FAQ's - APPX Software, Inc. : APPX Utility : APPX Installation & Configuration :
How to install Appx on a Cluster | |
ACTIVE/PASSIVE 2 NODE FAILOVER CLUSTER HOWTO APPX 4.2.6 works just perfectly in a Failover Cluster. There were only two details different in the install and configuration from a normal Windows 2003 Server Installation -- 1) the APPX TCP Client Manger (appxdsvc) needed special attention, 2) The APPX Registration needed two serial numbers - one for each NODE. The details: The cluster was a two node Active / Passive cluster with shared storage, via a SAN. NODE 1 Windows 2003 Enterprise Server with Local C: Drive on HBA 1 Server was a Member Server of a Windows 2003 Active Directory Domain NIC 1 on LAN NIC 2 on PRIVATE Heartbeat for Cluster HBA 2 Drive 1 was Quorum drive on SAN (Drive Q:) HBA 2 Drive 2 was Data drive on SAN (Drive R:) Serial Number of Drive C: was xxxx-xxxX NODE 2 Windows 2003 Enterprise Server with Local C: Drive on HBA 1 Server was a Member Server of a Windows 2003 Active Directory Domain NIC 1 on LAN NIC 2 on PRIVATE Heartbeat for Cluster HBA 2 Drive 1 was Quorum drive on SAN (Drive Q:) HBA 2 Drive 2 was Data drive on SAN (Drive R:) Serial Number of Drive C was xxxx-xxxY Installed APPX onto Drive R:\APPX.426 on the SAN. Gave APPX users "Modify" rights to the NTFS permissions of R:\APPX.426. Copied R:\APPX.426\tools\appxdsvc.exe to new directory C:\appx426tools\appxdsvc.exe on BOTH NODE 1 and NODE 2 On NODE that had currently possession of the SAN (NODE 1) I performed the following command from a command window: c:\appx426tools\appxdsvc.exe -install -engine=r:\appx.426\appx.exe I then rebooted the node with possession of the SAN (NODE 1). On NODE that now has current possession of the SAN (NODE 2) I performed the same command from a command window: c:\appx426tools=appxdsvc.exe -install -engine=r:\appx.426\appx.exe You now have APPX installed with a license server configured on both NODES. You can connect to the CLUSTER and initiate a failure on NODE with SAN possession, appx sessions will die, but new sessions can be connected to the CLUSTER without delay. You should import an APPX Cluster registration that has serial numbers from both NODE's C: Drives xxxx-xxxX and xxxx-xxxY. The advantage you have above, is the data is in a single place - hopefully on a RAID'd SAN, this gives data protection against drive failure, but the ease of single location - no need for mirroring software. But unlike most single data location installs, you have server redundancy, provided by the Cluster software. You can certainly make one server the more powerful if your budget dictates, you should just give the powerful server the priority over SAN possession. | |
[Append to This Answer] | |
2005-Feb-02 9:11am |
Previous: | Is there a work around or environment variable that suppresses the "keep key" behavior? |
Next: | How to convert an HFS or Vxfs filesystem from a nolargefiles system to a largefiles system |
|