|
This chapter contains the following sections:
To run NTDS Administrator:
- Log on to the Windows NT computer as an Administrator.
- From the Windows Start Bar click Programs,
then DIGITAL OpenVMS Disk Services, then NTDS
Administrator.
You use NTDS Administrator to:
For more information, see NTDS Administrator's online help.
On the... | Use... |
OpenVMS Alpha node | The following commands to
find out which disk services are currently available and which
Windows NT computers are connected to them:
| Windows NT computer | NTDS
Administrator and the Windows NT Event Viewer to find out which disk
services the computer is currently connected to. For information on
using the Windows NT Event Viewer, see Section 4.1. |
The following examples show how to use the NTDS SHOW SERVER and NTDS SHOW SERVICES commands:
- Use NTDS SHOW
SERVER to find out the cluster name that was specified on the
NTDS START SERVER command.
It also shows summary information about disk services available
on the local node.
$ NTDS SHOW SERVER
DIGITAL OpenVMS Disk Services for Windows NT
Cluster name: Media
PID of server process: 222000D2
Active disk services: 2
Connections: 1
Standby disk services: 1
- Use NTDS SHOW
SERVICES to find out more information about disk services on
the local node. If you omit the parameter, it shows information
about all the services that are currently available on the local
node.
If a service is active and a Windows NT computer is connected to
it, the Computer connected field contains the name of
the computer, otherwise it contains (none).
$ NTDS SHOW SERVICES
DIGITAL OpenVMS Disk Services for Windows NT
Disk services available on this node:
Service name: Music
Status: Active
Computer connected: (none)
Container file: DISK$MEDIA:[NTDS$CONT]MUSIC.NTDS;1
Service name: Sports
Status: Standby
Computer connected: (none)
Container file: DISK$MEDIA:[NTDS$CONT]SPORTS.NTDS;1
Service name: Movies
Status: Active
Computer connected: NTSERVER5
Container file: DISK$MEDIA:[NTDS$CONT]MOVIES.NTDS;1
- For an active service, use NTDS SHOW SERVICES with /FULL to get statistics on
reads and writes to the container file since the service became
active on the local node:
$ NTDS SHOW SERVICES "Movies" /FULL
DIGITAL OpenVMS Disk Services for Windows NT
Disk services available on this node:
Service name: Movies
Status: Active
Computer connected: NTSERVER5
Container file: DISK$MEDIA:[NTDS$CONT]MOVIES.NTDS;1
Read operations: 14350
Blocks read: 651367
Write operations: 43018
Blocks written: 4200499
- This example shows how to find out whether the service
Movies, which is active on the local node, has a standby
elsewhere in the OpenVMS Cluster:
$ RUN SYS$SYSTEM:SYSMAN
SYSMAN> SET ENVIRONMENT /CLUSTER
SYSMAN> SET PROFILE /PRIVILEGES=SYSPRV
SYSMAN> DO NTDS SHOW SERVICES "Movies"
SYSMAN> EXIT
When you shut down a Windows NT computer, it automatically flushes
to disk all the unwritten data in its local cache; this includes
data for its virtual disks as well as for its physically attached
local disks. The computer does not shut down until all of the data
is safely on disk.
If the computer has lost a connection to a disk service, for
example, because the service was stopped on the OpenVMS Cluster or the network link failed, the writes
are stalled and the shutdown pauses until the computer manages to
reconnect to the service and write the data to disk.
- Caution
- If the shutdown stalls,
don't turn the computer off or you will lose data!
Find out why the computer lost the connection to the disk
service and fix the problem. For example, check whether the
service is available on the OpenVMS Cluster, using the NTDS SHOW SERVICES command.
If necessary, use the NTDS START
SERVICE command to restart the service. The Windows NT computer
then automatically reconnects to the service, writes the data to
disk, and finishes shutting itself down.
If you turn the computer off before it manages to reconnect to the
disk service and write the data to disk, you may lose data stored
on the disk service's virtual disk. You could also lose data stored
on the computer's physically attached local disks if writes to those
disks have not yet been flushed from the Windows NT cache.
To make sure that your computer doesn't hang during shutdown, before
you shut it down, check that it hasn't lost the connection to any
disk service. You can do this by trying to access each of its
virtual disks, for example, by double clicking the disk icons in
My Computer or Windows NT Explorer.
Service failover is automatic in an OpenVMS Cluster. When the node
currently providing a service fails, the service automatically fails
over to a node that is configured to provide the standby service.
There may be times when you want to force a service to fail over to
a standby, for example, to balance the load across the nodes in your
OpenVMS Cluster.
To do this, stop the service on each of the nodes on which it is
currently available except for the target node, and then
restart it on those nodes.
For example, the Sports service is currently active on node GREEN
and has standbys on nodes BLUE and ORANGE. GREEN is heavily loaded,
so you decide to fail the service over to ORANGE:
$ RUN SYS$SYSTEM:SYSMAN
SYSMAN> SET ENVIRONMENT /NODE=(GREEN,BLUE)
SYSMAN> SET PROFILE /PRIVILEGES=(SYSNAM,SYSPRV)
SYSMAN> DO NTDS STOP SERVICE "Sports" /OVERRIDE 1
SYSMAN> DO NTDS START SERVICE "Sports" - 2
_SYSMAN> /CONTAINER=DISK$MEDIA:[NTDS$CONT]SPORTS.NTDS
SYSMAN> EXIT
- Force the service to fail over to the target node,
ORANGE, by using NTDS STOP
SERVICE with the /OVERRIDE qualifier to stop the service on
the other two nodes, GREEN and BLUE.
- Restart the service on the other nodes, GREEN and
BLUE, using the NTDS START
SERVICE command. This starts standby services on those
nodes.
To move a connection from one Windows NT computer to another:
- Disconnect the current Windows NT computer from the
service, using NTDS Administrator.
- On the target Windows NT computer, use NTDS Administrator
to add the service then connect to it. You may also need to
configure the communication settings (see the online help for
more information).
The drive letters assigned to the partitions on the virtual disk
may be different from those on the original Windows NT computer.
To change the drive letters, use Windows NT Disk Administrator.
If you want to share the virtual disk, do so in the normal way,
for example, using Windows NT Explorer.
If the Windows NT computer currently connected to a service fails,
it is only safe to connect another Windows NT computer to the
service if you are sure that either:
- The original computer will not restart (for example,
because it was destroyed in a fire), or
- The original computer will not be able to connect to the
OpenVMS Cluster when it restarts (for example, because you have
disconnected it from the network)
- Caution
- You can corrupt your data
if you connect another Windows NT computer to the service and the
original Windows NT computer restarts and makes contact with the
OpenVMS Cluster that is providing the service.
When the original Windows NT computer restarts it automatically
tries to reconnect to the service. If the service happens to fail
over from one node in the OpenVMS Cluster to another while it
is restarting, it might succeed in reconnecting to the service,
stealing the connection back from the other Windows NT computer. If
this happens, your data could be corrupted.
You can use Windows NT backup tools to back up and restore the
data stored on a virtual disk created by DIGITAL OpenVMS Disk Services for Windows NT.
You can use them to back up or restore either:
- Selected files on the virtual disk, or
- The whole virtual disk.
You can use OpenVMS backup tools to back up and restore a virtual
disk created by DIGITAL OpenVMS Disk
Services for Windows NT.
You back up and restore the OpenVMS container file that is used to
store the data for the virtual disk.
From OpenVMS, you cannot back up selected Windows NT files on the
virtual disk. You must back up the whole of the virtual disk.
If the container file is stored in a Files-11 volume, you must make
sure that the back up captures a consistent snapshot of the entire
container file at a single instant in time. You must make sure that
the container file you are backing up is not being modified during
the backup, otherwise your backup will be inconsistent.
To do this, during the backup, take the disk service off line by
either:
- Disconnecting the Windows NT computer from the service, or
- Stopping the disk service on the OpenVMS Cluster
Use NTDS Administrator to disconnect from the service before you
start the backup, and to reconnect the service when the backup is
finished.
This is the recommended method. It produces a backup that contains
all the updates written to the Windows NT computer's cache.
Disconnecting the Windows NT computer from the service flushes the
data from its cache.
When you back up the container file, use the /IGNORE=INTERLOCK
qualifier. Because you have not stopped the service, the server
software has the container file open so the backup will fail if you
do not use this qualifier.
Disconnecting from a disk service automatically closes all open
files in NTFS partitions on its virtual disk.
Note that you cannot disconnect from a disk service if its virtual
disk has a non-NTFS partition (for example, a FAT partition) and
there are any open files in the partition. In this case, do one of
the following:
- Close all the files in the partition, then disconnect
from the service and back up the container file.
- Use the other method to take the disk service off
line during the back up; stop the disk service on the OpenVMS
Cluster.
If the virtual disk has multiple partitions, the failed attempt
to disconnect from the service might succeed in taking the other
partitions off line and making them inaccessible. To get them back
on line again, use the Windows NT Disk Administrator to reassign
drive letters to them.
Use the NTDS STOP SERVICE
command with the /OVERRIDE qualifier to stop the service before you
start the backup, and the NTDS
START SERVICE command to restart it when the backup is finished.
Make sure that you use NTDS STOP
SERVICE with /OVERRIDE on every node that can provide
the service. Stop the service on the node currently providing the
service and on all of its standbys.
This method does not capture unwritten data in the Windows NT
computer's cache. It produces a backup that captures the same data
as you would find on disk if the Windows NT computer crashed.
This method is not recommended if users are writing data to the
virtual disk during the backup.
- While the service is stopped, all reads and writes to the
container file are stalled.
- If the Windows NT computer's cache becomes full because
it cannot issue writes to the container file, it may hang until
the service is restarted on the OpenVMS Cluster.
If the container file is stored in a Spiralog volume, you do
not have to take the disk service off line while you back up the
container file. You can issue the SPIRALOG SAVE command at any time;
it always captures a consistent snapshot of the entire container
file at an instant in time, even when users are updating the Windows
NT files stored in it.
The SPIRALOG SAVE command does not flush the Windows NT computer's
cache, so it captures the same data as you would find on disk if the
Windows NT computer crashed.
If you want to capture recent updates written to the Windows NT
cache, use NTDS Administrator to disconnect the Windows NT computer
from the service before you issue the SPIRALOG SAVE command.
You can reconnect the computer to the service after a few seconds
(as soon as the SPIRALOG SAVE /LOG command issues the FLUSHDONE
message).
When you do an incremental backup, you back up the changes to
the container file, not the whole container file.
Follow these steps to restore the whole virtual disk from OpenVMS:
- Restore the container file.
You can restore the container file to the same or a different
OpenVMS disk.
If the original container file still exists, we recommend that
you give the restored container file a different name. For
example, if you are restoring SPORTS.NTDS, call the restored
container file SPORTS_RESTORE.NTDS.
- Start a disk service for the restored container
file.
Use the NTDS START
SERVICE command to start a disk service for the restored
container file. For example:
$ SET PROCESS /PRIVILEGE=(SYSNAM,SYSPRV)
$ NTDS START SERVICE "Sports_restore" -
_$ /CONTAINER=DISK$MEDIA:[NTDS$CONT]SPORTS_RESTORE.NTDS
- Connect the Windows NT computer to the disk service
for the restored container file.
On the Windows NT computer, use NTDS Administrator to add then
connect to the service.
Note that the virtual disks associated with the restored and
the original disk services have the same disk signature. This
means that if you connect the same Windows NT computer to both
the original and the restored disk service, when you run Disk
Administrator you'll get a popup window asking if you would like
to change the disk signature of the restored disk service. Click
Yes to let Disk Administrator make this change.
Follow these steps to restore selected Windows NT files from
OpenVMS:
- Restore the container file.
You can restore the container file to the same or a different
OpenVMS disk.
Give the restored container file a different name from
the original container file. For example, if you are
restoring SPORTS.NTDS, call the restored container file
SPORTS_RESTORE.NTDS.
- Start a disk service for the restored container
file.
Use the NTDS START
SERVICE command to start a disk service for the restored
container file. For example:
$ SET PROCESS /PRIVILEGE=(SYSNAM,SYSPRV)
$ NTDS START SERVICE "Sports_restore" -
_$ /CONTAINER=DISK$MEDIA:[NTDS$CONT]SPORTS_RESTORE.NTDS
- Connect the Windows NT computer to the disk service
for the restored container file.
On the Windows NT computer, use NTDS Administrator to add then
connect to the service.
Note that the virtual disks associated with the restored and
the original disk services have the same disk signature. This
means that if you connect the same Windows NT computer to both
the original and the restored disk service, when you run Disk
Administrator you'll get a popup window asking if you would like
to change the disk signature of the restored disk service. Click
Yes to let Disk Administrator make this change.
- Copy the files you want to restore from the virtual
disk for the restored container file to the target locations.
- Disconnect the Windows NT computer from the disk
service for the restored container file.
On the Windows NT computer, use NTDS Administrator to disconnect
from the service.
- Stop the disk service for the restored container
file.
Use the NTDS STOP SERVICE
command to stop the disk service for the restored container file.
For example:
$ NTDS STOP SERVICE "Sports_restore"
- Delete the restored container file.
For example:
$ DELETE DISK$MEDIA:[NTDS$CONT]SPORTS_RESTORE.NTDS;1
This section describes the things you can do to optimize the
performance of your disk services.
- Use the following command to optimize the performance of
DIGITAL TCP/IP Services for OpenVMS:
$ UCX SET CONFIGURATION PROTOCOL TCP /NODELAY_ACK
Do this on all the OpenVMS Alpha nodes that provide disk
services.
- If the Windows NT computer has two or more network cards
configured for TCP/IP, make sure that it tries to connect to the
fastest one first whenever it connects to disk services.
To do this, make sure that the IP subnet associated with the
fastest card is at the top of the subnet list displayed by
the NTDS Administrator's Communication Settings screen (Figure 2-2). If necessary, use the
Up and Down buttons to change the order of the
entries in the list.
For example, if the computer has an FDDI card and an Ethernet
card, make sure the IP subnet associated with the FDDI card is
listed first.
- Make sure that the container file is on a disk that is
directly attached to each of the OpenVMS nodes that can provide
its disk service. You lose performance if the server has to use
MSCP to access the container file via another node.
- Give the Windows NT computer more memory.
The performance of a disk service is sensitive to the amount of
memory available to the computer connected to the service and the
workstations that access shares on it (the computers in tiers 2
and 3 in Figure 1-1). More memory means a
bigger cache, which means fewer I/Os are issued to disk.
On the OpenVMS Cluster, use NTDS SHOW SERVICES with
/FULL to check the ratio of reads and writes to disk. If it shows
a high proportion of reads, it could mean that the Windows NT
computer's cache is not big enough.
- Upgrade the network connection between the OpenVMS
Cluster and the Windows NT computer, for example from Ethernet
to FDDI. The heavier the I/O load to disk, the more significant
the speed of the network connection on overall performance.
Previous Topic | Next Topic | Contents | Index | Version Information
Legal
|