Description
The startrcrelationship command
starts a stand-alone relationship. The command fails if it is
used to start a relationship that is part of a consistency group.
This
command can be specified only to a relationship that is connected.
For a relationship that is idling, this command assigns a copy direction
(primary and secondary roles) and begins the copy process. Otherwise,
this command restarts a previous copy process that was stopped either
by a stop command or by some I/O error.
Note: A command in idling state is rejected if any of the indicated
secondary volumes is the target of an existing FlashCopy map.
If the FlashCopy mapping
is active, the remote copy cannot be started.
If an existing remote copy relationship is stopped with
-access specified but is restarted and the resultant secondary volume
(depending on the choice of primary) is mapped to a host of type
hide_secondary,
that volume is not presented to the host. This is true even though it is mapped for configuration
purposes. The mapped volumes are presented to the host if the:
- Host type is changed to a type other than hide_secondary
- Remote copy relationship is stopped and -access is specified
- Volume ceases to be a secondary volume because the remote copy relationship is being deleted or
switched
In the idling state, you
must provide the -primary parameter. In other
connected states, you can provide the -primary parameter,
but it must match the existing setting.
The
-force parameter
is required if consistency would be lost by starting a copy operation.
This situation can occur if input transactions occur on either the
primary or secondary volumes since the
ConsistentStopped or
Idling state
occurred. This situation occurs when the relationship is in either
of these states:
- ConsistentStopped but not synchronized
- Idling but not synchronized
After restarting a relationship
in either of these states, the data on the secondary volume is not
usable for disaster recovery until the relationship becomes consistent.
A Global Mirror relationship with
a cycling_mode of multi in either
of these states does not require the -force parameter
because a consistent secondary image is retained. However, if such
a relationship is in idling state and written data has been received
at the secondary volume, the -force flag is still
required, because the secondary volume has a divergent image that
cannot represent a consistent earlier state.
The
-force parameter is not required
if the relationship is in one of the following states:
- InconsistentStopped
- InconsistentCopying
- ConsistentSynchronized
However, the command does not fail if you specify the
-force parameter.
A Global Mirror relationship with
a cycling mode of:
- none uses the non-cycling Global Mirror algorithm
- multi must:
- Use a change volume that is configured at the primary volume (or
the command fails)
- Use a change volume that is configured at the secondary volume
(or the command fails)
- Perform multiple cycles of cycling
After creating a background
copy the relationship remains in copying state, waiting for the remainder
of the period time to expire before performing a new cycle. If the
secondary change volume is unconfigured when the background copy completes,
the relationship stops as if there is no cycle period.
Relationships
that are active-active must have a state of idling to
be started. (You must specify -primary to determine
which of the master and auxiliary copies become the primary when starting
an idling relationship.)
Use this command
to:
- Restart the active-active relationship copy process
and retain the historical disaster recovery copy that access has been
granted to (which might have been used while the up-to-date copy was
offline)
- Switch back to an up-to-date copy in the same state it was before stoprcrelationship
-access was specified. Any changes that are made to the historical
copy are discarded
Remember: If you switch back to the up-to-date copy,
you might have to take host actions to prepare for the volume data
changing.
After specifying
this command, if the secondary copy is not a historical copy of the
primary relationship, it cannot be used for disaster recovery (and
disaster recovery availability is restored after the copies are resynchronized).
This situation can occur when:
- The new primary is the historical copy, which means the new secondary
copy contains data that is from a later point in time than the data
that the primary contains
- The secondary copy is the historical copy and was modified between
the specification of stoprcrelationship -access and startrcrelationship
-primary command (Which means the secondary copy represents
a divergent data image)
This command copies only the regions that are needed to resynchronize
the two copies.