This version of the Availability Manager has updated configuration and other files. Some of the updates are due to the new features in this version, and some of the updates are to remove text and configuration logicals that are not used by the Availability Manager.

The default location for configuration and log files for the Availability Manager are now in the user’s directory instead of the installation directory. This allows you to set various customizations independent of other Availability Manager users on the system.

The files are stored by default in the AMDS$AM subdirectory. Config and Log subdirectories within the AMDS$AM subdirectory contain the configuration and log files, respectively.

For Windows systems, the default setup creates the folder AMDS$AM in the user’s folder (typically C:\Users\<username>). The configuration and log locations can be overridden by creating the AMDS$AM_CONFIG and AMDS$AM_LOG environment variables before starting the Data Analyzer or Data Server.

The Windows protocol driver allows the Data Analyzer and Data Server to communicate with OpenVMS systems on a local LAN. You must have Administrator privilege on the Windows system in order to install and use this driver. If you do not have Administrator privilege system, or only connect to a remote Data Server, you do not need to install the driver.

A screen in the kit's installation wizard allows you the option of installing the driver. By selecting the checkbox, the driver is installed, requiring Administrator privilege. By leaving the checkbox clear, the Data Analyzer is installed without the driver, and the installation proceeds without the need for Administrator privilege.

The Data Server configuration utility has been added to the Availability Manager Windows installation to configure the IP port number for the network adapters on the system. The new VSI Availability Manager Data Server Guide for Microsoft Windows provides the configuration process instructions.

For OpenVMS, there is only one port number necessary when running the Data Server, and it is specified by the /PORT_NUMBER qualifier. The port numbers available for use are the same as those described in the VSI Availability Manager Data Server Guide for Microsoft Windows.

Monitoring large OpenVMS systems or clusters, especially when viewing the Data Analyzer over a remote connection, could result in a number of DCSLOW and DCCOLT events due to the number of packets exceeding what the Data Analyzer or Data Server could service. In extreme cases, this could result in long delays in updating data in Data Analyzer displays.

The Windows protocol driver has been enhanced to queue up bursts of incoming network packets, allowing the Data Analyzer or Data Server time to service them. The Data Analyzer/Data Server have also been enhanced to allow for concurrent writes of network packets to OpenVMS systems, allowing for faster data collection. Each of these enhancements allows the Data Analyzer and Data Server to service a greater number of OpenVMS systems.

The read queue size and the number of concurrent writes are set to 10 and 2 by default. These values work well for monitoring a large OpenVMS cluster. However, if you experience the above symptoms, you might find it helpful to adjust the read queue length.

HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\AmNdis60\Configuration

This enhancement is the OpenVMS driver equivalent of the enhancements made to the Windows driver in Section 3.2.4, “Windows Driver Queues for Network Packets Enhances Data Analyzer/Data Server Performance”.

The read queue size is set to 20 by default. This value works well for monitoring a large OpenVMS cluster. However, if you experience a number of DCSLOW and DCCOLT events, you might find it helpful to adjust the read queue length.

The default value can be overridden by adjusting the value of AMDS$NUM_RCV_PACKETS in the file SYS$MANAGER:AMDS$LOGICALS.COM. If this value is not in your file, then you can copy the section from the file SYS$MANAGER:AMDS$LOGICALS.TEMPLATE.

Under rare circumstances, the Data Collector would crash when generating a Hello packet. This packet is read by the Data Analyzer, and causes the Data Analyzer to start monitoring the system.

The problem was in queueing a future request to generate next Hello packet, using an OpenVMS TQE for the request. The problem has been fixed.

In OpenVMS Patch kit VMS732_LAN-V0700 introduced a mechanism for tracking buffers holding network traffic from the network. Any software that receives these buffers needs to return them in a specified manner. With this change, using the Data Analyzer on an OpenVMS system for a while would suddenly show the PTHLST event for all remote systems. The workaround was to set SYSGEN parameter LAN_FLAGS bit 16 on.

RMDRIVER has been updated to return the buffers as specified by OpenVMS. Setting LAN_FLAGS bit 16 is no longer necessary.

The Single Disk pane would display the disk operation rate and queue length with values one hundred times the actual value. The values were not correctly scaled to display the actual values. This has been fixed.