Professional Documents
Culture Documents
Synchrony Platform
Version 4.3
11 May 2010
Copyright Axway Software, 2010 All rights reserved. This documentation describes the following Axway software: Synchrony. No part of this publication may be reproduced, transmitted, stored in a retrieval system, or translated into any human or computer language, in any form or by any means, electronic, mechanical, magnetic, optical, chemical, manual, or otherwise, without the prior written permission of the copyright owner, Axway Software. This document, provided for informational purposes only, may be subject to significant modification. The descriptions and information in this document may not necessarily accurately represent or reflect the current or planned functionalities of this product. Axway Software may change this publication, the product described herein, or both. These changes will be incorporated in new versions of this document. Axway Software does not warrant that this document is error free. Axway Software recognizes the rights of the holders of all trademarks used in its publications.
Preface
The purpose of this document is to help you with the installation and management of Synchrony components in a cluster environment. It does not replace the Synchrony Installation and Prerequisites Guide, but it gives you more information on: Cluster architecture Planning an installation of Synchrony on a cluster Management basics of HACMP, Sun Clusters, HP Serviceguard, Veritas Storage Foundation, High Availability, Red Hat Linux Cluster Suite and Windows MSCS
This document is intended for users that have to install a Synchrony component or solution in a cluster architecture.
ASSUMED KNOWLEDGE
In this guide, we assume that the reader has a working knowledge of UNIX and Windows.
RELATED DOCUMENTATION
Synchrony Cluster is accompanied by: Synchrony Installation and Prerequisites Guide (HTML) All Synchrony Cluster documentation is available from the Axway Support website: https://support.axway.com.
Preface
Contents
6 6
6 8
Advanced cluster architecture 2 Installation prerequisites HACMP Sun Cluster HP Serviceguard Red Hat Cluster Suite Veritas Storage Foundation MSCS 12 UNIX users License keys
UNIX license keys Windows license keys
9 11 11 11 11 12 12
13 13
13 14
Cluster installation Cluster installation checklist Installing the first node Managing Synchrony scripts on UNIX
Script descriptions Script installation
15 15 16 17
17 17
18
18 19
Installing additional nodes Installing scripts on the second node Testing a Synchrony Component on the second node 4 Cluster configuration
19 20 20 21
Contents
21
21 21
Applying patches
Automator Composer Gateway Integrator InterPlay Messaging ProcessManager Sentinel Transfer CFT
22
22 22 22 23 23 23 23 24 24
25 25
25 25
HACMP
HACMP Concepts HACMP management basics
25
25 27
Sun Cluster
Sun Cluster concepts Sun Cluster management basics
32
32 33
HP Serviceguard
HP Serviceguard concepts HP Serviceguard management basics
36
36 37
39
39 39 40 40 42 44
46
46 46 47
MSCS 51
MSCS Concepts MSCS: Management 51 52
Contents
1 Cluster architecture
IN THIS CHAPTER
This chapter describes the basics of cluster architecture. It includes a description of both standard and advanced cluster architecture.
Network
Client
Server + Disk
Figure 1: Standard architecture with one server
In a classic cluster environment, two servers are used. They share a disk and the same address. Each server has a local disk from where it boots and where the applications can store static files (binary, jar, configuration files, etc.). The data is stored on the shared disk. When one server is active: it keeps the IP address, mounts the shared disk, and starts the applications. The other server is passive but it checks the status of the active server (i.e. the communication between the two nodes) using a heartbeat. If there is an error or if the first node does not respond, the second node becomes active. It takes the IP address, mounts the shared disk and starts the applications. Before doing this step, it verifies that the other node has really stopped. This operation is called "failover." For the client application, the scenario is exactly the same as if the application had been stopped and immediately restarted.
First case
In Figure 1, the server on the left is active and the client applications are connected using the IP address "ClusterIP". http://ClusterIP
Client
Network
Local Disk
Local Disk
Shared Disk
Second case
In Figure 2, a failover is done and the server on the right becomes active. The client application can open a new connection using the same IP address ("ClusterIP").
http://ClusterIP
Client
Network
Local Disk
Local Disk
Shared Disk
These environments are not standardized and consequently the commands are different on each type of cluster.
Cluster architecture
Cluster behavior
In this section, we assume that the two nodes of the cluster are stopped. Follow these steps to start the cluster environment: The first node is started. It boots and starts its environment. The Cluster Manager detects that it is alone. So it decides to start the group. This group contains a disk, an IP address and an application. The disk is attached to the CPU and the shared directory is mounted. Then, the application is started. After a few seconds the Cluster Manager, which has been set up in the resource definition, begins to monitor the application periodically. When the first IP connection with the address attached to the group arrives, the node accepts it. Then, the second node is started. It boots and starts the Cluster Manager. It joins the cluster and sees that the other node is running and active. It goes into passive mode. A failover can occur for different reasons: A planned action A non recoverable error in a group o A hardware problem In all of these cases, the first node becomes inactive. The second node detects the failure and starts its own processes. It attaches the disk, sends a request to the IP router to clean its routing table and starts the application.
o o
When the first IP connection with the address attached to the group arrives, the new active node accepts it.
Cluster architecture
Figure 4:
nodes
Cluster architecture
10
2 Installation prerequisites
IN THIS CHAPTER
The installation prerequisites described in this chapter are provided in addition to the standard Synchrony components installation prerequisites.
HACMP
You can install a Synchrony Component into a pre-existing HACMP cluster configuration that already contains groups and resources. Alternatively, you can create a new cluster with a group dedicated to the Synchrony component you plan to install. The group that you select as the destination group for your installation must have the following minimum set of resources: an IP Address a hard drive For more information, refer to the section in chapter 5 on HACMP concepts. The creation of HACMP clusters and groups is beyond the scope of this documentation. For detailed information, refer to IBMs documentation.
Sun Cluster
You can install a Synchrony Component into a pre-existing Sun Cluster configuration that already contains groups and resources. Alternatively, you can create a new cluster with a group dedicated to the Synchrony component you plan to install. The group that you select as the destination group for your installation must have the following minimum set of resources: an IP Address a hard drive For more information, refer to the section in chapter 5 on Sun concepts. The creation of Sun clusters and groups is beyond the scope of this documentation. For detailed information, refer to Suns documentation.
HP Serviceguard
You can install a Synchrony Component into a pre-existing HP Serviceguard configuration if the package is already installed.
Installation prerequisites
11
The package that you select as the destination group for your installation must have the following minimum set of resources: an IP Address a hard drive For more information, refer to the section in chapter 5 on HP Serviceguard concepts. The creation of an HP Serviceguard package is beyond the scope of this documentation. For detailed information, refer to your HP documentation.
MSCS
You can install a Synchrony component into a pre-existing Windows cluster configuration if the MSCS cluster already contains groups and resources. Alternatively, you can create a new cluster with a group dedicated to the Synchrony component you plan to install.
Installation prerequisites
12
The cluster that you select as the destination cluster for your installation must have the following minimum set of resources: an IP Address a Network Name a hard drive For more information, refer to the section in chapter 5 on MSCS concepts. The creation of MSCS clusters is beyond the scope of this documentation. For detailed information, refer to the MSCS documentation on the Microsoft website.
UNIX users
In UNIX, users can be classified in two main categories: root user: a user with special administration permissions standard user: a standard user without special permissions The first category puts a root user together with, for example, the DBA (Database Administrator) or MQSeries Administrator. The second category includes the application users who only have read permissions for the use of objects or files managed by the administrators. They work mainly from their home directories. A standard user is required to install Synchrony components. The only exception to this rule is the installation of the component Automator, which requires a root user. A standard user is required for installations on a cluster, but a root user is required to configure the Cluster Manager (HACMP). There are steps which are done by a standard user dedicated to the installation of Synchrony and other steps which are done by a root user. The type of user required is indicated in the introduction paragraph of each section. The terms used are: root user for the first category (cluster administrator) standard user for the second category
License keys
UNIX license keys
Synchrony components need two license keys to be installed on a cluster, one per node. For Windows, a third key may be needed and this case is described in the next section. The table below lists the license key information for each Component:
Component Automator Composer Gateway Number of license keys or specific behavior Only one key for the cluster One key per node One key per node
Installation prerequisites
13
Number of license keys or specific behavior One key per node One key per node One key per node Only one key for the cluster One key per node One key per node
Installation prerequisites
14
3 Cluster installation
IN THIS CHAPTER
This chapter describes how to install Synchrony components in a cluster environment. Before starting the installation, the following information is necessary (in addition to the standard installation information): The hostname of each node The shared directory of the group where the Component will be embedded The IP address of the group where the Component will be embedded The installation of a Synchrony Component on a cluster configuration is done directly by the Synchrony Installer. The following is performed: Installation on the first node Installation on the second or additional nodes To complete these steps, the Installer needs to be able to reach the shared disk. Consequently, the group of the cluster where this disk is embedded must be active on the nodes where the steps are executed.
3. Install the Component on the first node (refer to section Installing the first node). 4. Install the scripts and test the Component (refer to sections Managing Synchrony scripts on UNIX and Testing a Synchrony Component on the first node). 5. Move the group to the second node (refer to chapter 5 Cluster management). 6. Install the component on the second node (refer to section Installing additional nodes). 7. Test the Component on the second node (same test as step 3). 8. Configure the Cluster Manager (refer to chapter 5 Cluster management). 9. Test for failover (refer to chapter 5 Cluster management).
Cluster installation
15
Node 1
Node 2
Database
Local Disk
From the Synchrony Installer follow these steps to install the first node: 1. In the Installation mode screen select Advanced. 2. In the Architecture type screen select Cluster. 3. In the Cluster installation screen select First node. 4. In the Synchrony directories screen select the path to your local Installation directory and type the path to your Shared directory.
16
The short name of the component without a version number (integrator, messaging, etc.)
Example: The script to start Synchrony Integrator is named start_integrator_cluster. Note: For Red Hat Linux, there is only one script with the syntax: component>_cluster
Script installation
These scripts are provided as a kit built with two sub-directories: common and Component. As the name indicates, the common directory can be shared by all Synchrony components.
Cluster installation
17
The Component directory contains two mains scripts: Install_<component> Env_<component> To install the kit: 1. Locate the directory where the cluster scripts must be installed. This information is provided by the configuration manager. 2. If the component is going to be installed first, copy the two directories into the sub-directory. 3. If a Synchrony component has already been installed, copy only the Component directory. 1. In the directory of the component, run the script Install_<component>. After these steps, the Component directory contains the management scripts described in the previous section. Each directory dedicated to a component includes a script that contains all of the parameters used by the other scripts. This script works as a .profile file on UNIX. It initializes the environment variables used by the component. It contains two main variables used to switch to the UNIX installation user and to locate the shared directory of the cluster. It can also contain some variables used by the component itself. It has the following syntax: Env_<component> You must update this file with the following parameters:
Parameter LOCUSR DIR Value The UNIX user used to install the component The local installation directory
Testing on UNIX
The following tests must be run: 1. Run the status script to verify that the parameters in the script are correct and that the return code of the script is 1 (Not on Sun Cluster). 2. Run the start script. 3. Verify that the component is running. 4. Run the status script to verify that the script detects the activity of the component, i.e. that the return code is 0 (Not on Sun Cluster).
Cluster installation
18
5. Run the stop script. 6. Verify that the component is correctly stopped. 7. Restart the component using the restart script. 8. Verify that the component is running. 9. Do some smoke tests with the component. For these tests, use the IP address of the group of the cluster.
Testing on Windows
The following tests must be run: 1. If the component can be started in the command mode, start and stop it to validate the installation. 2. If the component has been installed in service mode, start the service. 3. Do some smoke tests with the component. For the test, use the IP address of the cluster group.
Cluster installation
19
Node 1
Node 2
Local Disk
During the second step, the Installer retrieves information from the configuration file stored on the shared disk. The Installer clones the local disk of the additional node. The installation is launched on the second or an additional node using the following options: 1. In the Installation mode screen select Advanced. 2. In the Architecture type screen select Cluster. 3. In the Cluster installation screen select Additional node. 4. In the Synchrony directories screen select the path to your local Installation directory and type the path to your Shared directory. After this step, the Installer retrieves the parameters used during the installation of the first node. Depending on the Synchrony Component, it may display these parameters. It performs the installation on the local disk, but it does not update the shared disk or the database.
20
4 Cluster configuration
IN THIS CHAPTER
This chapter describes how to install Synchrony components in different situations as well as how to apply patches to the components in a cluster environment.
Cluster configuration
21
Applying patches
Note: Applying patches must be performed in the advanced installation mode of the Synchrony Installer. Applying patches to a cluster environment is different than a standard environment. Some Synchrony components support "Hot Patching", which means that you can apply a patch to the passive node while the Component is in use on the other node. This section describes how to apply patches to each Component that supports clustering. In each component example below A is the active node and B is the passive node.
Automator
To apply a patch to Automator (Note: This method is valid for the Production and the Domain Sever (The Modeling Server cannot be installed on a cluster): 1. Make sure A is the active node. 2. Stop Automator: In HACMP & HP Serviceguard use the suspend script. In Sun Cluster use the disable command (need to be a root user). o In MSCS use the stop command. 3. Apply the patch.
o o
4. Restart Automator:
o o o
In HACMP & HP Serviceguard use the restart script. In Sun Cluster use the enable command. In MSCS use the start command.
Composer
To apply a patch to Composer: 1. Make sure A is the active node. 2. Stop Composer: In HACMP & HP Serviceguard use the suspend script. In Sun Cluster use the disable command (need to be a root user). o In MSCS use the stop command. 3. Apply the patch.
o o
4. Restart Composer:
o o o
In HACMP & HP Serviceguard use the restart script. In Sun Cluster use the enable command. In MSCS use the start command.
Gateway
To apply a patch to Gateway: 1. Make sure that A is the active node. 2. Apply the patch to node B (the passive node).
Cluster configuration
22
3. Switch nodes (failover). 4. Apply the patch to node A, which is now passive.
Integrator
To apply a patch to Integrator: 1. Make sure that A is the active node. 2. Apply the patch to node B (the passive node). 3. Switch nodes (failover). 4. Apply the patch to node A, which is now passive.
InterPlay
To apply a patch to InterPlay, you must follow these steps: 1. Make sure that A is the active node. 2. Apply the patch to node B (the passive node). 3. Switch nodes (failover). 4. Apply the patch to node A, which is now passive.
Messaging
To apply a patch to Messaging: 1. Make sure that A is the active node. 2. Apply the patch to node B (the passive node). 3. Switch nodes (failover). 4. Apply the patch to node A, which is now passive.
ProcessManager
To apply a patch to ProcessManager, you must follow one of these sets of steps depending on if the patch can be applied while the component is in use. If the patch can be applied while the Process Manager is in use: 1. Make sure that A is the active node. 2. Apply the patch to node B (the passive node). 3. Switch nodes (failover). 4. Apply the patch to node A, which is now passive. If the patch cannot be applied while the ProcessManager is in use: 1. Make sure A is the active node. 2. Stop ProcessManager. In HACMP & HP Serviceguard use the suspend script. o In Sun Cluster use the disable command (need to be a root user). o In MSCS use the stop command. 3. While ProcessManager is stopped, apply the patch to A, the active node. During this step, the patch asks if it must run an update to the database, answer YES.
o
Cluster configuration
23
In HACMP & HP Serviceguard use the restart script. In Sun Cluster use the enable command. In MSCS use the start command.
Sentinel
To apply a patch to Sentinel: 1. Make sure A is the active node. 2. Stop Sentinel. In HACMP & HP Serviceguard use the suspend script. o In Sun Cluster use the disable command (need to be a root user). o In MSCS use the stop command. 3. Apply the patch.
o
4. Restart Sentinel.
o o o
In HACMP & HP Serviceguard use the restart script. In Sun Cluster use the enable command. In MSCS use the start command.
Transfer CFT
To apply a patch to Transfer CFT, you must follow these steps: 1. Make sure that A is the active node. 2. Apply the patch to node B (the passive node). 3. Switch nodes (failover). 4. Apply the patch to node A, which is now passive.
Cluster configuration
24
5 Cluster management
IN THIS CHAPTER
This chapter includes basic cluster management guidelines for Cluster Administrators.
HACMP
HACMP Concepts
This section describes selected HACMP concepts that are useful for installing Synchrony components in AIX cluster configuration.
Clusters
In HACMP, a cluster is a configuration of multiple nodes. The HACMP software supports from two to thirty-two nodes in a cluster.
Cluster management
25
Nodes
Nodes form the core of an HACMP cluster. A node in the Synchrony HACMP environment is a processor that runs: AIX HACMP software One or more Synchrony component applications In an HACMP cluster, each node is identified by a unique hostname. A node may own a set of resources: disks, volume groups, file systems, networks, network addresses, and applications. Typically, a node runs a Synchrony component server application that accesses data on the shared external disks. Each node has access to one or more shared external disk devices. A shared external disk device is a disk physically connected to multiple nodes. The shared disk stores mission-critical data typically mirrored or RAID-configured for data redundancy. A node in an HACMP cluster must also have internal disks that store the operating system and the application binaries. These disks are not shared.
Resource groups
Resource groups are a set of cluster resources handled as one unit by HACMP. Resource groups are configured by the user. You can configure startup, failover and fallback policies, customizing the resource group behavior to your needs. An HACMP resource group contains a set of different resources. Synchrony components are viewed in HACMP as Application resources.
Resources
A resource in HACMP terminology is a cluster entity, such as a disk, file system, network interface, application monitor or Synchrony component server that is made highly available.
Application resources
Applications are managed by defining the application in HACMP as an application resource. The application server includes application start and stop scripts. HACMP uses these scripts when the application needs to be brought online or offline on a particular node, to keep the application highly available. This object must be associated with a resource group.
Cluster management
26
F1=Help F9=Shell
F2=Refresh F10=Exit
F3=Cancel Enter=Do
F8=Image
Cluster management
27
The name of the script that stops the Synchrony component (refer to section Managing Synchrony scripts on UNIX)
X
F4=List F8=Image
Cluster management
28
Field Monitor name Application Server(s) to monitor Monitor mode Monitoring method Monitor interval Hung monitor signal Stabilization interval Restart count Restart interval Action on application failure Notify method Cleanup method Restart method
Value Generally, the same name as the component to be monitored. For example Integrator. The name of the component to be monitored. For example; Integrator. Keep Long-running-monitoring The name of the script to be used to monitor the Synchrony component (refer to Managing Synchrony scripts on UNIX). Enter 30 Enter 9 Enter 30 Enter 3 Enter 198 Enter failover Leave blank Leave blank The name of the script to be used to restart the Synchrony component (refer to Managing Synchrony scripts on UNIX).
Cluster management
29
Change/Show resources for a resource group (standard) Do not select: Change/Show a resource group
3. From the list of available cluster groups, select a group. The following screen displays Rg_Http as an example.
Change/Show All Resources and Attributes for a Resource Group Type or select values in entry fields. Press Enter AFTER making all desired changes. [Entry Fields] Rg_Http jack joe
Resource Group Name Participating Nodes (Default Node Priority) Startup Policy Fallover Policy Fallback Policy
Online On First Available Node Fallover To Next Priority Node In The List Never Fallback + + + + +
Service IP Labels/Addresses [william] Application Servers [test] Volume Groups [san] Use forced varyon of volume groups, if necessary true Filesystems (empty is ALL for VGs specified) [/intranet]
F4=List F8=Image
In this example, one application named test is managed by HACMP. It is possible to have a set of applications managed in the same group. In the example used in this document, the list of applications made are with the resources Test and Integrator. 4. Make your changes and validate the updates with the enter key. The following screen is displayed:
COMMAND STATUS Command: OK stdout: no stderr: no
F2=Refresh F9=Shell
F6=Command /=Find
Cluster management
30
Verification to be performed on the following: Cluster Topology Cluster Resources Retreiving data from available cluster nodes. Verifying Cluster Topology...
----------- follow-up to the verify command -----------------[MORE...53] F1=Help F8=Image n=Find Next F2=Refresh F9=Shell F3=Cancel F10=Exit F6=Command /=Find
F1=Help F9=Shell
F2=Refresh F10=Exit
F3=Cancel Enter=Do
F8=Image
Cluster management
31
3. When the target node is selected, the Manager starts the group and displays the following screen:
COMMAND STATUS Command: running stdout: yes stderr: no
Before command completion, additional instructions may appear below. Attempting to bring group Rg_Http until this indicator Wait offline on node joe.
switches to OK Waiting for cluster to process the resource group movement request....
Waiting for the cluster to stabilize...
Sun Cluster
Sun Cluster concepts
This topic describes selected Sun cluster concepts that are useful for installing Synchrony components in a Sun cluster configuration. This topic includes the following sections: Clusters Nodes Groups Applications
Clusters
At Sun, a cluster is a configuration of multiple nodes. The Sun cluster software supports from two to thirty-two nodes in a cluster.
Nodes
Nodes form the core of a Sun cluster. A node in the Synchrony Sun cluster environment is a processor that runs: Solaris Sun cluster software One or more Synchrony component applications
Cluster management
32
In a Sun cluster, each node is identified by a unique hostname. A node may own a set of resources: disks, volume groups, file systems, networks, network addresses, and applications. Typically, a node runs a Synchrony component server application that accesses data on the shared external disks. Each node has access to one or more shared external disk devices. A shared external disk device is a disk physically connected to multiple nodes. The shared disk stores mission-critical data typically mirrored or RAID-configured for data redundancy. A node in a Sun cluster must also have internal disks that store the operating system and application binaries. These disks are not shared.
Groups
Groups are a set of cluster applications handled as one unit by Sun Cluster. Groups are configurable by the user. You can configure startup, failover and fallback policies, customizing the group behavior to your needs. A Sun Cluster group contains a set of different applications.
Resources
Synchrony components are managed by Sun Cluster after being defined as a resource. The resource includes application start and stop scripts. Sun Cluster uses these scripts when the application needs to be brought online or offline on a particular node, to keep the application in high availability mode. This object must be associated with a group.
Cluster management
33
Cluster management
34
Type: Type_version: Group: R_description: Resource_project_name: Enabled{node1}: Enabled{node2}: Monitored{node1}: Monitored{node2}: Resource: Type: Type_version: Group: R_description: Resource_project_name: Enabled{node1}: Enabled{node2}: Monitored{node1}: Monitored{node2}:
SunW.nfs:3.2 3.2 RGnfs default True True True True Synchrony_Integrator SunW.gds:6 6 RGnfs default True True True True
res-nfs-splus
res-nfs-hanfs
Synchrony_Integrator
You can refer to that all the resources are running on node2.
Cluster management
35
HP Serviceguard
HP Serviceguard concepts
Clusters
At HP, a cluster is a configuration of multiple nodes. The HP Serviceguard cluster software supports from two to thirty-two nodes in a cluster.
Cluster management
36
Nodes
Nodes form the core of a HP cluster. A node in the Synchrony HP Serviceguard environment is a processor that runs: HP-UX HP Serviceguard software One or more Synchrony component applications
In an HP Serviceguard, each node is identified by a unique hostname. A node may own a set of resources: disks, volume groups, file systems, networks, network addresses, and applications. Typically, a node runs a Synchrony component server application that accesses data on the shared external disks. Each node has access to one or more shared external disk devices. A shared external disk device is a disk physically connected to multiple nodes. The shared disk stores mission-critical data typically mirrored, or RAID-configured for data redundancy. A node in an HP Serviceguard must also have internal disks that store the operating system and application binaries. These disks are not shared.
Package
Packages are a set of cluster applications handled as one unit by HP Serviceguard. Packages are configurable by the user. You can configure startup, failover and fallback policies, customizing the group behavior to your needs. The applications handled by the package are called Services.
Service
Synchrony components are managed by HP Serviceguard after being defined as a service. The service includes application start scripts. HP Serviceguard uses these scripts when the application needs to be brought online on a particular node, to keep the application in high availability mode. When this script stops, HP Serviceguard takes it into consideration and enters the failover process.
HP Serviceguard: Management
HP Serviceguard can be managed using a command line interface or a graphical user interface (GUI). The GUI is not always available at some sites, therefore this section describes the command line interface.
Cluster management
37
1. In the control file, locate the procedure customer_defined_run_cmds and add the components stop scripts:
# START OF CUSTOMER DEFINED FUNCTIONS # This function is a place holder for customer define functions. # You should define all actions you want to happen here, before the service is # started. You can create as many functions as you need. function customer_defined_run_cmds { # ADD customer defined run commands. : # do nothing instruction, because a function must contain some command. <INSTALLATION-DIRECTORY>/Scripts/Component1/stop_Component1_cluster <INSTALLATION-DIRECTORY>/Scripts/Component2/stop_Component2_cluster <INSTALLATION-DIRECTORY>/Scripts/Component3/stop_Component3_cluster test_return 51 }
2. In the *.conf file (configuration file), locate the section where the services are defined and add the following lines:
SERVICE_NAME Component1 SERVICE_FAIL_FAST_ENABLED no SERVICE_HALT_TIMEOUT 300 SERVICE_NAME Component2 SERVICE_FAIL_FAST_ENABLED no SERVICE_HALT_TIMEOUT 300 SERVICE_NAME component3 SERVICE_FAIL_FAST_ENABLED no SERVICE_HALT_TIMEOUT 300
3. Synchronize the two nodes using the rcp command: rcp -r `pwd` OTHER_NODE:` 4. Check the syntax of the *.conf file: cmcheckconf -p <configuration file>
Cluster management
38
Clusters
A cluster is a configuration of multiple nodes.
Cluster management
39
Nodes
Nodes form the core of a Linux Cluster Suite. A node in the Synchrony Linux Cluster Suite environment is a processor that runs: Red Hat Cluster Suite for Red Hat Enterprise Linux 5 One or more Synchrony component applications
In a Linux Cluster Suite, each node is identified by a unique hostname. A node has a set of resources: file systems (GFS/NFS), networks, network addresses, and applications. Typically, a node runs a Synchrony component server application that accesses data on the shared external disks. Each node has access to one or more shared external disk devices. A shared external disk device is a disk physically connected to multiple nodes. The shared disk stores mission-critical data typically mirrored or RAID-configured for data redundancy. A node in a Linux Cluster Suite must also have internal disks that store the operating system and application binaries. These disks are not shared.
Resources
Synchrony components are managed by Linux Cluster Suite after being defined as a resource. The resource includes application start and stop scripts. Linux Cluster Suite uses these scripts when the application needs to be brought online or offline on a particular node, to keep the application in high availability mode. This object must be associated with a service.
Services
Services are a set of cluster applications handled as one unit by Linux Cluster Suite. Services are configurable by the user. You can configure startup and failover policies, customizing the service behavior to your needs. A Linux Cluster Suite services contains a set of different resources.
This section gives an overview of these two methods, but we strongly advise you to read the Red Hat documentation. The first method is strongly recommended by Red Hat. If the workstation used does not support an X Windows environment, the second method can be used. All the commands must be done by the root user (refer to section UNIX users.).
GUI procedure
Open an X Windows environment and enter the command: system-config-cluster Refer to next page.
Cluster management
40
Typically, the Synchrony component scripts are defined in the Resources property:
Cluster management
41
The link between the resource and the cluster is defined in the Service property.
Cluster management
42
Example: If the header of the cluster.conf file contains the following line <cluster config_version="9" name="name_of_the_cluster"> It must be copied into a new file named cluster.conf.10. The config_version tag must be incremented. Example: <cluster config_version="10" name="name_of_the_cluster"> In order for the updates to be committed, this step is mandatory. Then, the file can be updated. It is possible to validate the configuration with the command: rg_test_test <file_name> Then, commit the update by entering the command: ccs_tool update <file_name> This command commits the new configuration and synchronizes the other nodes of the cluster.
Cluster management
43
With the parameter -i <delay> it displays cluster status and refresh the status every <delay> seconds. With the parameter -s <service> it displays the status of the specified service. Example: The command clustat i 10 s synchrony displays the following status: Service Name Owner (Last) State ------- -------- ---------synchrony node10.example.com started The screen is refreshed every 10 seconds.
Cluster management
44
Example of a complete cluster.conf file: This example comes from a test bed used for this documentation. <?xml version="1.0"?> <cluster config_version="18" name="jstd-aaclus01"> <fence_daemon post_fail_delay="0" post_join_delay="3"/> <clusternodes> <clusternode name="cluster-10.example.com" nodeid="1" votes="1"> <fence> <method name="1"> <device domain="system1" name="cluster-10-fence"/> </method> </fence> </clusternode> <clusternode name="cluster-11.example.com" nodeid="2" votes="1"> <fence> <method name="1"> <device domain="system2" name="cluster-11-fence"/> </method> </fence> </clusternode> </clusternodes> <cman expected_votes="1" two_node="1"/> <fencedevices> <fencedevice agent="fence_xvm" name="cluster-10-fence"/> <fencedevice agent="fence_xvm" name="cluster-11-fence"/> </fencedevices> <rm> <failoverdomains/> <resources> <script file="/etc/init.d/synchrony-messaging" name="synchronymessaging"/> <script file="/etc/init.d/synchrony-process-manager" name="synchrony-process-manager"/> <script file="/etc/init.d/synchrony-integrator" name="synchrony-integrator"/> <script file="/etc/init.d/synchrony-event-router" name="synchrony-event-router"/> <ip address="192.168.0.200" monitor_link="1"/> <fs device="/dev/sda" force_fsck="1" force_unmount="1" fsid="64480" fstype="ext3" mountpoint="/opt/shared" name="shared" options="" self_fence="0"/> </resources> <service autostart="1" name="synchrony"> <ip ref="192.168.0.200"> <fs ref="shared"> <script ref="synchrony-event-router"> <script ref="synchrony-messaging"/> <script ref="synchrony-process-manager"/> <script ref="synchrony-integrator"/> </script> </fs> </ip> </service> </rm> </cluster>
Cluster management
45
Clusters
A cluster is a configuration of multiple nodes.
Nodes
Nodes form the core of a Veritas Storage Foundation. A node in the Veritas Storage Foundation environment is a processor that runs: Veritas Storage Foundation One or more Synchrony component applications In a Veritas Storage Foundation, each node is identified by a unique hostname. A node has a set of Services Group which contains resources as the definition of shared file system, the shared IP address and the application (Synchrony). Typically, a node runs a Synchrony component server application that accesses data on shared external disks. Each node has access to one or more shared external disk devices. A shared external disk device is a disk physically connected to multiple nodes. The shared disk stores mission-critical data typically mirrored or RAID-configured for data redundancy. A node in a Veritas Storage Foundation must also have internal disks that store the operating system and application binaries. These disks are not shared.
Groups
Groups are a set of Resources which have to run together on the same node.
Resources
Resources are a set of cluster applications handled as one unit by Veritas Storage Foundation. Resources are configurable by the user. You can configure startup and failover policies, customizing the services to your own needs. A Veritas Storage Foundation service contains a set of different resources.
Cluster management
46
Description Contains the description of the application itself Contains the definition of the shared IP address These 3 resources define the shared disk The name of the mounted disk can be found in the Mount definition
GUI procedure
The Veritas Cluster Manager must be installed on the PC. Start it and connect to one of the configuration nodes. The following screen is displayed, follow the instructions:
Cluster management
47
This view gives the description of a resource type IP address (the shared IP address).
Cluster management
48
Typically, the Synchrony component is defined as a resource type application which is include in a Service Group. Creation of a resource application To create a resource application: 1. Right-click on the target group and select Add Ressource.
Cluster management
49
3. Enter the full pathname to the start, stop, status and clean scripts.
For example:
Cluster management
50
MSCS
MSCS Concepts
This section describes some selected MSCS concepts that are useful for installing Synchrony components in Windows cluster configurations. Clusters and nodes Groups and resources Virtual servers Failover: Moving groups
Resources
A group has different types of resources. A resource is any entity that can provide a service to a client, can be taken offline and brought online by the MSCS clustering software. The network applications, data files and other tools you install on cluster nodes are cluster resources. A resource is hosted on only one node. You configure the resource to failover from one node to the other in response to a failure. MSCS organizes resources by type.
Resource groups
A resource group consists of dependent or related resources. Dependent resources require other resources in order to operate successfully. Because all resources in a group move between nodes as a unit, dependent or related resources never span a single group boundary (resources cannot be dependent on resources in other groups). A group can run on only one node at a time.
Cluster management
51
Virtual servers
MSCS groups that contain an IP Address resource and a Network Name resource (along with other resources) are published to clients on the network under a unique server name. Because these groups appear as individual servers to clients, they are called virtual servers. Users access applications or services on a virtual server in the same way they would access an application or service were it on a physical server. End users do not need any special training to use clustered resources and do not even need to know that the system is clustered.
MSCS: Management
This section briefly describes procedures you can use to view and manage cluster installations with the MSCS Cluster Administrator and includes the following sections: Opening the MSCS Cluster Administrator About MSCS group resources Viewing cluster resource properties Viewing and modifying MSCS resource groups Displaying active groups Manually moving a group from one node to another Verifying the status of individual nodes
Cluster management
52
Cluster management
53
Cluster management
54
7. In the Dependencies window select the resources. Generally, resources are the disk, the IP address and the Network Name (resources must be online before the application). 8. Click Next. 9. In the Generic service parameters window type the name of the service created during the installation of the Component. 10. Select Use network name for computer name. This changes the hostname retrieved by the application with the Network Name of the group. 11. Click Next. 12. In the Registry replication window, do not add anything (leave it blank). 13. Click Finish to save the changes.
Cluster management
55
Cluster management
56