Managing Copy and Merge Operations (Integrity servers and Alpha)
Copies and merges performed by the volume shadowing
software are regulated automatically by the locking software and by
the setting of SHADOW_MAX_COPIES. The SET SHADOW command, introduced in OpenVMS Alpha Version 7.3–2 and extended
in OpenVMS Version 8.2, provides better control over the order of
copies and merges and allows you to specify the systems on which the
copy operations must take place.
All SET SHADOW qualifiers pertain to shadow sets
(DSAn:), and some can also be applied to individual shadow set members (ddcu:), as described
in Table 4-3. For most qualifiers
that take a shadow set as a parameter, the /ALL qualifier can be used
in place of the shadow set name to indicate that the requested action
applies to all shadow sets on the system.
The qualifiers remain in effect until the device
(shadow set or shadow set member) is dismounted. If the device is
remounted (in the case of a shadow set member, returned to the shadow
set from which it was dismounted), the qualifier must be specified
again. The SET SHADOW command requires the
SYSPRV privilege.
NOTE: The qualifiers /DELETE, /DISABLE, /ENABLE, /NAME, and /POLICY
are used only to manage host-based minimerge (HBMM) operations and
do not apply to other operations. If you specify any other (non-HBMM)
qualifiers in a command that includes HBMM qualifiers, the command
fails. For more information about HBMM, see Chapter 8.
The following example shows how to specify qualifiers for
a shadow set:
$ SET SHADOW DSAn:/qualifier/qualifier
Table 4-3 SET SHADOW Command Qualifiers
Qualifier
Function
/ABORT_VIRTUAL_
UNIT {DSAn:|/ALL}
Aborts
mount verification on the specified shadow set or on all shadow sets
in mount verification on the system. Use this qualifier when you
know that the unit cannot be recovered. When you use this qualifier,
the shadow set must be in mount verification. The shadow set aborts
mount verification immediately on the system from which the command
is issued. If the shadow set is not in mount verification, this command
returns the error %SYSTEM-E-UNSUPPORTED, unsupported operation or
function.
After this command completes, the shadow set must still be
dismounted. Use the following command to dismount the shadow set:
$ DISMOUNT/ABORT/OVERRIDE=CHECKS DSAn
/ALL
Causes the command to operate on all shadow
sets that are mounted on the system from which the command is issued.
/ALL can be used
instead of DSAn:in most commands that take a shadow set device specification as a
parameter, except with the /DEMAND_MERGE, /DELETE, /EVALUATE=RESOURCES,
and /POLICY qualifiers or with any qualifier that operates only on
individual shadow set members (for example, /MEMBER_TIMEOUT and /FORCE_REMOVAL.
/CONFIRM/NOCONFIRM
(default)
Specifies whether
a query is made before each merge operation to confirm that the operation
must be performed on the designated shadow set.This qualifier can be used
only in conjunction with the /DEMAND_MERGE qualifier.The following
responses are valid in response to the query:
Affirmative: YES, TRUE, or 1.
Negative: NO, FALSE, 0 (zero), or pressing the Return
key.
End the process: QUIT or Ctrl/Z.
When you enter ALL, the command continues to process,
but no further prompts are given.
You can enter word responses in uppercase
or lowercase letters, and words can be abbreviated to one or more
letters. If you enter an illegal response, DCL redisplays the prompt.
For examples, see the SET SHADOW examples in the HP OpenVMS DCL Dictionary.
/COPY_SOURCE {ddcu:|DSAn:|/ALL}
Specifies which source member of a shadow
set to use as the source for read data during full copy operations
when a third member is added to a shadow set that contains two full
members. This qualifier affects only those copy operations that do
not use disk copy data (DCD) commands. The source specified by this
qualifier persists until the shadow set is dismounted.Some storage controllers,
such as the HSG80, have a read-ahead cache, which significantly improves
a device’s read performance. Copy operations normally alternate
reads between the two source members, which effectively nullifies
the benefits of the read-ahead cache. This qualifier lets you force
all reads from a single, specified source member for the duration
of a copy operation.In addition to improving copy performance, /COPY_SOURCE
can be used to prevent read operations from a specific shadow set
member that is considered unreliable. By specifying only the reliable
shadow set member, the copy operations can continue to completion.
The unreliable shadow set member can be removed once the copy operation
completes successfully. If a shadow set (DSAn) is specified, all reads
for full copy operations are performed from the device that is the
current "master" member, regardless of physical location of the disk.If
a shadow set member (ddcu:) is specified, that member is used as
the read source for all copy operations. This setting allows you to
choose any source member. For example, you can choose a source member
that is at the same site as the member being added, rather than using
a master member that is not at the same site. If /ALL is specified,
all reads for full copy operations on all currently mounted virtual
units are performed from the master member.
/DELETE {DSAn:|/NAME}
Used only in conjunction with /POLICY=HBMM,
/DELETE removes a host-based minimerge (HBMM) policy from a specified
shadow set, or deletes an HBMM named policy from the entire cluster.
For example, the following command removes the policy that is currently
associated with shadow set DSA1:
$ SET SHADOW /DELETE DSA1 /POLICY=HBMM
In contrast, the following command removes COMPANY_POLICY
from the cluster:
$ SET SHADOW /DELETE /NAME=COMPANY_POLICY /POLICY=HBMM
You cannot delete the NODEFAULT policy, nor can you
specify /ALL with /DELETE.
/DEMAND_MERGE
Initiates a merge operation on the
specified shadow set. This qualifier is useful if the shadow set was
created with the INITIALIZE/SHADOW command without the use of the
/ERASE qualifier. For more information about using /DEMAND_MERGE, see “Using /DEMAND_MERGE to Start a Merge Operation”.
You
cannot specify /ALL with /DEMAND_MERGE.
An OPCOM message
is displayed for each shadow set indicating that a demand merge has
been invoked and recording the process ID (PID) of the process that
executed the command. For example:
Demand Merge requested for _DSA721:,
PID: 2760009A
/DISABLE=HBMM
{DSAn:| /ALL}
Disables host-based minimerge (HBMM) on the specified shadow
set or clusterwide on all shadow sets.
HBMM is the only supported value for /DISABLE,
and it must be included.
/DISABLE=SPLIT_READ_LBNS
Disables the split behavior
of LBNs and as a result the reads are alternated between the source
shadow set members having the same read_cost and device queue length.
/ENABLE=HBMM
Enables host-based minimerge (HBMM)
on the specified shadow set or across the entire cluster if an applicable
HBMM policy exists.
HBMM is the only supported value for /DISABLE,
and it must be included.
/ENABLE=SPLIT_READ_LBNS
Logically divides the shadow
set members having the same read cost into equal groups of LBNs. When
a virtual unit performs a read, it does so by reading from the corresponding
LBN group. This results in the maximum usage of the controller read-ahead
cache.
/EVALUATE=
RESOURCES
Forces the
system to evaluate whether it must act on most shadow copy and merge
operations currently being managed on the system. It cancels
most operations and then, based on the value of system parameter SHADOW_MAX_COPY
and the copy/merge priority of each shadow set, it evaluates the order
in which the pending copies and merges should be restarted.
RESOURCES is the only supported value for /EVALUATE, and it
must be included.
EVALUATE does not apply to MSCP-based
minimerge operations. MSCP-based minimerge operations are not subject
to cancellation and restart by /EVALUATE.
This command
is intended to be used after changing the value of the dynamic system
parameter SHADOW_MAX_COPY or after issuing a SET SHADOW /PRIORITY=n
command for a shadow set. After a suitable delay, all available SHADOW_MAX_COPY
slots on the system are allocated using the priority list.
/FORCE_REMOVAL
ddcu
Expels the specified shadow
set member from the shadow set. The specified device must be a member
of a shadow set that is mounted on the system where the command is
issued. You cannot specify /ALL with /FORCE_REMOVAL.
If connectivity
to a device has been lost and the shadow set is in mount verification,
this qualifier causes the member to be expelled from the shadow set
immediately.
If the shadow set is not currently in mount
verification, no immediate action is taken. If connectivity to a
device has been lost but the shadow set is not in mount verification,
this qualifier lets you flag the member to be expelled from the shadow
set as soon the shadow set enters mount verification. If no action
has been taken on the specified member and you wish to clear the flag,
use /NOFORCE_REMOVAL.
If the shadow set is dismounted
before the member is expelled, the FORCE_REMOVAL request expires.
/LOG
Instructs the volume shadowing software to
display a brief message confirming that the SET SHADOW command completed. If /OUTPUT is also specified, this information
is written to the output file.
/MEMBER_TIMEOUT
=n ddcu:
Specifies the timeout
value to be used for a shadow set member. The specified device must
be a member of a shadow set that is mounted on the system where the
command is issued.
The value supplied by this qualifier overrides
the system parameter SHADOW_MBR_TMO for this specific device. Each
member of a shadow set can be assigned a different MEMBER_TIMEOUT
value.The valid range for n is 1 to 16777215
seconds.
The timeout value set by /MEMBER_TIMEOUT does
not persist after the shadow set is dismounted.
/MVTIMEOUT {=
n DSAn:|=n /ALL}
Specifies
the mount verification timeout value to be used for all shadow sets
on the cluster or for the shadow set, specified by its virtual unit
name (DSAn:). The specified shadow set must be mounted on the system
where the command is issued. The value supplied by this qualifier overrides the value
specified by the system parameter MVTIMEOUT for this specific shadow
set.
NOTE: You cannot change the value of MVTIMEOUT for a system disk.
Any attempt to do so results in an error.
The valid range for n is 1 to 16777215 seconds.
The timeout value set by /MVTIMEOUT does not persist after
the shadow set is dismounted.
/NAME=policy-name
Used with /POLICY=HBMM to define
a named host-based minimerge (HBMM) policy or used with /DELETE to
delete a policy. The policy is defined clusterwide. See detailed descriptions
under /DELETE and /POLICY.
Policy names are case insensitive
and must consist of 1 to 64 characters. Only letters, numbers, the
dollar sign ($), and the underscore (_) are allowed.
If
you create a default policy, you must assign it the name DEFAULT.
Creates
or deletes a policy for host-based minimerge (HBMM).
HBMM is
the only supported value for the /POLICY qualifier, and it must be
included. You can optionally specify a named policy, including DEFAULT,
or you can specify NODEFAULT to indicate that the shadow set to which
it is applied is not to use HBMM, including any DEFAULT policy. For
information about specifying policies and using the DEFAULT and NODEFAULT
names, see Chapter 8: “Host-Based Minimerge (HBMM) ”.
When /POLICY is
specified with /DELETE, it removes either a specified HBMM named policy
or the HBMM policy for a specific shadow set. You cannot delete the
NODEFAULT policy.
When /POLICY is specified with /NAME,
it defines a clusterwide named policy. When no qualifiers others than
/NAME and /DELETE are specified, /POLICY defines a policy for a specific
shadow set.
Deleting bitmaps with the DELETE/BITMAP command
causes a bitmap to be deleted. However, the shadowing software recognizes
this condition and starts a new bitmap immediately. To disable HBMM
bitmaps, you must use the command SET SHADOW/DISABLE=HBMM.
When defining a policy, you use
five keywords (MASTER_LIST, COUNT, RESET_ THRESHOLD, MULTIUSE, and
DISMOUNT) to control the placement and management of HBMM bitmaps.
An HBMM policy specification consists of a list of these keywords
enclosed within parentheses. Only the MASTER_LIST keyword is required.
If the COUNT and RESET_THRESHOLD keywords are omitted, default values
are applied.
The MULTIUSE and DISMOUNT
keywords specify the number of bitmaps to be converted to multiuse
bitmaps during the automatic and manual removal of members respectively.
If MULTIUSE is omitted, the automatic minicopy on volume processing
is not enabled. As a result, none of the HBMM bitmaps are converted
to multiuse bitmaps. If DISMOUNT is omitted, a maximum of six HBMM
bitmaps can be used as multiuse bitmaps.
MASTER_LIST=list
The MASTER_LIST
keyword is used to identify a set of systems as candidates for a master
bitmap. The list value can be a single
system name; a parenthesized, comma-separated list of system names;
or the wildcard character, as shown in the following examples:
MASTER_LIST=Node1
MASTER_LIST=(NODE1,NODE2,NODE3)
MASTER_LIST=*
When the system list consists of a single system name or the
wildcard character, parentheses are optional.
An HBMM policy must include at least one MASTER_LIST. Multiple
master lists are optional. If a policy has multiple master lists,
the entire policy must be enclosed with parentheses, and each constituent
master list must be separated by a comma as shown in the following
example:
There is no significance to the position of a system name in
a master list.
COUNT=n
The COUNT keyword specifies the number of systems
in the master list that can have master bitmaps. Therefore, the COUNT
keyword and its associated MASTER_LIST must be enclosed within a single
parenthetical statement.
The COUNT value specifies the number of systems on which you
want master bitmaps. It does not necessarily mean that the first n systems in the list are chosen.
When the COUNT keyword is omitted, the default value is six
or the number of systems in the master list, whichever is smaller.
You cannot specify more than one COUNT keyword per master list.
Examples:
(MASTER_LIST=(NODE1,NODE2,NODE3), COUNT=2
(MASTER_LIST=(NODE1,NODE2,NODE3),COUNT=2),
(COUNT=2, MASTER_LIST=(NODE4,NODE5,NODE6))
RESET_THRESHOLD=n
The RESET_THRESHOLD
keyword specifies the number of blocks that can be set before the
bitmap can be cleared. Each set bit in a master bitmap corresponds
to a set of blocks to be merged, so this value can affect the merge
time.
Bitmaps are eligible to be cleared when the RESET_THRESHOLD
is exceeded. However, the reset does not occur immediately when the
threshold is crossed. For more information about selecting a value
for this attribute, see the OpenVMS Support for Host-based
Minimerge (HBMM) document.
The reset threshold is associated with a specific HBMM policy.
Therefore, the RESET_THRESHOLD keyword can be defined only once in
a policy specification. Because its scope is the entire policy, the
RESET_THRESHOLD keyword cannot be specified inside a constituent master
list when the policy uses multiple master lists.
When the RESET_THRESHOLD keyword is omitted, the value of 1,000,000
is used by default.
Example:
(MASTER_LIST=*, COUNT=4, RESET_THRESHOLD=100000)
In a policy with multiple master lists,
a given system name can appear in only one master list. A shadow set
need not be mounted to have an HBMM policy defined for it. For more
/POLICY examples, see the SET SHADOW Examples section in
the HP OpenVMS DCL Dictionary .
MULTIUSE=n
The MULTIUSE keyword enables automatic minicopy
on volume processing. n specifies
the number of existing HBMM master bitmaps to be converted to MULTIUSE
bitmaps in the event that a shadow set member is removed from the
shadow set by the shadowing driver.
During loss of connectivity to a site or controller, shadowing
may remove a member from the shadow set. When the member is added
back to the shadow set, a full shadow copy occurs.
By converting a few of the HBMM bitmaps to multiuse bitmaps,
all the writes that are performed to the shadow set are recorded.
Thus, when the member is added back to the shadow set, the multiuse
bitmap can be used for a minicopy operation. This is much faster than
a full copy operation.
The value of n cannot exceed the implied
or explicit value of COUNT. If MULTIUSE is not specified, bitmaps
are not converted to multiuse and a full copy operation is required.
Fatal drive errors that remove a shadow set member do not cause a
multiuse conversion because the drive must be replaced, and therefore
requires a full copy operation. For more information, see “Multiuse Property for Host-Based Minicopies”
DISMOUNT=n
The DISMOUNT keyword allows all the 12 write bitmaps
to be used by Shadowing as multiuse bitmaps, thereby reducing the
single point of failure of single minicopy master bitmaps. n specifies the number of HBMM bitmaps to be converted
to multiuse bitmaps every time a member is dismounted from a shadow
set using the following command:
DISMOUNT/POLICY=MINICOPY
/PRIORITY=n DSAn:
Overrides the current default priority setting. Priorities range from
0 (lowest) to 10000 (highest). The default priority is 5000. A shadow
set with a priority of 0 is never considered for a merge or a copy
on the system. When a recovery operation (that is, either a merge
or a copy) is needed on multiple shadow sets, the shadow sets are
recovered in priority order from highest to lowest. The priority setting
is system specific; any change in priority made on a single system
does not propagate to the entire cluster and does not persist across
a system reboot. Once this qualifier has been applied to a shadow
set that is mounted, the setting persists across any subsequent DISMOUNT
and MOUNT commands.
Enables you to modify the default “cost”
assigned to each shadow set member (ddcu:). By modifying
the assignments, you can bias the reads in favor of one member of
a two-member shadow set, or, in the case of three-member shadow sets,
in favor of one or two members of the set over the remaining members.
The device specified must be a member of a shadow set that is mounted
on the node where the command is issued.
The valid range
for the specified cost (n) is 1 to 65,535 units.
You cannot specify /ALL with /READ_COST.
The shadowing
driver assigns default READ_COST values to shadow set members when
each member is initially mounted. The default value depends on the
device type and its configuration relative to the system mounting
it. The following list of device types is ordered by the default READ_COST
assignments, from the lowest cost to the highest cost:
DECram device
Directly connected device in the same physical location
Directly connected device in a remote location
DECram served device
Default value for other served devices
The value supplied by the /READ_COST qualifier
overrides the default assignment. The shadowing driver adds the value
of the current queue depth of the shadow set member to the READ_COST
value and then reads from the member with the lowest value.
Different systems in the cluster can assign different costs
to each shadow set member.
When you specify a shadow set
(DSAn) instead of a shadow set member,
the read cost setting for all shadow set members reverts to the default
read cost settings established automatically by the shadowing software.
The specified shadow set must be mounted on the system where the command
is issued. You an specify any value for the cost; the value is ignored,
and the setting reverts to the default settings.
After
you have applied this qualifier to a member, the setting remains in
effect as long as the member is part of the shadow set. If the member
is removed from the shadow set and later returned, this qualifier
must be specified again.
If the /SITE command qualifier
has been specified, the shadowing driver takes site values into account
when it assigns default READ_COST values. In order for the shadowing
software to determine whether a device is in the category of “directly
connected device in a remote location,” the /SITE command qualifier
must have been applied to both the shadow set and the shadow set member.
Reads requested for a shadow set from a system at site 1 are
performed from a shadow set member that is also at site 1. Reads requested
for the same shadow set from site 2 can read from the member located
at site 2.
NOTE: DECram can shadow a DECram disk to a physical disk. However,
be aware that in the current implementation of Volume Shadowing for
OpenVMS, if the physical disk goes away, you are writing to a volatile
disk.
/RESET_COUNTERS
Resets the shadowing-specific
counters that are maintained for each shadow set. Counters that are
reset to zero (0) are:
HBMM Reset Count
Copy Hotblocks
Copy Collisions
SCP Merge Repair Count
APP Merge Repair Count
You can display the
current settings of the counters using the SHOW SHADOW command. The HBMM Reset Count refers to the number of times the
RESET_THRESHOLD value is met. The RESET_THRESHOLD is the setting that
determines how frequently a bitmap is cleared. Using the HBMM Reset
counter, you can gauge the rate of threshold resets.
/RECOVERY_
OPTIONS=DELAY_
PER_SERVED_
MEMBER=n
Allows the
system manager to adjust the rating assigned to a system based on
a delay assessed for each MSCP served shadow set member on that system.
The value specified by this qualifier overrides the value established
by the SHADOW_PSM_RDLY system parameter. The default delay for each
MSCP served member is 30 seconds and the valid range for the specified
delay is 0 through 65,535 seconds. When a copy or merge operation
is needed on a shadow set that is mounted on multiple systems, OpenVMS
Volume Shadowing attempts to perform this work on a system that has
a local connection to all of the shadow set members. Systems are rated
with a penalty (delay time) assessed for each shadow set member that
is MSCP served to the system. No delay is added for local members,
so a system with all locally accessible shadow set members is likely
to perform the work before a system where one or more members is served.
IF /ALL is also specified, the specified delay is applied to all currently
mounted shadow sets. See TBS for more information.
/SITE = n {ddcu:| DSAn:}
Indicates to the shadowing
driver the site location of the specified shadow set (DSAn:) or shadow set member (ddcu:).
The
SHADOW_SITE_ID system parameter defines the default site location
of the shadow set. You can override the default location of the shadow
set with this qualifier.
The valid range for the site
location, represented by n, is 1 through 255.
If /ALL is specified, all shadow sets are assigned the new
value. The shadow set’s member site values remain unchanged.
After you apply this qualifier, the setting remains in effect
until you change it using a SET SHADOW/SITE command.
This qualifier can improve read performance because the member
that is physically local to the system is the preferred disk from
which to read, provided that you specify the /SITE qualifier for each
shadow set member and for the shadow set. (In a Fibre Channel configuration,
shadow set members at different sites are directly attached to the
system. For the Volume Shadowing and OpenVMS Cluster software, there
is no distinction between local and remote in multiple-site Fibre
Channel configurations.)
/STALL=WRITES[=nnn]
Pauses the write operations
for nnn seconds. The default time is SHADOW_MBR_TMO.
If a value is not specified for nnn, the lock
on write operations is released after SHADOW_MBR_TMO seconds. For
example:
SET SHADOW DSA42 /STALL=WRITES
In this example, the writes are
stalled to the shadow set for a period of SHADOW_MBR_TMO seconds.
SET SHADOW DSA42 /STALL=WRITES=60
In this example, the writes are
stalled to the shadow set for a period of 60 seconds.
/NOSTALL=WRITES[=nnn]
Releases the lock on write
operations after the specified period (nnn seconds).
After the specified period is over, writes are allowed to the shadow
set members. See the following example:
SET SHADOW
DSA42 /STALL=WRITES=60
SET SHADOW DSA42 /NOSTALL=WRITES=30
In this example, initially the
write operations are locked for a period of 60 seconds. On providing
the /NOSTALL qualifier, the writes are allowed to the shadow set after
a period of 30 seconds.
Using /DEMAND_MERGE to Start a Merge Operation
The /DEMAND_MERGE qualifier was created to force
a merge operation on shadow sets that were created with the INITIALIZE/SHADOW
command without specifying the /ERASE qualifier. The /DEMAND_MERGE
qualifier ensures that all blocks not in use by active files are the
same. The system manager can enter this command at a convenient time.
If the /ERASE qualifier was not used when the shadow set was created
with /INITIALIZE/SHADOW, and the SET SHADOW/DEMAND_MERGE command has not been executed, then the higher overhead of a full
merge operation on this shadow set is encountered after a system failure.
System managers can also use the SET SHADOW/DEMAND_
MERGE command if the ANALYZE/DISK/SHADOW command found differences
between the members of the shadow set (see “Using ANALYZE/DISK/SHADOW to Examine a Shadow Set”).
SHOW SHADOW Management Functions
The SHOW SHADOW command reports
on the status of the specified shadow set and indicates whether a
merge or copy operation is required, depending on the qualifier that
you specify. If a merge or copy operation is required, this command
reports whether it is pending or in progress. The qualifiers are
described in this section. To use this command, specify the shadow set’s virtual unit name, followed by the qualifiers
you want to use, as shown in the following example:
$ SHOW SHADOW DSAnnnn:/qualifier/qualifier/
/ACTIVE
This qualifier returns one of three possible states:
Merge or copy is not required
Copy is in progress on
node nnnnx at LBN xxxx
Merge is in progress on
node nnnnx
/COPY
This qualifier returns one of three possible states:
Copy is not required
Copy is pending
Copy is in progress on
node nnnnx at LBN xxxx
/MERGE
This qualifier returns one of three possible states:
Merge is not required
Merge is pending
Merge is in progress on
node nnnnx at LBN xxxx
/OUTPUT=file-name
This qualifier outputs any messages to the specified
file.
Example 4-8 shows
sample output from the SHOW SHADOW command:
Example 4-8 SHOW SHADOW Sample Output
$SHOW SHADOW DSA716:
_DSA716: TST716
Virtual Unit SCB Status: 0001 - normal
Local Virtual Unit Status: 00000010 - Local Read
Total Devices 2 VU_UCB 810419C0
Source Members 2 SCB LBN 000009C8
Act Copy Target 0 Generation 00A15F90
Act Merge Target 0 Number EDA9D786
Last Read Index 0 VU Site Value 5
Master Mbr Index 0 VU Timeout Value 3600
Copy Hotblocks 0 Copy Collisions 0
SCP Merge Repair Cnt 0 APP Merge Repair Cnt 0
Device $252$DUA716 Master Member
Index 0 Status 000000A0 src,valid
Ext. Member Status 00
Read Cost 42 Site 5
Member Timeout 120 UCB 8116FF80
Device $252$DUA1010
Index 1 Status 000000A0 src,valid
Ext. Member Status 00
Read Cost 500 Site 3
Member Timeout 120 UCB 811DD500
