Recovering From A Corrupted Quorum Drive - IBM totalstorage 326 User Reference

Powering on the Model 326 when clustering is active
1. Power on any UPS that you powered off in the powering off procedure, and
2. Power on any network hubs or switches that you powered off in the powering off
3. Power on all 5192 Network Attached Storage Storage Units and 3534 Managed
4. Power on each 5191 RAID Storage Controller. After about three to four minutes,
5. Power on the node that you shut down last in the powering off procedure.
6. After the node comes up, start Cluster Administrator on that node and make
7. If no problems exist and all clustered resources are online, power on the node

Recovering from a corrupted Quorum drive

Clustering relies on data stored on the Quorum disk to maintain resource
synchronization between the two nodes in the cluster. In the event of a power loss
to both nodes or a hardware failure that corrupts the Quorum data, the cluster
service might not start, leading to the following event log error:
Event ID: 1147
Source: ClusSvc
Description: The Microsoft Clustering Service encountered a fatal error.
The Quorum drive data must be available so the cluster service can confirm that
the cluster configuration on the local node is up to date. If it cannot read the log,
the cluster service does not start to prevent the loading of old configuration data.
To restore the Quorum disk, a Microsoft Windows Backup utility backup of the
System State of the boot drive (C:) of one node must be available. Backing up the
entire boot drive also saves the System State. Backing up the System State
automatically saves the Quorum log and other cluster files.
A Microsoft tool is needed as part of the Quorum restore procedure. This tool is
called Clusrest.exe and can be downloaded from the Microsoft Web site at the
following URL:
http://download.microsoft.com/download/win2000platform/clusrest/1.0/NT5/EN-
US/clusrest.exe
The Quorum restore procedure involves restoring the system state and cluster state
to the node followed by execution of the Clusrest.exe tool. Upon completion of the
restore, the node should rejoin the cluster and return to normal operation.
1. Restore the entire boot drive of the node if needed. Otherwise, restore the
2. Ensure that the cluster service is stopped on the other node.
50
Model 326 User's Reference
allow it to return to normal operation.
procedure.
Hubs. Give the 3534 Managed Hubs about three minutes to start up.
the storage controllers will have completed their startup routine. You can verify
this by making sure that for each drive in the storage controller and for each
drive in the storage unit, the status LED (on the top front of the drive) is solid
green (not blinking) for at least five seconds.
sure that all resources are in an online state or shortly return to that state.
that you shut down first in the powering off procedure. Each resource for which
that node is the preferred owner will fail back to that node and return to an
online state.
System State to the node.