Imaging Server Does Not Fail Over

Problem:  When an imaging session is in progress in a clustering environment and a failover occurs, IMGSERV.NLM does not unload properly.

Explanation:  The load and unload commands for IMGSERV.NLM are issued in the ZFDSTART.NCF and ZFDSTOP.NCF files, which are integrated to execute from the load and unload script of the virtual server. The <<y appendage in the script unloads IMGSERV.NLM properly if the unload command is issued through ZFDSTOP.NCF.

However, if IMGSERV.NLM is unloaded through the ZFDSTOP.NCF file from the virtual server unload script, the unload message is first sent to the shared resource console with an accompanying message: Unload module Y/N? The yes answer (already imbedded in the <<y appendage) is sent to the system console, where it is not expected. In this way, IMGSERV.NLM never unloads because it is waiting to receive the Yes command.

Action:  Use the following steps to spell out the "unload imgserv.nlm <<y" in the shared resource unload script.

1. From ConsoleOne^® running on a Windows workstation, click the cluster object > click the shared resource object > click the Unload Script tab to open the script.

2. At the beginning of the shared resource unload script, type the following commands:

   Java -killzenWSImp

   Java -killzenWSRem

   Java -killzenWSInv

   delay 8

   Unload imgserv <<y

   pxestop.ncf

3. Close the shared resource unload script.