IBM totalstorage 326, Справочник пользователя

Страница: 82 / 240

own the volume, you cannot select the volume when creating a new schedule
through the New Persistent Image Schedule panel (under Schedules).
3. Use the Schedules panel on the other engine to create the same schedule
that you created on the original engine, with all of the same parameters (start
time, frequency, number to keep, and so on).
4. Use the Cluster Administrator to move the disk group that contains the
volume back to the original engine.
v
Volume restore of the system volume (C: drive) is not supported. If you attempt
to restore a persistent image containing the system volume, the restore operation
will not take place.
v
Volume restore of a data volume might require a reboot of the node. You will be
notified by the Restore Persistent Images panel whether a reboot is required
after a restore operation is initiated.
v
When you restart the Model 326 (“restart” in this case means that with both
nodes down, the node that was shut down last is restarted first so that it initially
owns all of the shared data volumes), Persistent Storage Manager (PSM) takes
two actions:
1. Loading
2. Mapping
During loading, PSM loads existing persistent images from the cache files on
each of the volumes. The loading time depends on the amount of cache data
there is to read. Cached data is used by PSM to maintain the persistent images,
and the more cache data there is, the longer it takes to load the persistent
images, and thus the longer it might take the Model 326 to become fully
operational after a restart.
During mapping, PSM makes the loaded persistent images accessible through
the file system by mounting each of them as a virtual volume underneath the
persistent images directory on the real volume for which the persistent image
was created. Mapping takes place five minutes after the real volume has been
mounted. The mapping time varies with the number of persistent images, as well
as the size of the volume.
As an example, suppose that on your Model 326, you defined a 1 TB volume
with 50 percent of the volume allocated to the cache (500 GB cache), and that
you had 20 persistent images on the volume, using 100 GB (20 percent) of the
cache (based on the write activity to the volume since the first persistent image
was created). You would observe an increase in the Model 326 startup time of
roughly 3 minutes, 20 seconds over what it would be without any persistent
images on the volume. Then, once the Model 326 has become fully operational,
all 20 persistent images would become accessible within another 18 minutes
(including the five minutes that PSM waits after the volume comes up to begin
the mapping).
When a volume is failed over between nodes, then PSM must perform persistent
image loading and mapping on the node to which the volume is moving, just as it
does when the “first node” is restarted.
In the failover scenario, loading must take place before the volume can be
brought online on the node (when the clustered disk resource is shown as being
Online in Cluster Administrator). Then, as in the restart case, mapping begins five
minutes after the volume comes online.
70
Model 326 User’s Reference