin.exe (a graphical user interface tool).

7 Cluster Administration Using Cluster.exe
You can run Cluster.exe from a Microsoft® Windows NT® Workstation or Microsoft Windows NT Server with Service Pack 3 or higher or from other computers running Microsoft Windows® Cluster.exe is installed by default on all cluster nodes. Cluster.exe is also included when you install Cluster Administrator (Cluadmin.exe) on other computers. You can perform the administrative tasks for Cluster service from the Windows command prompt without using the provided graphical interface. While the graphical method provides easier administration and displays the status of cluster resources, Cluster.exe provides the capability for an automated batch file and for administering over a slow wide area network (WAN).

8 Cluster Administration Using Cluadmin.exe
You use Cluster Administrator (%Systemroot%\Cluster\Cluadmin.exe) to perform most of the administrative functions for Cluster service. It is installed by default on both cluster nodes during installation. In addition, Cluadmin can be installed on any Microsoft Windows NT Workstation or Microsoft Windows NT Server with Service Pack 3 or higher or from other computers running Microsoft Windows on the network. Each node of the cluster will have a shortcut to the Cluadmin.exe in the Administrative Tools folder. Using Cluster Administrator, you can create resources and groups, configure the cluster, and set resource failover and failback policies. You can simultaneously connect to multiple clusters in Cluster Administrator.

9 Using Cluster.exe Cluster Commands
Scripting Administration Using Cluster.exe

10 When you install the Cluster Administrator, the command line Cluster
When you install the Cluster Administrator, the command line Cluster.exe program is also installed in the %Systemroot%\System32 folder. You can also use Cluster.exe from a command line or a script to automate many cluster administration tasks. Applications and administration tools use remote procedure calls (RPC) to communicate with the Cluster.exe program.

11 Cluster Commands The basic cluster syntax is: CLUSTER [cluster name] /option The cluster name is optional. If you do not specify a cluster name, Cluster.exe attempts to connect to the cluster that is running on the node on which you are running Cluster.exe.

12 To test connectivity with a cluster, or to ensure that you can use Cluster.exe, check the version name by using the command: CLUSTER [cluster name] /version For specific administration commands using Cluster.exe, see the Job Aid, Cluster Commands, in the Appendix.

13 Scripting Administration Using Cluster.exe
You can script all of the Cluster.exe commands. You can use scripts to build or rebuild a cluster, or to make global changes across many clusters. You can run the scripts through a batch file, for example *.bat or *.cmd. You can use Cluster.exe with a scripting tool, such as Microsoft Visual Basic® Scripting Edition.

14 Demonstration: Administration Using Cluster.exe

15 In this demonstration the instructor will run a script that creates a group, IP number, and network name using Cluster.exe. On the London computer, open Cluster Administrator in one window, and open a command prompt in another window. From the command prompt, change the directory to c:\moc\2087a\labfiles\mscs and run myip.bat. The batch file pauses before and after every remark.

16 REM Line 1 – 4: Create a Group:Mygroup bring the group online, set owners and get status of the group. 1.CLUSTER mycluster group Mygroup /create 2.CLUSTER mycluster group Mygroup /Online 3.CLUSTER mycluster group Mygroup /Setowners:nodea,nodeb 4.CLUSTER mycluster GROUP mygroup /status

17 REM Line 5: Create the IP Address resource: myip.
5.CLUSTER mycluster RESOURCE myip /create /Group:mygroup / Type:"Ip Address"

18 REM Line 6 – 10: Define the IP Address parameters and check Status.
6. CLUSTER mycluster RESOURCE myip /priv network="Cluster Public" 7. CLUSTER mycluster RESOURCE myip /priv address= 8. CLUSTER mycluster RESOURCE myip /priv subnetmask= 9. CLUSTER mycluster RESOURCE myip /priv EnableNetBIOS=1 10.CLUSTER mycluster RESOURCE myip /Stat

19 REM Line 11:View the myip properties.
11.CLUSTER mycluster RESOURCE myip /prop REM Line 12: View the private properties. 12.CLUSTER mycluster RESOURCE myip /priv REM Line 13: Bring online and wait 60 sec. for completion. 13.CLUSTER mycluster RESOURCE myip /Online /Wait:60 REM Line 14: Create a network name resource in mygroup. 14.CLUSTER mycluster RESOURCE mynetname /Create /Group:mygroup /Type:"Network Name"

20 REM Line 15 – 17: Set parameters for mynetname and bring online.
15.CLUSTER mycluster RESOURCE mynetname /priv Name="ClusterSvr" 16.CLUSTER mycluster RESOURCE mynetname /Adddependency:myip 17.CLUSTER mycluster RESOURCE mynetname /Online /Wait:60

21 REM Line 18: Check the status of the Cluster.
18.CLUSTER mycluster GROUP mygroup /Stat REM Line 19: Move the group Mygroup to NodeB 19.CLUSTER mycluster GROUP Mygroup /MoveTo:NodeB /Wait:120

22 Using Cluadmin.exe Cluadmin.exe Console Tree Tool Bar Details Pane
Cluster Administrator – [CLUSTER (.)] File View Window Help For Help, press F1 CLUSTER Groups Cluster Group Resources Cluster Configuration DC1 Active Groups Active Resources Network Interfaces DC2 Name State Owner Resource Type Cluster IP Address Online DC1 IP Address Cluster Name Online DC1 Network Name Disk Z: Online DC1 Physical Disk Console Tree Tool Bar Details Pane

23 When you have installed Cluster service, the graphical user interface of Cluster Administrator (Cluadmin.exe) is installed by default in the %Systemroot%Cluster folder. It is possible to remotely administer a server cluster running Windows 2000 from a client running Windows Cluster Administrator is part of the Admin Pack that is included with Windows 2000 Server editions. You can install Cluster Administrator on all of the Windows family of operating systems.

24 Cluster Administrator
The three main areas of Cluster Administrator are the tool bar, the details pane, and the console tree.

25 The Toolbar You can use the toolbar to change the look of the details pane or to change properties of objects. You can click the icons in the toolbar as an alternative to right-clicking objects in the details pane or the console tree.

26 The Details Pane The right window of Cluster Administrator is the details pane. You can display four views in this area: Large Icons, Small Icons, List, and Details. In the Details view, you can view properties of the containers in the console tree, such as Name, State (for example, online or offline), and Resource type.

27 The Console Tree The console tree in the left window displays objects that can have properties and can hold other objects. Containers such as Groups, Resources, and Cluster Configuration are represented as folders. Cluster Node containers are represented as computer icons. The Groups container displays all of the groups that are configured for the cluster. Groups define the units of failover. For example, when one resource in a group fails and it is necessary to move the resource to an alternate node, all of the resources in the group are moved to the alternate node.

28 The Resources container displays a list of all of the resources of every group. You can see which group the resource belongs to, and you can manage each resource in this view. You use the Cluster Configuration container to keep track of the networking parameters of Cluster service. There is a container for every node in the cluster. Each Node container shows the Active Groups, Active Resources, and Network Interfaces that the node currently controls. When a group fails over from one node to another, the group is removed from the controlling node’s Active Groups container and displayed in the new controlling node’s Active Groups container.

29 Connecting to a Cluster Using Cluadmin.exe
When you start Cluster Administrator, connections that were open when you closed the previous session are restored by default. If no connections were open, Cluster Administrator presents an Open Connection to Cluster dialog box and you can specify which cluster you need to administer. You can also access specific clusters or nodes by adding multiple cluster names, node names, or Internet Protocol (IP) addresses. Note: To prevent all previous cluster connections from reopening, run Cluadmin.exe from a command prompt with the –noreconnect switch. The – noreconnect switch prompts you for a connection to a cluster.

30 You can specify the cluster by using its cluster network basic input/output system (NetBIOS) name or IP address that you assigned at installation. In addition, you can specify the NetBIOS name or IP address of one of the cluster nodes. Connecting to a node instead of the cluster virtual server is useful in case of problems with the cluster name or cluster IP address resources, for example, if the RPC service fails to respond, you would be able to connect to the cluster from one of the nodes in the cluster. From the server running Cluster service, you can press PERIOD to connect to the cluster root by using a local procedure call (LPC) rather than an RPC.

31 The title bar in Cluster Administrator reflects what you entered at the Open Connection to Cluster dialog box. For example, if you entered CLUSTERNAME, the title bar would show Cluster Administrator – CLUSTERNAME (CLUSTERNAME). Entering NODENAME would produce Cluster Administrator – CLUSTERNAME (NODENAME). Connecting to the local server by pressing PERIOD results in Cluster Administrator – CLUSTERNAME (.) in the title bar. Note: If you connect to the cluster root by using the NODENAME and that node goes down, all of the objects become unavailable in Cluster Administrator until the node is back online.

32 Configuring the Cluster Properties
Setting the General and Quorum Properties Changing the Network Priority Changing the Network Usage Changing Security

33 When you install Cluster service, you make choices in terms of where you locate the quorum resource (cluster files) and where you assign network properties and usage. After you have installed the cluster, you may need to change these configurations. To make any of these changes, you can use the cluster name Properties option in the Cluster Administrator tool. To access cluster properties, click the cluster name and either right-click the cluster name, and then click Properties, or on the File menu, click Properties.

34 Changing Cluster Properties
In the cluster name Properties dialog box, you can select the following tabs: � General to change the cluster NetBIOS name and description. � Quorum to change the quorum location. � Network Priority to change the priority order for the internal cluster communications. � Security to add or remove cluster administrators.

35 Setting the General and Quorum Properties
Cluster Administrator – [CLUSTER (server1)] File View Window Help For Help, press F1 Networks Cluster Private Public Network Interfaces SERVER1 Active Groups Active Resources SERVER2 Name State Description Groups Resources Cluster Configuration SERVER1 Up SERVER2 Up New Configure Application Rename Properties CLUSTER Properties General OK Cancel CLUSTER Quorum Network Priority Security Quorum resource: Disk Z Cluster maintenance files Apply Partition: Z: (Clusdrive) Root path: \MSCS\ Reset quorum log at: 64 KB

36 If you need to change the name or description for the cluster, you select the General tab of the cluster name Properties dialog box. If you need to change the quorum resource or configure the quorum log, such as installing a new drive that you will use for the quorum resource, you select the Quorum tab in the cluster name Properties dialog box.

37 Changing Cluster Name and Description
The properties on the General tab are Name and Description. The following table lists the General tab options and their usage. Option Use Name To change the name of the cluster. This option will also change the cluster name resource in the cluster group. Description To provide a logical description of the cluster. This box is often used to reference the physical location of servers or resources.

38 Changing the Quorum Resource Location
The properties on the Quorum tab are Quorum resource, Partition, Root path, and Reset quorum log at. The following table lists the Quorum tab options and their usage. Option Use Quorum resource To select the shared drive to use as the quorum resource. Partition To select the partition on the shared drive which stores the quorum log. Root path To specify the folder in which you will store the quorum log. This setting defaults to \mscs\ Reset quorum log at To configure the size at which the log will start overwriting earlier entries. This setting defaults to 64 kilobytes (KB). If you have many resources in the cluster, you may want to increase the reset limit.

39 Changing the Network Priority
Cluster Administrator – [CLUSTER (server1)] File View Window Help For Help, press F1 Networks Cluster Private Public Network Interfaces SERVER1 Active Groups Active Resources SERVER2 Name State Description Groups Resources Cluster Configuration SERVER1 Up SERVER2 Up New Configure Application Rename Properties CLUSTER Properties General OK Cancel CLUSTER Quorum Network Priority Security Networks used for internal cluster communications: Private Cluster Connection Public Cluster Connection Apply Move Up Move Down Properties… Internal cluster data will only be transmitted on the highest priority available network. Use the Move Up button to raise the priority level of a network. Use the Move Down button to lower the priority.

40 If you need to change the priority in which Cluster service will use the network connections between the nodes of the cluster, you select the Network Priority tab of the cluster name Properties dialog box. The tab also displays the state and subnet mask of this network connection. The following table lists the Network Priority tab options and their usage. Option Use Move Up Raises the priority of the selected network. Move Down Lowers the priority of the selected network. Properties Displays the Properties dialog box for the selected network. You can change the network name in addition to the network usage settings.

41 Note: Whichever mixed or private network is highest in the list has the top priority and will be used for node-to- node communication. A network dedicated for public access only will not appear on this page, and will never be used for private communication between the nodes.

42 Changing the Network Usage
Cluster Administrator – [CLUSTER (server1)] File View Window Help For Help, press F1 Name Network State Adapter Address Description SERVER1 Public Up Intel® 8… SERVER2 Public Up Intel® 8… CLUSTER Groups Cluster Group Resources Cluster Configuration Resource Types Networks Cluster Private Netwo SERVER1 Active SERVER2 Active Groups Active Resources Network Interfaces New Configure Application Properties Rename Private Cluster Connection Properties General OK Cancel Private Cluster Connection Name: Private Cluster Connection Description: Apply Enable this network for cluster use This network performs the following role in the cluster: Client access only (public network) Internal cluster communications only (private network) All communications (mixed network) State: Up Subnet mask:

43 If you need to reconfigure how Cluster service will use a network, expand Cluster Configuration, expand Networks, right-click the network name, and then click Properties.

44 The following table lists the General Network Property tab options and their usage.
Use Name The descriptive name given to the network during installation. Description A description of the network. Enable this network for cluster use Selecting this check box allows Cluster service to use the network adapter. This check box is selected by default. Client access only (public network) Selecting this option will cause Cluster service to use this network adapter for communication with clients. No node-to-node communication will take place when using this network adapter. Internal cluster communications only (private network) Selecting this option will cause Cluster service to use this network adapter for node-to-node communication only. All communications (mixed network) Selecting this option will cause Cluster service to use the network adapter for node-to-node communication and for communication with clients. This option is selected by default.

45 Changing Security

46 Microsoft Cluster service can be configured to allow or deny administrative responsibility to users or groups. To configure cluster security, go to the Properties page of the cluster and select the Security tab. Access to the cluster can be granted or denied only to the local Administrators group on each node, domain users, and global groups. Do not attempt to grant or deny access to local users or local groups (other than the local Administrators group). The local administrator account and local system account always have access to the cluster and cannot be removed from the cluster.

47 There are only two levels of permission for a cluster that can be assigned in Cluster Administrator, Full Control or No Access. If you have Full Control administrative permissions you can fully administer the cluster. However, if you have No Access permission, you cannot connect to the cluster in Cluster Administrator. The privileges necessary to administer a cluster can be granted (or revoked) only on a cluster-wide basis. Cluster service has no facility for granting more granular privileges to cluster resources.

48 Creating a Group Cluster Administrator – [CLUSTER (server1)]
File View Window Help Creates a new cluster group CLUSTER Groups Resources Cluster Co Resour Netwo Clu Pul Netwo SERVER1 Active Network Interfaces SERVER2 Active Groups Active Resources Delete Rename Bring Online Take Offline Move Group Name State Owner Resource Type Cluster IP Address Online DC1 IP Address er Name Online DC1 Network Name Online DC1 Physical Disk New Configure Application Group Resource Properties New Group Disk Group 1 < Back Cancel Next > Name: Disk Group 1 Description: This wizard will help you quickly set up a new resource group. To continue, click Next. Preferred Owners List all preferred owners on the right, and then arrange them in the order of preference. Add -> <- Remove Move Up Move Down Available nodes: Name Preferred owners:

49 Groups enable you to combine resources into larger logical units
Groups enable you to combine resources into larger logical units. Typically a group is made up of related or dependent resources, such as applications and their associated peripherals and data. However, you can also establish groups with resources that are unrelated and independent to balance the load or for administrative convenience. At any point in time, only one node owns a group. A single group always owns a resource. These relationships ensure that all of a group’s members reside on the same node. You can fail over groups or move them from one system to another as atomic units that Cluster service manages. Therefore, at any given instance, different systems across the cluster cannot own different resources within the same group.

50 You can set a node as a preferred owner of a group
You can set a node as a preferred owner of a group. The preferred owner setting is used for a failback policy for the group. If one node is selected as the preferred owner of the group, the group will always fail back to the original owner. If both nodes are selected as preferred owners, when the group fails over, it will continue to run on the new owner. The group has no failback policy. If you do not select any preferred owners for the group, the group can run on either node and will not have a failback policy.

51 The rules governing groups include:
� A group is active on one, and only one, node of a cluster at a time. � Different nodes in the cluster cannot own different resources within the same group. � A resource cannot belong to more than one group. � A dependency tree cannot span groups.

52 Creating a New Group To create a new group, in the Cluster Administrator tool, right-click Groups, select New, and click Group. Enter the group name and description. Select any Preferred owners only to enable the failback policy.

53 Changing Group Properties
In the cluster group Properties, you can select the following tabs: General to name or rename the group, give a description to the group and set the preferred owners. Failover to configure a threshold and period for the restart function. Failback to configure a failback policy to prevent or allow a failback.

54 Creating a Cluster Resource
Cluster Administrator – [CLUSTER (.)] File View Window Help Creates a new cluster group CLUSTER Groups Cluster Group Project Files Resources Cluster Configuration Resource Types Networks Network Interfaces DC1 Active Groups Active Resources DC2 Cascade Tile Horizontally Tile Vertically Close All View New Window Disk Z: Online DC1 Clust Cluster IP Address Online DC1 Clust Cluster Name Online DC1 Clust New Configure Application Group Resource Name State Owner Grou

55 A cluster resource is any physical or logical component that can be:
Brought online and taken offline. Managed in a server cluster. Hosted (owned) by only one node at a time.

56 When you create a cluster resource, you must step through a series of dialog boxes. You will need to designate the group to which you want the resource to belong. After you have created a resource and added it to a group, you can move it to another group. When moving a resource, all dependencies relating to that resource will also move.

57 Changing Resource Properties
If you need to change the properties of an existing resource, click on the resource and right-click to select Properties. You can select the following tabs to change the resource properties. Tab Use General To name or rename the resource, give a description to the group, set the possible owners and choose a separate resource monitor. Dependencies Advanced To configure a restart policy. Parameters To configure individual resource settings.

58 Resource Type and Possible Owners

59 A resource can only belong to one group
A resource can only belong to one group. Some resources may have dependencies. You add resources to a group through configuring a new resource within Cluster Administrator. You can also create or add resources from the Resources folder.

60 To create a resource, click File, select New, and then click Resource.
The following table lists the New Resource page options and their usage. Option Use Name The unique name of the resource. You can change the resource name, because internal resources are identified by their globally unique identifier (GUID). Description An optional description of the resource. Resource Type A drop-down list of available resources. See the following table on the next page for a list of all of the default resources. Group A drop-down box of available groups in the cluster. Run this resource in a separate Resource Monitor Sets the resource to run in a separate Resource Monitor. This option is useful for a resource that is not functioning correctly. In this case, the resource will only affect its own Resource Monitor instead of affecting a Resource Monitor that is monitoring multiple resources.

61 Naming the Resource You name a resource for administrative use. For example, there might be ten file share resources on a cluster. You would need to use a naming convention to logically identify each resource. In the New Resource dialog box, you need to enter the resource name and the group that you want the resource to belong to. All of the resources have similar configuration settings. You can enter a description of the resource, which is optional.

62 The following table lists the default resource types.
1. DHCP Service 9. Network Name 2. Distributed Transaction Coordinator 10. Network News Transport Protocol (NNTP) Server Instance 3. File Share 11. Physical Disk 4. Generic Application 12. Print Spooler 5. Generic Service 13. Simple Mail Transfer Protocol (SMTP) Server Instance 6. Internet Information Services (IIS) Service Instance 14. Time Service 7. IP Address 15. WINS Service 8. Message Queuing Note: Some cluster-aware applications add resources to the default resource type list. For example, if you install Microsoft Exchange 2000 on a cluster, you need to open the New Resource dialog box and add the appropriate Exchange resources to a group.

63 Possible Owners You configure failover for a resource by setting its Possible owners. A possible owner can be any node in the cluster that is allowed to run the resource. You can restrict a resource from running on specific nodes within the cluster. In Windows 2000 Advanced Server, a Cluster service resource can have only two possible owners because there are only two nodes in the cluster.

64 If you want a resource or group to run on one node without failover capabilities, select only one possible owner. In such a situation, a failover will not occur because there is no node defined as a possible owner for one or more of the resources within a group. In the Possible Owners dialog box, indicate the nodes on which you can bring the new resource online. In the next dialog box, select any dependencies that this new resource might have.

65 Dependencies and Parameters

66 Resources may or may not have dependencies
Resources may or may not have dependencies. All resources have different parameters, and a unique Parameters dialog box appears depending on the resource type.

67 Dependencies When you create a new resource, you must specify the dependencies of this resource. Dependencies are resources that Cluster service must bring online before the new resource can be brought online. Many of the resource types that ship with Cluster service have requirements for dependencies. The only resource types that do not usually depend upon another resource are the physical disk and IP address. Resources that have mandatory dependencies do not allow you to go to the next screen until you have added them to the Resource dependencies box.

68 Some resources do not require any dependencies, but as a best practice the resource should have a dependency. For example, a File Share resource has no mandatory dependency; however as a best practice a File Share resource should have a Network name of a virtual server and a disk resource.

69 Parameters You will also need to enter information in the Parameters dialog box that is unique to the type of resource. The Parameters dialog box will be different for every type of resource that is added to the cluster. For example, a File Share resource will have a Share name, Path, Comment, User Limit, Permissions and Advance options, while an IP address resource parameter will include the IP address and the subnet mask.

70 Failover and Failback Configuring Failover Configuring Failback
Causing and Monitoring Failover

71 Cluster service automatically initiates failover when it detects a failure on one of the cluster nodes. Because each cluster node monitors both its own processes and the health of the other node, the need for failover is detected with minimal delay. In most cases, Cluster service can detect that a node has failed and begin failover in less than five seconds.

72 A resource group may migrate or fail over to another node for many reasons:
Failure of the original node. Failure of one or more resources in the group at the original node. The resource group prefers to execute at the other node (failback). The operator requests the group to move.

73 If a resource fails, all others resources who are dependent on the failed resource will also fail. Cluster service will attempt to restart the resource. After a configurable amount of restarts, Cluster service will take all of the resources in that group offline and move the group to another cluster node. Cluster service will then attempt to start all of the resources on the new node. If the resource that originally failed does not start, Cluster service will take this resource and all of its dependent resources offline to prevent the group from failing back to the original node.

74 Failover and Failback Policies
Failover and failback policies taken as a whole determine how a group will behave when a failure occurs. Failover occurs when a resource in a group has exceeded its restart threshold, or when multiple resources in a group exceed the group failover threshold. Failover can also occur when a resource fails to restart per the restart threshold. A related process, failback, occurs when a group moves back to the node on which it was active prior to a failover.

75 The following table shows whether you configure the group or the resource property for failover or failback policies. Configure Failover/Failback Preferred owners Group Property Failback Possible owners Resource Property Failover Failover Threshold Restart Threshold

76 Resource Configuration
Configuring Failover Group Configuration Resource Configuration Failover Configuration Groups Possible Owner Resources X Failover Policy Restart Policy

77 The failover policy is a combination of group and resource configurations. To set up a failover policy, you need to configure: A group failover threshold. Possible owners at the resource level. A restart threshold for each resource in the group.

78 Group Failover Policy In the properties of the group, you set the group failover threshold and period on the Failover tab. The following table shows the Failover tab options and their descriptions. Option Description Threshold The maximum number of times that the group is allowed to fail over during the failover period. If this number is exceeded within the failover period, Cluster service will take the group offline. The default setting is ten. Period The maximum number of hours during which the failover threshold cannot be exceeded. If the threshold is exceeded within the failover period, Cluster service will take the group offline. The default setting is six hours.

79 Assuming the default settings, if a group were to fail over for an eleventh time during the six-hour period, the group would be taken offline. If you wish to reset the counter for the failover threshold, you must stop and restart Cluster service on the node that currently owns the group.

80 Resource Failover Policy
To fail over a resource, more than one node needs to be identified as a possible owner. You can change the Possible owners of this resource on the General tab of the resource name Properties dialog box. Note: The resource cannot fail over unless all of the resources in the group list the same possible owners. In the Properties dialog box of the resource, you set the resource failover policy on the Advanced tab by entering values for Restart, Looks Alive/Is Alive poll intervals, and Pending timeout.

81 The following table shows the options and their descriptions for the Advanced tab in the resource Properties dialog box.

82 “Looks Alive” poll interval
Option Description Restart Specifies that Cluster service attempts to restart the resource when it fails. ● Affect the group Specifies whether to fail over the group when this resource fails. ● Threshold The number of times during the specified threshold period that Cluster service will attempt to restart the resource before the group fails over. The default is three. ● Period The number of seconds during which the specified number of attempts to restart the resource must occur before the group fails over. The default is 900 seconds. “Looks Alive” poll interval How often Cluster service checks the registered state of the resource to determine whether the resource appears to be active. You can use either the default interval 5,000 milliseconds for the resource type or specify another interval [in milliseconds]. “Is Alive” poll interval How often Cluster service checks the thread of the resource type when checking to determine whether the resource is active. You can use either the default interval 6,000 milliseconds for the resource type or specify another interval [in milliseconds]. Pending timeout The amount of time in seconds that a resource in a pending state [online pending or offline pending] has to resolve its status before Cluster service puts the resource in the failed state. The default is 180 seconds.

83 Failback Configuration
Configuring Failback Group Configuration Failback Configuration Groups Preferred Owner Resources Failback Policy X

84 To set a resource to fail back means that Cluster service will move a group back to a preferred owner after Cluster service detects that the owner is back online. A group by default has no preferred owners. If you do not configure a preferred owner for a group, when it fails to another node and starts successfully, the group will remain there until an administrator moves the group, or another failover occurs. If you want a group to always run on a specific node if that node is available, select this node as the preferred owner. The Failback tab in the cluster group Properties dialog box allows the administrator to either prevent failback or set failback for a group. To access the Failback tab, click the Failback tab in the group’s Properties dialog box.

85 Failback Settings The following table lists the four failback settings, which are the same for all of the groups. Option Description Prevent failback When the node returns to the online state, the group will not fail back to the node on which it was running before failover. Allow failback When the node returns to the online state, the group will fail back to the original node on which it was running before failover. Immediately Failback will occur as soon as Cluster service detects that the original node is back online. Failback between Failback will only occur between specified hours.

86 Causing and Monitoring Failover
Online Online Pending Offline Offline Pending Failed

87 You can test a failover and failback policy, in addition to resource dependencies, by initiating a failure of a resource. You can test a failover by initiating a failure from either Cluster Administrator tools (Cluadmin.exe or Cluster.exe). To initiate a failure with Cluadmin.exe, right-click the resource, and then click Initiate Failure. If the failover policy includes a restart value, the resource will restart. If you continue to cause the resource to fail, the failover policy will eventually exceed the restart threshold and move the group containing the resource to another possible owner. This technique is useful for testing a newly-created resource, or changes in a resource policy.

88 Each resource has a registered state during the normal operation or when the resource is being failed over. The table lists the resource states. Resource Registered State Description Online The resource is online and working properly. Online Pending The resource is waiting for a dependent resource to come online. Offline The resource is offline. The resource is either preparing to move to another node or an administrator has taken the resource offline. It is offline when it is first created. Offline Pending The resource is waiting for a resource that depends upon it to come offline. Failed The resource could not successfully come online and has failed.

89 Lab A: Creating Cluster Resources

90 Objectives After completing this lab, you will be able to:
Create an IP address resource. Create a name resource. Test for failover of cluster resources.

91 Prerequisites Before working on this lab, you must be familiar with the concepts in Course 2087A, Module 5, “Cluster Administration.” You must also have a server cluster installed and running on both nodes.

92 Lab Setup To complete this lab, you need the following:
Two computers running Microsoft Windows Advanced Server, each with a small computer system interface (SCSI) adapter installed. Each computer in a student pair must have a different SCSI ID. A shared external SCSI disk. Each computer is connected to the shared disk with SCSI cables. The shared external SCSI disk ID must be different than either of the SCSI cards of the computers in the student pair. Two network adapter cards in each node of the cluster.

93 Scenario In this exercise, you will create a virtual server. The virtual server requires a name resource and an IP resource. One student will complete Exercise 1 and create the IP address resource. The other student will then complete Exercise 2 and create a name resource with a dependency of the IP address that was created in Exercise 1. Students will work together to fail over the resources back to Node B. The following exercises will refer to your computers as Node A and Node B. For this lab, Node A will be the computer from which you will assign the IP address, and Node B will be the computer where you create the virtual server and associate it with the IP address in the first exercise.

94 Exercise 1: Create an IP Address Resource
In this exercise, you will create a virtual IP address resource and bring the resource online. The virtual IP is a prerequisite for the virtual server to be installed in Exercise 2.

95 To create an IP address resource
Complete this task from the Node A computer. 1. Click Start, point to Programs, point to Administrative Tools, and then click Cluster Administrator. 2. In the console tree, expand Cluster_Name, then expand Groups. 3. Right-click Cluster Group, point to New, and then click Resource. 4. Refer to the reference table to complete the New Resource dialog box: Name: Virtual Server IP Description: Virtual Server IP Number Resource Type: IP Address Group: Cluster Group 5. Click Next.

96 6. In the Possible Owners dialog box, click Next.
7. In the Dependencies dialog box, click Next. 8. Use the following information to complete the TCP/IP Address Parameters dialog box: • Virtual Server IP Address: Enter the virtual server IP address of Node A as shown on the table. The variable x in the cluster IP address is your classroom number. Subnet mask: Network: Cluster Public 9. Click Finish. 10. In the Cluster Administrator dialog box, click OK. 11. In Cluster Administrator, select Cluster Group, right-click the Virtual Server IP, and then click Bring Online. 12. Close all of the windows.

97 Exercise 2: Create Name Resource
In this exercise, you will create a name resource. When this resource is associated with the virtual IP address created in Exercise 1, you have created a virtual server. Other resources could be configured to run on this virtual server.

98 To create a name resource
Complete this task from the Node B computer. 1. Using Cluster Administrator, right-click Cluster Group, point to New, and then click Resource. 2. Use the following information to complete the New Resource dialog box: Name: Virtual Server Name Description: Virtual Server Name Resource Type: Network Name Group: Cluster Group 3. Click Next. 4. In the Possible Owners dialog box, click Next. 5. In the Dependencies dialog box, under Available Resources, select Virtual Server IP, then click Add.

99 6. Click Next. 7. In the Network Name Parameters dialog box, in the Name: box, type the name of the virtual server from the lab setup sheet and then click Finish. 8. In the Cluster Administrator dialog box, click OK. 9. In Cluster Administrator, select Cluster Group, right- click the Virtual Server Name, and then click Bring Online.

100 Exercise 3: Testing Failover
In this exercise, you will fail over a group from Node A to Node B in the cluster. You will then fail over the same group back to Node B.

101 To move the resources on Node B to Node A
Complete this task from the Node B computer. 1. Click Start, point to Programs, point to Administrative Tools, and then click Cluster Administrator. 2. In the console tree, expand Cluster, and then expand Groups. 3. In the details pane, notice that the owner of all the resources of the Cluster Group belongs to Node B. 4. Right-click on Cluster Group, and then click Move Group. 5. In the details pane, notice that the resources state progresses from offline, to online pending, and then to online. 6. In the details pane, notice that the owner of all the resources of the group now belongs to Node A.

102 Exercise 4: Using Cluster.exe
In this exercise, you will fail over a group from Node B to Node A by and back to Node B by using Cluster.exe.

103 To move a group from Node A to Node B by using Cluster.exe
Complete this task from the Node B computer. 1. Click Start, and then click Run. 2. In the Run command box, type CMD and then click OK. 3. From the command prompt type: CLUSTER clustername GROUP “Cluster Group” /moveto:NodeB /wait:120 4. You should receive a status that the group is Online on Node B.

104 To move a group from Node B to Node A by using Cluster.exe
Complete this task from the Node A computer.

105 1. Click Start, and then click Run.
2. In the Run command box, type CMD and then click OK. 3. From the command prompt type: CLUSTER clustername GROUP “Cluster Group” /moveto:NodeA /wait:120 4. You will receive a status that the group is Online on Node A.

106 Review Administration Tools Configuring the Cluster Properties
Creating a Group Creating a Cluster Resource Failover and Failback


Download ppt "Module 5: Cluster Administration"

Similar presentations


Ads by Google