How to recover a failed VCS\ Exchange cluster node back into original Organization and Site.
| Article:TECH11151 | | | Created: 2001-01-16 | | | Updated: 2004-01-21 | | | Article URL http://www.symantec.com/docs/TECH11151 |
Problem
How to recover a failed VCS\ Exchange cluster node back into original Organization and Site.
Solution
VERITAS Cluster Server for NT (VCS) provides for
failover if a node fails. Recovering a failed Exchange node requires a precise
procedure. In this example, there is a 2-node VCS\Exchange cluster. The servers
are named VCS102 and VCS103. VCS103 failed in such a way that the entire
Operating System must be reinstalled. VCS102 is hosting Exchange and is
functioning correctly..
Prerequisites:
-
A full backup of both systems.
-
Remove VCS103 from the Exchange Service Group List. This can be done from the
Cluster Manager > Tools > System Manager (Figure
1).
-
A disk drive with at least 100 MB free. This cannot be a system drive or a drive
on VCS102. This drive will temporarily have a reassigned drive
letter.
Figure 1
The procedure is as follows:
On VCS103:
1.
Install Operating System, same Service Pack as on VCS102, current drivers and
proprietary software according to manufacturer requirements.
2.
Install VERITAS Cluster Server into the existing cluster. Reboot
VCS103.
3.
Install Exchange Agent for VCS.
4.
Open Disk Administrator (or Volume Manager if it is employed). Choose a disk
(either from an internal drive on VCS103 or shared storage ) that does not
contain the original Exchange installation. Make sure that there is at least 100
MB free on the disk. Name it the same drive letter that Exchange was installed
on.
For example, the original installation for Exchange
was on Shared Storage Disk 1. During installation, Disk 1 was named E: A
separate disk is selected for the installation of Exchange on VCS103. Disk 2 is
chosen because it has enough free disk space. Give Disk 2 the drive letter E:.
This makes the original data on the live Exchange installation safe from being
overwritten. The new Exchange installation will have drive letter E in its
configuration. It will be identical to VCS102.
5.
VNAME VCS103. For example, if the virtual name that was used before is
VEXCH, the command would be:
%SYSTEMROOT%\PROGRAM FILES\VERITAS\CLUSTER
SERVER\bin\exchange\vname -add VEXCH
6.
Install Exchange on VCS103. Install as new site. However, make sure that the
same site name and organization name are used as on VCS102.
7.
Exit the Exchange Installation.
8.
Open Control Panel > Services.
9.
Put all of the Microsoft Exchange Services to manual.
10.
Set the VERITAS High Availability Service to log in as the same account that you
specified on VCS102.
11.
Close Control Panel.
12.
Reboot VCS103.
13.
Install Exchange Service Pack if necessary. Make sure it is the same version as
on VCS102.
14.
Reboot VCS103.
15.
Unmount the drive that was named drive letter E in step 4. Set to Dot Not
Assign Drive Letter or back to original drive letter (before it was
E).
16.
Add VCS103 back into Exchange Service Group List. It had been removed in the
prerequisite step above.
17.
In Cluster Manager, select File > Save Configuration.
18. In Cluster Manager, right-click on the Exchange Service Group (in the left
pane) and select Switch to. Select VCS103. The Service Group will
failover to VCS103.
After successful failover, the cluster is back in
service.
Thank you.