Sun StorageTek Availability Suite 4.0 Remote Mirror Software Administration Guide

This chapter describes the Remote Mirror software's /usr/sbin/sndradm command and options.

The topics in this chapter include:

• Log File
• Syntax Summary
• From Which Host Are Commands Issued?
• Enabling and Disabling Volume Sets
• Synchronizing and Updating Volume Sets
• Reconfiguring or Modifying a Volume Set
• Displaying Volume Set Information
• Enabling or Disabling Autosynchronization
• Setting the Asynchronous Queue
• Tuning the Asynchronous Queue

---

Log File

The /var/adm/ds.log file contains operation log messages for the Remote Mirror commands. For example, these log messages show a volume set enabling and updating:

Oct 17 13:48:10 sndr: sndradm -e atm10 /dev/vx/rdsk/oracle816/oratest /dev/vx/rdsk/oracle816/oratest_bm atm20 /dev/vx/rdsk/oracle816/oratest /dev/vx/rdsk/oracle816/oratest_bm ip sync g oragroup

Successful

Oct 17 13:48:10 sv: enabled /dev/vx/rdsk/oracle816/oratest

Oct 17 13:48:10 sv: enabled /dev/vx/rdsk/oracle816/oratest_bm

Oct 18 10:23:54 sndr: sndradm -u atm10 /dev/vx/rdsk/oracle816/oratest /dev/vx/rdsk/oracle816/oratest_bm atm20 /dev/vx/rdsk/oracle816/oratest /dev/vx/rdsk/oracle816/oratest_bm ip sync g oragroup

Starting

---

Syntax Summary

Syntax

sndradm switches [options]

For example, to update the secondary volume from the primary volume:

sndradm -u [-g io-groupname] [-C tag] [-n] [[-f volset-file] |volset | set-name]

The /usr/sbin/sndradm command uses the command switches shown in TABLE 5-1. TABLE 5-2 shows the commands sorted in alphabetical order and TABLE 5-3 shows the options.

TABLE 5-4 describes the commands and synchronization situations performed from a primary host, secondary host, or both.

Command Switches

Command Options

---

From Which Host Are Commands Issued?

TABLE 5-4 describes the commands and synchronization situations which can be performed from a primary host, secondary host, or both.

---

Enabling and Disabling Volume Sets

Use the following commands to enable and disable volume sets.

Enabling Volume Sets

The sndradm -e and -E commands enable the Remote Mirror volume set (volset) and start logging changes to the volumes. Also use these commands to create one-to-many, many-to-one, and multihop volume sets. See One-to-Many, Many-to-One, and Multihop Volume Sets.

Caution - When creating volume sets, do not create secondary or bitmap volumes using partitions that include cylinder 0. Data loss might occur. See VTOC Information.

When you first enable a set, the software assigns a default name of shost:sdev, where shost is the secondary host name and sdev is the secondary volume partition name, separated by a colon (:). The volume set name is referred to in this document as set-name.

After executing these commands, you can use the shost:sdev name for a volume set each time you issue a command, instead of specifying the complete primary and secondary host, volume, and bitmap information for a volume set.

sndradm -e

This command sets the bitmap to indicate that a full synchronization between volumes is required. It also enables logging and adds the local volume set (including the local bitmap volume) to the storage volume (sv) driver control.

Syntax

sndradm -e [-n] [[-f volset-file] | volset]

sndradm -E

Caution - Before using sndradm -E, ensure that the volumes are already synchronized by means other than the Remote Mirror software (for example, restoring the volumes from tape or other media). Otherwise, the secondary or target volume will contain inconsistent data.

This command clears the bitmap logging to indicate that the specified volumes are fully synchronized and enables logging. It also adds the local volume set (including the local bitmap volume) to the storage volume (sv) driver control.

Syntax

sndradm -E [-n] [[-f volset-file] | volset]

After you enable a volume set, you can synchronize or update the volumes. See Starting a Full Synchronization and Starting an Update Resynchronization for more information.

Disabling Volume Sets

Use the sndradm -d command when the primary and secondary volumes no longer need to be associated with each other as software volume sets.

This command discards any active logging information in the bitmap volumes. See Enabling and Disabling Volume Sets to re-enable software replication and Starting a Full Synchronization to re-establish identical replicated sets.

sndradm -d

When you issue this command, the software terminates all replication services between the specified primary and secondary volumes and discontinues the relationship between these volume sets. It also discards any active bitmaps that track temporary differences between primary and secondary volumes.

Syntax

sndradm -d [-g io-groupname] [-C tag] [-n] [[-f volset-file] |volset | set-name]

Adding and Deleting Point-in-Time Copy Software Volumes

Use the sndradm -I command to add or delete Point-in-Time Copy software volumes for use with enabled software volume sets. You can use this command on the secondary and primary hosts to offer data protection during reverse and forward synchronization update operations. You also use this command to offer data protection during an automatic or manual resynchronization.

The sndradm -I a command creates configuration entries marked with the ndr_ii key. The ndr_ii entries contain an additional state field that the kernel uses to determine when point-in-time copies must be made. The kernel notifies the Remote Mirror software synchronization daemon whenever a synchronization is started and waits for sndrsyncd to perform any necessary copies before allowing the synchronization to proceed.

The daemon is also notified when any Remote Mirror software resynchronization starts or finishes executing. The daemon performs Point-in-Time Copy operations on the secondary or target host, if configured to do so.

Consider the following when configuring a volume set:

• The master volume is the Remote Mirror secondary volume.
• The shadow and bitmap volumes cannot be a volume currently in use by the Point-in-Time Copy software or Remote Mirror software. If the volumes are in use, data loss can occur as the result of this misconfiguration.
• Before the reverse synchronization occurs, the point-in time copy software enables the volume set as a dependent set. If the synchronization fails for any reason and the secondary volume data is lost or corrupted, you can restore the Remote Mirror secondary volume (the master volume in this set) by updating it with the shadow volume contents.

See also:

• Choosing Automatic or Manual Resynchronization
• Using the Remote Mirror Software With the Point-in-Time Copy Software
• Autosynchronization.

sndradm -I a

This command adds a Point-in-Time Copy volume group entry to the Sun StorageTek configuration.

Syntax

sndradm -I a master-vol shadow-vol bitmap-vol

where master-vol, shadow-vol, and bitmap-vol are the full path to the raw device nodes for configured Point-in-Time Copy volumes. In this case, the master-vol is the Remote Mirror secondary volume.

Caution - The shadow and bitmap volumes cannot be a volume currently in use by the point-in-time copy or Remote Mirror software. If the volumes are in use, data loss can occur as the result of this misconfiguration.

See the Sun StorageTek Availability Suite 4.0 Point-in-Time Copy Software Administration Guide for more information about these volume sets.

sndradm -I d

This command deletes a Point-in-Time Copy volume group entry from the Sun StorageTek configuration.

Syntax

sndradm -I d master-vol shadow-vol bitmap-vol

where master-vol, shadow-vol, and bitmap-vol are the full path to the raw device nodes for configured Point-in-Time Copy volumes.

Stopping Replication and Starting Logging

Use the sndradm -l command to start bitmap logging on enabled volume sets and I/O groups. Issue the sndradm -l command as follows:

• From the primary host if a synchronization is in progress.
  The primary host is in logging mode and the secondary hosts are in "need sync" state. This state is cleared by either performing a synchronization or by updating the secondary volumes.
• From the secondary host if the primary host or volume has failed.
  When you place the secondary host into logging mode, the primary host continues in replication mode.
• From either host if there is no replication occurring.

To resume Remote Mirror software operations after using the sndradm -l command, use the sndradm -u command to perform an update resynchronization or the sndradm -m command to perform a full resynchronization. Note also that, when issued from the secondary host, the sndradm -l command does not work on the secondary volume for any volume that is currently synchronizing.

sndradm -l

This command enables you to stop replication between primary and secondary volumes and to start logging data changes on these volumes. The Remote Mirror software continues logging even if the operations of a volume set or I/O group are interrupted.

If all volume sets in an I/O group are replicating (meaning that the secondary volumes contain a valid copy of the corresponding primary volumes), all sets in the group enter logging mode automatically whenever one volume set enters logging mode. This scheme ensures that the secondary volumes contain a valid copy.

Syntax

sndradm -l [-g io-groupname] [-C tag] [-n] [[-f volset-file] |volset | set-name]

---

Synchronizing and Updating Volume Sets

The following commands enable you to synchronize and update volume sets.

See also Restarting an Initial Synchronization and Performing a Reverse Copy or Update After a Primary Volume or Network Link Failure.

Restarting an Initial Synchronization

If, after enabling the volume sets, the initial synchronization operation using the sndradm -m or sndradm -u commands is interrupted, the software restarts the synchronization, according to the conditions described in the following table:

Performing a Reverse Copy or Update After a Primary Volume or Network Link Failure

As specified in Rehearsing Disaster Recovery, do the following tasks after the network link or disk failure is repaired:

1. Place the primary and secondary host volumes into logging mode if they are not already in logging mode.
See Stopping Replication and Starting Logging.

2. Perform the reverse copy or update.

Starting a Full Synchronization

Use the sndradm -m command when all of the following conditions exist:

• The primary volume has the desired contents.
• The contents of the primary and secondary volumes may be inconsistent.
• No logging information exists to incrementally resynchronize the volumes.
• You want to completely overwrite the contents of the secondary volume.

Use the sndradm -m -r command when all of the following conditions exist:

• The secondary volume has the desired contents.
• The contents of the primary and secondary volumes may be inconsistent.
• No logging information exists to incrementally resynchronize the volumes.
• You want to completely overwrite the contents of the primary volume.

You can use the primary volume during reverse synchronization
(sndradm -m -r). The primary volume shows a consistent volume image of the latest data as soon as the reverse synchronization starts. If your application had been writing to the secondary volume as part of a failure or disaster rehearsal, you can move the application back to the primary volume when the reverse synchronization starts. See also Rehearsing Disaster Recovery.

If a synchronization initiated by the sndradm -m or sndradm -m -r commands is interrupted, use the sndradm -u update command to finish executing the synchronization. You can use the sndradm -p, sndradm -P, and dsstat commands to view volume set status.

See also Restarting an Initial Synchronization and Performing a Reverse Copy or Update After a Primary Volume or Network Link Failure

sndradm -m

This command enables you to start a full copy operation from the primary volume to the secondary volume. It also enables replication concurrently from the primary volume to the secondary volume so that any new writes to the primary volume are also replicated to the secondary volume.

Syntax

sndradm -m [-g io-groupname] [-C tag] [-n] [[-f volset-file] |volset | set-name]

sndradm -m -r

This command enables you to start a full reverse copy operation from the secondary volume to the primary volume. It also enables replication from the primary volume to the secondary volume so that any new writes to the primary volume are also replicated to the secondary volume.

Syntax

sndradm -m -r [-g io-groupname] [-C tag] [-n] [[-f volset-file] |volset | set-name]

Starting an Update Resynchronization

Use the sndradm -u command when all of the following conditions exist:

• The primary volume has the desired contents.
• Replication was stopped by using the sndradm -l command or the network link was interrupted and bitmap logging was activated automatically.
• You want to overwrite those segments of the secondary volume that are different from the primary volume as defined in the bitmaps.
• You intend to start replication from the primary to the secondary volume.

Use the sndradm -u -r command when all of the following conditions exist:

• The secondary volume has the desired contents.
• Replication was stopped using the sndradm -l command or the network link was interrupted and logging was activated automatically.
• You want to overwrite those segments of the primary volume that are different from the secondary volume as defined in the bitmaps.
• You intend to start replication from the primary to the secondary volume.

If a synchronization initiated by the sndradm -m or sndradm -m -r commands is interrupted, use the sndradm -u update command to finish executing the synchronization. You can use the sndradm -p, sndradm -P, and dsstat commands to view volume set status.

See also Restarting an Initial Synchronization and Performing a Reverse Copy or Update After a Primary Volume or Network Link Failure.

sndradm -u

This command resynchronizes the secondary volume from the primary volume. It updates the secondary volume according to changes based on the bitmaps maintained while replication was stopped. It also enables concurrent replication between the primary and secondary volumes so that any new write operations to the primary volume are also replicated to the secondary volume.

Syntax

sndradm -u [-g io-groupname] [-C tag] [-n] [[-f volset-file] |volset | set-name]

sndradm -u -r

This command resynchronizes the primary volume from the secondary volume. It updates the primary volume according to changes based on the bitmaps maintained while replication was stopped. It also enables concurrent replication between the primary volume and secondary volumes so that any new write operations to the primary volume are also replicated to the secondary volume.

Syntax

sndradm -u -r [-g io-groupname] [-C tag] [-n] [[-f volset-file] |volset | set-name]

Waiting for Synchronization to Finish Executing

Use the sndradm -w command in one of the following situations:

• When you need to be sure that the copy or update operation is finished executing before you use a primary application (like a database application) or another software command on the specified volume set or I/O group
• Before disabling a volume set, so that any write or copy operations can finish executing
• As part of a script executing commands

sndradm -w

This command enables you to make the software wait until an in-progress copy or update synchronization finishes executing. It prevents another software command from executing on the specified volume set or I/O group.

Syntax

sndradm -w [-g io-groupname] [-C tag] [-n] [[-f volset-file] |volset | set-name]

---

Reconfiguring or Modifying a Volume Set

The following commands enable you to reconfigure or modify a volume set.

Changing the Bitmap Volume in a Volume Set

Use the sndradm -R b command when you need to assign a new bitmap to a volume set.

sndradm -R b

This command enables you to assign a new bitmap to an existing volume set. You can change the bitmaps on the primary or secondary host. This command copies any data from the old bitmap to the new bitmap.

Syntax

sndradm -R b {p|s} new-bitmap-name [-C tag] [-n] [[-f volset-file] |volset | set-name]

• To change a primary host bitmap, issue the command from the primary host.
• To change a secondary host bitmap, issue the command from the secondary host.
• Optionally, to ensure consistency for status reporting, issue the command from both hosts. For example, to change a secondary host bitmap, issue the command from the secondary host first and then issue it from the primary host.

Moving a Volume Set to a Different I/O Group

Use the sndradm -R g command to move volume sets to a different I/O group.

To move multiple sets to a single io-groupname, keep the io-groupname the same and specify different set-names on the command line.

See the following procedures:

• To Remove Volume Sets From an I/O Group
• To Move Volume Sets to Another I/O Group
• To Remove Volume Sets From an I/O Group and Change the Replication Mode

sndradm -R g

This command enables you to move volume sets to a different I/O group, thereby updating an existing I/O group. You must specify at least one set-name. Issue this command from the primary and secondary hosts.

Syntax

sndradm -R g io-groupname [-g io-groupname] [-C tag] [-n] [[-f volset-file] |volset | set-name]

1. Place the I/O group volume sets into logging mode:

# sndradm -g io-groupname -l

2. Remove volume sets:

• To remove a volume set from an I/O group, use quotation marks to specify a null group as follows:


# sndradm -R g "" {volset | set-name}

• To remove all volume sets from an I/O group:


# sndradm -R g "" -g io-groupname

• To remove selected sets from an I/O group, enter the volume set information in a file and use the -f volset-file option:


# sndradm -R g "" -g io-groupname -f volset-file

3. After you finish changing the sets, perform an update synchronization using the sndradm -u command.

1. Place the I/O group volume sets into logging mode:

# sndradm -g io-groupname -l

2. Move volume sets:

• To move all volume sets in one I/O group to another I/O group:


# sndradm -g io-groupname -R g new-io-groupname

• To move selected sets to an I/O group, enter the volume set information in a file and use the -f volset-file option with these commands.


# sndradm -R g new-io-groupname -f volset-file

3. After you finish changing the sets, perform an update synchronization using the sndradm -u command.

Changing the Replication Mode of a Volume Set

Use the sndradm -R m command to change the replication mode of a volume set.

sndradm -R m

This command enables you to reconfigure the replication or mirroring mode of a volume set.

Syntax

sndradm -R m {sync|async} [-C tag] [-n] [[-f volset-file] |volset | set-name]

Considerations

• Mixed mode I/O group - Creating a mixed I/O group, where mirroring modes are asynchronous for one set and synchronous for another set, is not allowed.
• Volume set requirements - All volume sets in the I/O group must have the same primary host, secondary host, and mirroring mode.

1. Place the I/O group volume sets into logging mode:

# sndradm -g io-groupname -l

2. Choose one of the following commands:

• To remove the volume set from its group:


# sndradm -R g ""{volset | set-name}

• To remove all volume sets from an I/O group:


# sndradm -R g "" -g io-groupname

3. Change the volume set replication mode of each volume set:

# sndradm -R m {sync|async} set-name

4. If desired, add the volume sets you have changed to a different or the previous I/O group:

# sndradm -R g io-groupname [[-f volset-file] | volset | set-name]

See also Moving a Volume Set to a Different I/O Group

5. After you finish changing the sets, perform an update synchronization using the sndradm -u command.

Updating the Volume Set Information

Use the sndradm -R -f command to change the current volume sets being operated on by the Remote Mirror software. This command is useful when you use a volume set file and have made changes to it.

The fields for the volume set file specified using the -f option are similar to those in the volset, described in TABLE 5-3:

phost pdev pbitmap shost sdev sbitmap ip {sync|async} [g io-groupname] [C tag]
[q qdev]

See Setting Up a Volume Set File and the rdc.cf man page for more information.

sndradm -R -f

This command enables you to update or reconfigure the current volume set information from a specified volume set file.

Syntax

sndradm -R -f volset-file [-g io-groupname] [-C tag] [-n]

Updating the Disk Cluster Tag Name

Use the sndradm -R C command in cases where the disk group name or cluster resource tag are not indicated by the volume path name. This command does not affect remote volumes and cannot be used in nonclustered environments.

You can also move all volume sets in one cluster tag to another cluster tag. Use the command as follows:

sndradm -C tag -R C new-tag

sndradm -R C

This command enables you to update or reconfigure the current disk group name or cluster resource tag of the local volume in a volume set. tag here is defined as the disk group name or cluster resource tag.

Syntax

sndradm -R C tag [-g io-groupname] [-C tag] [-n] [[-f volset-file] |volset | set-name]

---

Displaying Volume Set Information

Use the following commands to display information about volume sets.

Displaying Volume Set and I/O Group Status

When using the sndradm -p and sndradm -P commands to display status information, the reported status also includes information on the state of the set. Use the volset | set-name options to view the status for a single volume set. Use the -g io-groupname and -C tag options to filter the output so that it includes only those volumes that are part of io-groupname or tag.

sndradm -p

This command displays a brief version of the Remote Mirror software status for volume sets and I/O groups. The following is example output of this command on the primary host:

# sndradm -p

/dev/rdsk/c4t96d0s1 										-> fast7:/dev/rdsk/c2t0d0s1

/dev/rdsk/c4t97d0s1 										-> fast7:/dev/rdsk/c2t0d0s1

/dev/rdsk/c4t98d0s1 										-> fast7:/dev/rdsk/c2t0d0s1

Syntax

sndradm -p [volset | set-name] [-g io-groupname] [-C tag]

sndradm -P

This command displays a detailed version of the Remote Mirror software status for volume sets and I/O groups. The following is example output of this command on the primary host:

# sndradm -P

/dev/rdsk/c4t96d0s1 									-> 

fast7:/dev/rdsk/c2t0d0s1

autosync: off, max q writes: 4096, max q fbas: 16384, mode: sync

/dev/rdsk/c4t98d0s1 										-> fast7:/dev/rdsk/c2t0d0s1

autosync: off, max q writes: 4096, max q fbas: 16384, mode: sync

The reported status also includes information on the state of the set:

Syntax

sndradm -P [volset | set-name] [-g io-groupname] [-C tag]

Listing of Volume Sets and I/O Group Names

Use the sndradm -i command to display volume set and I/O group names. This command displays volume set information in a volume set file format. See Section , Setting Up a Volume Set File.

sndradm -i

This command enables you to display a list of the current volume set and I/O group names in use. The following is an example of the output.

Note that the output is formatted as it would be entered in a volset-file:
phost pdev pbitmap shost sdev sbitmap ip {sync|async}.

# sndradm -i

fast7 /dev/rdsk/c2t0d0s1 /dev/rdsk/c2t1d0s0 fast8 /dev/rdsk/c4t96d0s1 /dev/rdsk/c6t0d0s4 ip sync

fast7 /dev/rdsk/c2t0d0s1 /dev/rdsk/c2t1d0s3 fast8 /dev/rdsk/c4t97d0s1 /dev/rdsk/c6t0d0s6 ip sync

fast7 /dev/rdsk/c2t0d0s1 /dev/rdsk/c2t1d0s4 fast8 /dev/rdsk/c4t98d0s1 /dev/rdsk/c6t0d0s8 ip async

Syntax

sndradm -i [volset | set-name] [-g io-groupname] [-C tag]

Use the -g io-groupname and -C tag options to filter the output so that it includes only those volumes that are part of io-groupname or tag.

Displaying Link Status

Use the sndradm -H command to check the status of the link connecting primary and secondary hosts.

sndradm -H

This command displays the currently configured primary and secondary host machine names, volumes, and bitmap volumes for the selected sets. It also shows whether the link between machines is active or inactive. For example:

# sndradm -H atm-fred:/dev/vx/rdsk/freddg/sndr_vol01

Report SNDR link health? (Y/N) [N]: y

SNDR: atm-ethel /dev/rdsk/c3t9d0s3 /dev/rdsk/c6t0d0s4atm-fred /dev/vx/rdsk/freddg/sndr_vol01 /dev/rdsk/c6t0d0s6

Inactive

• Active indicates that replication or synchronization operations are occurring on the link used by the specified volume set or that all sets are in logging mode.
• Inactive indicates that the network link might be down.

Syntax

sndradm -H shost:sdev

where shost is the secondary host name and sdev is the secondary volume partition name, separated by a colon (:). When you first enable a set, the Remote Mirror software assigns a default volume set name of shost:sdev.

If you specify no arguments, the Remote Mirror software displays all configured volume sets. However, for best display results, specify a volume set using shost:sdev.

---

Enabling or Disabling Autosynchronization

Use the sndradm -a command to enable or disable autosynchronization. See Choosing Automatic or Manual Resynchronization for more information about autosynchronization.

sndradm -a

This command enables or disables the Remote Mirror software autosynchronization. If Remote Mirror software autosynchronization is enabled on the primary host, the synchronization daemon attempts to resynchronize volume sets if the system reboots or link failures occur.

The default state is disabled.

Syntax

sndradm -a {on|off} [-g io-groupname] [-C tag] [-n] [[-f volset-file] |volset | set-name]

Using Autosynchronization with Sun Cluster

Consider the following when using autosynchronization with Sun Cluster:

• If you want automatic resynchronization to occur in the event of a cluster failover, turn on the autosync feature. With this feature enabled, any cluster failover will automatically put the Remote Mirror volume sets back into replication mode after an update occurs.
  However, in a two-node cluster with a one-to-one configuration and with autosynchronization enabled, an interrupted reverse synchronization does not restart automatically. Issue the following command to restart the reverse synchronization process:

# sndradm -n -r -m

• If you want to manually force clusters to failover, make sure you allow all Remote Mirror components to fully recognize the condition, including the remote host that is not in the cluster. In practice, this means that you should not attempt an immediate update sync after the failover. You should wait at least thirty seconds after the completion of the scswitch command and before starting an update sync to allow time for Sun Cluster to complete its logical host interface reconfiguration.

---

Setting the Asynchronous Queue

These commands apply only to memory-based queues. They have no effect on disk queues.

Use the following commands to set the maximum number of writes or 512-byte blocks that can be queued in asynchronous mode.

See Tuning the Asynchronous Queue.

sndradm -W

This command enables you to set the maximum number of write operations to queue for asynchronous mode volume sets. The default value number is 4096. For example, set this value to 1000 to ensure that the secondary volume is never more than 1000 write operations behind the primary volume.

Syntax

sndradm -W value [-g io-groupname] [-C tag] [-n] [[-f volset-file] |volset | set-name]

sndradm -F

This command enables you to set the maximum number of 512-byte blocks in the asynchronous queue. The default number is 16384, allowing about 8 Mbytes of data per queue.

Syntax

sndradm -F value [-g io-groupname] [-C tag] [-n] [[-f volset-file] |volset | set-name]

---

Tuning the Asynchronous Queue

These commands apply only to memory-based queues. They have no effect on disk queues.

The queues containing the Remote Mirror software asynchronous write operations are stored in system kernel memory. You can tune or adjust these queues so that they are correctly sized for your Remote Mirror software use and volume data set size. Left untuned, the queues can appear to slow performance and consume more system memory than is desirable.

Consider the following steps when tuning asynchronous queues:

1. The number of volume sets and whether a group contains a large number of volume sets in asynchronous mode (such a group shares a single queue and might affect performance).

2. Physical system memory.

3. Network latency.

The number of volume sets is most important. Large queue settings matched with a small number of volume sets ensures better performance than large queue settings with a large number of volume sets. However, try not to size queues so large that you exhaust kernel memory and degrade your system's performance.

This section includes the following topics:

• Asynchronous Mode and the Queue
• Sample Output For a Correctly-Sized Queue
• Sample Output for an Incorrectly-Sized Queue

Asynchronous Mode and the Queue

The Remote Mirror software's asynchronous mode helps when your network has sufficient bandwidth to handle many write transactions or when your replicating application (such as a database) performs burst-writes at a rate higher than the network limits allow.

Generally, you use asynchronous mode for a volume set when the I/O write rate is less than the throughput of the network you are using for replication. This mode provides low latency rates to the software, resulting in faster response at the host (see Asynchronous Replication Mode).

As the number of write operations in the queue increases, the transactions can exceed the ability of the queue and network to process them (depending on network bandwidth and latency). The Remote Mirror software appears slow and reports less than optimal write transaction performance. Unless you tune (or size) the queue correctly, asynchronous mode performance might appear to be the same as or worse than synchronous mode, depending on the volume data set size.

Consider changing the software's default queue size if:

• The amount of system memory cannot support the number of asynchronous mode volume sets in the queue.
• The replicating application writes more data than the queue can store, affecting performance.
• You do not want the application's write operations to become "out of sync" with the primary host, where pending write operations try to queue faster than the queue or network can handle them.
• You do not care how "out of sync" the secondary host is from the primary host, you care only about response time on the primary host.

You can change the asynchronous queue size by using the sndradm -W and
sndradm -F commands. The Remote Mirror software sets the following default sizes for the queue:

1. Type the following to display the current queue size:

# sndradm -P

/dev/vx/rdsk/rootdg/ds4-clone   ->      nws:/dev/vx/rdsk/rootdg/ds4-sndr-s

autosync: on, max q writes: 4096, max q fbas: 16384, mode: async

The kstat command also displays queue information:

2. Perform any of the following kstat(1M) commands to show more information.

• To show all sets, type:


# kstat sndr:::maxqfbas

• To show the first instance (0) in the queue, type:


# kstat sndr:0::maxqfbas

• To show more information, type:


# kstat sndr:::maxqitems # kstat sndr:::async_throttle_delay

Sample Output For a Correctly-Sized Queue

The following kstat(1M) kernel statistics output shows information about the asynchronous queue. In this example, the queue is sized correctly.

# kstat sndr:0:setinfo

module: sndr                            instance: 0

name:   setinfo                         class:    storedge

        async_block_hwm                 878

        async_item_hwm                  483

        async_throttle_delay            0

        maxqfbas                        16384

        maxqitems                       4096

        primary_host                    regina

        primary_vol                     rootdg/ds4-clone

This example shows only a portion of the command output needed for this section; the kstat command actually displays more information. The following table explains the settings and statistics shown in the example.

Sample Output for an Incorrectly-Sized Queue

The following kstat(1M) kernel statistics output shows information about the asynchronous queue, which is incorrectly sized.

# kstat sndr:4:setinfo

module: sndr                            instance: 4     

name:   	setinfo                         				 class:    storedge

        async_block_hwm                 16380

        async_item_hwm                  2045

        async_throttle_delay            16497

        maxqfbas                        16384

        maxqitems                       4096

        primary_host                    andrea

        primary_vol                     rootdg/ds-forall

This example shows only a portion of the command output needed for this section; the kstat command actually displays more information.

This example shows the default queue settings, but the application is writing more data than the queue can handle. The async_block_hwm value of 16380 indicates that the application is approaching the limit for 512-byte blocks. The next few I/O operations might not be placed into the queue.

The value of async_throttle_delay indicates that the application has been delayed 16497 x 2 milliseconds. In this example, if the system has sufficient memory, increase the size of maxqfbas using the sndradm -F command.

---

Managing Disk Queues

You can have memory-based or disk-based queues. Disk-based queues have some advantages over memory-based queues. See Disk-Based Asynchronous Queues for further information. You can configure a disk queue when you first enable a set or you can use these commands later.

You can add, remove, and replace disk queues for sets and groups of sets with the disk queue CLI commands. You can also set each queue for blocking or non-blocking mode. See Blocking Mode and Non-blocking Mode for more information.

sndradm -q

This command enables you to add, remove, or replace a disk queue to a set or to a group.

Syntax

sndradm -g io-groupname -q a vol

sndradm -g io-groupname -q d

sndradm -g io-groupname -q r newvol

sndradm -q a vol shost:sdev

sndradm -q d shost:sdev

sndradm -q a newvol shost:sdev

For example, this is how to add a disk queue to a Remote Mirror set:

# sndradm -q a /dev/md/rdsk/d2 ns-81:/dev/md/rdsk/d221

sndradm -D

This command enables you to set the disk queue blocking mode for a set or group.

sndradm -D {block | noblock} set

For example, to change from blocking mode to nonblocking mode, type:

# sndradm -D noblock ns-81:/dev/md/rdsk/d221

Change Remote Mirror tunable? (Y/N) [N]: y

---

Setting the Number of Asynchronous Threads

You can change the number of asynchronous threads from the default value of 2. You can set the number of threads to a higher number limited only by practical considerations.

sndradm -A

This command enables you to set the number of asynchronous threads that drain the asynchronous queues.

sndradm -A asyncthreads set

See Multiple Asynchronous Flusher Threads for more details.

^{1 (TableFootnote) async_block_hwm and async_item_hwm show the maximum number that has been put into the queue since replication started. They do not show the current number in the queue.}

Sun StorageTek Availability Suite 4.0 Remote Mirror Software Administration Guide

819-6148-10

Copyright © 2006, Sun Microsystems, Inc. All Rights Reserved.