Micro Focus Enterprise Cloud Services Silk Login     Login     Register
Enterprise Cloud Status
Show Changes
Print
Recent Changes
Find References
History

Search

Recent Topics

ElasticClusterServer

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Elastic Cluster Server
.

Home > Components > ElasticClusterServer

Summary
Advanced use of the EnterpriseLink Cluster Server for use in elastic compute cloud or similar virtual environments.
Note
The following is reproduced from protected content on the EnterpriseLink Information Site and describes the Micro Focus EnterpriseLink Cluster Server background and configuration in a non-cloud environment. This component is automatically installed and configured as part of every MFECS cloud image.

Topics

The ElasticClusterServer is for Micro Focus Server, Micro Focus Studio, Micro Focus OnWeb and Micro Focus EnterpriseLink customers under current maintenance contract and is available for download from www.enterpriselink.com --> Administrators --> Downloads --> EnterpriseLink 5.0 --> Elastic Cluster Server

Background

The EnterpriseLink ClusterServer provides high availability and scalability of a multiple node EnterpriseLink Integration Server environment. It provides a single system image made up of independent nodes shared through a network. In this way individual components may fail but the system as a whole continues to operate. One EnterpriseLink customer using a 12-node cluster reported 99.97% up-time (2 hours downtime) in one calendar year, improving to 100.0% up-time (no scheduled downtime or unscheduled failures) the next. That environment supports over 10,000 named users and typically 2500 concurrent users in a 24 x 7 operating environment.

See TalkAboutClustering to learn about basic ClusterServer design and operation together with best practices involved in deploying it in a production environment. See notes on ClusterServerRecovery and ClusterServerReporting covering those specific areas.

This topic discusses advance use of the EnterpriseLink ClusterServer to support cloud and non-EnterpriseLink environments in a stand-alone environment (one in which no other EnterpriseLink components have been installed).

Elastic Compute Characteristics

Elastic compute environments such as the Amazon Elastic Compute Cloud put additional requirements on fail-over and load balancing components.

There is no difference between the ElasticClusterServer and the ClusterServer other than how they are installed and used. The standard distribution for Windows, UNIX or Linux version of EnterpriseLink (post WebSync 10) contains the updates described here. The ElasticClusterServer is also available for installation without the EnterpriseLink Server and, indeed, without a Web server. Administration and operation is accomplished from a stand-alone Web page.

ElasticClusterServer

Stand-alone Installation

The ElasticClusterServer is part of the EnterpriseLink installation but is also available in a simplified stand-alone installation that doesn't require Install Shield, Microsoft Installer or RPM packaging. Download the stand-alone component from the Micro Focus EnterpriseLink Information Site or Micro Focus SupportLine Web site Copy the downloaded file to the directory that will be used as the run-time execution environment.

On Windows

 cd your-instllation-directory
 unzip eelcs.zip
 elcs -install -start

On UNIX an Linux

 cd your-instllation-directory
 gzip -d eelcs.tar.gz | tar xf -
 sh ./setup.sh

This will create the registry configuration settings, install the operating system service or daemon, and start that service. The default configuration provides for minimum operation with a default license key suitable for testing purposes. See ElasticClusterServer#Configuration for configuration.

Installation will set the service to auto-start when the machine is rebooted. Start the service with the command

 service EnterpriseLinkClusterServer start

or

 /etc/rc.d/rc2.d/S2elcs start

To stop and remove the service,

 service EnterpriseLinkClusterServer stop

or

 /etc/rc.d/rc2.d/S92elcs stop

Stand-alone Configuration

The elcs.html file extracted is a RESTful Web 2.0 interface to the same configuration channel the cluster servers used to talk among themselves. The elcs.html file can be placed under a Web server such as IIS or Apache or it can be used stand-alone by opening the file in a browser.

Note
Under Microsoft Internet Explorer, accept the security warnings that appear. Under Mozilla Firefox, allow local JavaScript execution.

When the configuration file is opened it attempts to connect to the cluster server running on the same machine with IP address 127.0.0.1 and the default cluster port 4125.

It takes about 30 seconds after start before the ClusterServer answers to requests. During this time you may see not connected on the status line.

Fig. 1. Initial state of the Elastic Cluster Server configuration page

Change the IP address and port number will connect and download the configuration from that ClusterServer

Note
The Cluster Server accepts HTTP requests, including those used to extract its configuration, only if it has a destination Web server to which to send those requests. The default configuration sets up a single Web server node, with host name localhost and port 80. At least during the initial configuration you must have a Web server listening on the same machine as the Cluster Server.

If cookies are enabled on your browser the most recent set will be saved in your temporary internet folder and be recalled the next time the page is opened. If a successful connection is made the full configuration is displayed on the page. Changing the IP address or hitting the Refresh button will reread the configuration from the Cluster Server.

Fig. 2. The full cluster server configuration

Note
If the ClusterServer configuration page does not operate you can manually change the configuration by executing regedit (Windows) or editing the EnterpriseLink.cfg file (UNIX/Linux). Registry changes are seen by a running ClusterServer and take effect immediately.
 HKEY_LOCAL_MACHINE\Software\MicroFocus\EnterpriseLink\5.0\ClusterControl

Like the standard ClusterServer the ElasticClusterServer requires a per-server license key for the machine on which it runs. When the ClusterServer is first installed a temporary key is created based on the Cluster IP address 127.0.0.1. If you change the Cluster Server IP address you will have to obtain one of two actions,

  1. Obtain a new EnterpriseLink ClusterServer license key from Micro Focus SupportLine with the following attributes,
    1. Permaent key,
    2. One server,
    3. IP address of Cluster Server,
  2. Use an EnterpriseLink License Server elsewhere on the network.
    1. Install the EnterpriseLink product filling in the License Server Q & A, or
    2. Use regedit to add the following registry keys
 HKEY_LOCAL_MACHINE\Software\MicroFocus\EnterpriseLink\5.0
   \License\PrimaryServerIP="10.20.30.40"
 HKEY_LOCAL_MACHINE\Software\MicroFocus\EnterpriseLink\5.0
   \License\PrimaryServerPort="4124"

Fig. 3. Switching from the built-in license key to the standard License Server

By default, the ElasticClusterServer is set up to work with EnterpriseLink Integration Servers. EnterpriseLink uses session cookies based on the cookie name as project, identified in the URL after .stn/. See Fig. 2.

The ClusterServer can operate with non-EnterpriseLink Web applications that manage state through session cookies. Fig. 4 shows the configuration settings for ASP .NET applications that use ASPSESSIONID.

Fig. 4. Changing the session identification configuration parameters

A video of operating system, network and cluster server configuration in a cluster environment is shown here:

Operations

Operation of the ElasticClusterServer is unchanged from that of the ClusterServer in an EnterpriseLink Integration Server environment. It acts as a proxy server listening on the Cluster Server IP Address and Cluster Server Listen Port and distributes sessions (i.e., multiple HTTP connection requests identified by a common Session Marker HTTP cookie) to available Web Server Nodes. It determines the viability of those nodes to accept work either statically through configuration (offline nodes don't accept any work, quiesce nodes don't accept new session allocations) or dynamically by issuing an HTTP GET request to the Check URL every Check Interval amount of time. If the On Connection Failure condition is met the node is marked offline or quiesce as configured (online indicates that connection failure should not deter session allocation). In addition, the Web server responding to the Check URL may return a page or XML message containing the Check Pattern (if specified). In this case the appropriate Check Action is carried out.

To ensure there is no single point of failure in the environment a ClusterServer will communicate with its Cluster Server Peers through one of two ways. One, via a broadcast message sent to the Broadcast Network every Heartbeat amount of time or, two, via a unicast message to the Cluster Server IP Address itself (both messages over the Broadcast Port). No response within a Fail Over number of seconds will cause a shift from the primary to a secondary cluster server. If the Cluster Server Adapter and Netmask are given an IP address take-over is attempted between the two (the IP address is enabled on the adapter with the mask). If a separate file elcs-acquire-ip is found then the address take over isn't accomplished but that script is invoked to use Elastic IP Address assignment in a Cloud Computing environment or Dynamic DNS host name.

Continuation of the video of the cluster operation is shown here:

Usage

The cluster server responds to the following commands under the special URL /elcs/admin/cmd. Command execution can be disabled or restricted to administrators by other configuration settings.

Command Description
statistics report session count statistics for each node
sessions report information on each session
configuration reload initial configuration
monitorlog turn on extended logging
stopmonitorlog turn off extended logging
memory report on memory usage statistics
disable disable the Cluster IP address
shutdown orderly shutdown of the service
xpeers * Generate Cluster Server control XML peer Cluster Server status
xnodes * Generate Cluster Server control XML Web Server node status
xsessions * Generate Cluster Server control XML session status
xconfig * Process Cluster Server control XML configure
xcommand * Process Cluster Server control XML command from POST data
exit immediate process exit, simulating an uncatchable event
crash simulate internal error and initiate an orderly shutdown
help this short usage
empty return the elcs.html file, the stand-alone configurator
* part of the self-configuration and auto-scaling features; internal use only

Fig. 5. Table of ElasticClusterServer commands.

See Also