Table of Contents
Advertisement
Microsoft Cluster Server, which controls the disk resource failover, waits a certain
period, called the pending timeout, for the disk to come online. (During the
loading phase, the disk resource is shown as being in Online Pending state.)
With a default value of 180 seconds (3 minutes) for the pending timeout, this
time interval might be exceeded because of the time it takes to load the
persistent images on the volume. If this occurs, the delay might cause Cluster
Server to mark the disk as Failed and to not be available to either Model 326
node. Other dependent resources (IP addresses, network names, file shares, and
so on) might also fail.
For this reason, it is recommended that you increase the pending timeout value
for all clustered resources to 1200 seconds (20 minutes). To do this, open
Cluster Administrator, select Resources from the left pane to display all clustered
resources in the right pane, and then for each resource listed in the right pane:
1. Right-click the resource name and select Properties.
2. Select the Advanced tab.
3. Change the Pending timeout value to 1200 (seconds).
4. Click Apply, then click OK.
v PSM imposes a limit of 1 terabyte (TB) of cached data, across all volumes on the
Model 326. For this reason, you should ensure that the total configured size of all
cache files on the Model 326 is not greater than 1 TB.
You can do this by accessing Persistent Storage Manager, then going to the
Volume Settings page, and making sure that the total of all values in the Cache
Size column is 1 TB or less. (You can access Persistent Storage Manager
through the Persistent Storage Manager link on the IBM NAS Admin console on
the Model 326 desktop, or by starting the Windows 2000 for Network Attached
Storage user interface and then selecting Disks, then Persistent Storage
Manager.)
If the total is greater than 1 TB, you should reduce the size of the cache on one
or more of the volumes by selecting the volume from the list, then clicking
Configure, then selecting a smaller value from the "Cache size" drop-down list
and clicking OK.
Note: You cannot change the size of the cache on a volume that has persistent
images. You must delete all persistent images on the volume before
changing the cache size. You should try to reduce the cache size on a
volume that has no persistent images, if possible, before deleting any
persistent images.
If more than 1 TB of cache is configured on the Model 326, the following can
occur (note that a volume for which a persistent image has never been created is
considered to have a cache size of zero, regardless of how large its cache is
configured to be):
– When the Model 326 is restarted, PSM prevents a volume from being
mounted on the file system (prevents it from being accessible) if that volume's
PSM cache would increase the total size of all cache files (on all volumes
mounted to that point) above 1 TB, and an error message is written to the
system event log. The event source is psman5, and the text of the error
message is:
There is insufficient memory available.
– When a volume is failed over between nodes, then PSM running on the "new"
node will behave as it would if the volume were being mounted during a
restart: if that volume's PSM cache would increase the total size of all cache
files on that node above 1 TB, then PSM blocks the mount and writes the
Chapter 6. Managing and protecting the network and storage
71
Advertisement
Table of Contents
Troubleshooting
