Metro/Global Mirror operations
This chapter describes some general considerations for Metro Mirror and Global Mirror, when used within the context of Metro/Global Mirror. It provides hints and tips that are related to specific operations of Metro/Global Mirror. It includes the following topics:
33.1 Overview
This chapter includes some general operations that are preferred practices for a Metro/Global Mirror environment. Metro/Global Mirror has characteristics beyond a simple combination of Metro Mirror and Global Mirror. Some of the topics described here are preferred practices that ensure that you take special care because of the complexity of Metro/Global Mirror.
Figure 33-1 shows a general Metro/Global Mirror configuration. Host I/O is directed to the local site labeled H1. Metro Mirror is used to synchronously mirror data to the intermediate site labeled H2 and then Global Mirror asynchronous mirrors data to the remote recovery site labeled H3.
Figure 33-1 Metro/Global Mirror configuration
33.2 General considerations for storage failover
A storage failover changes the access point of the data from the local to the remote storage system. The application is started by using the remote storage system. This has an a serious impact on a production site. This failover results in downtime for the applications because they must be started on the storage system of the secondary site. Depending on the configuration of the storage environment, access to the remote storage must be configured after the failover and before the applications can be restarted.
A storage failover is activated only for situations such as:
•Conditions that result from a disaster where the required infrastructure components are now inaccessible
•Maintenance operations at the local H1 site
•Migration of applications
Disaster recovery tests can be performed without performing a failover of the applications. Chapter 35, “Metro/Global Mirror disaster recovery test scenarios” on page 393 describes how to perform these tests.
Before you perform a failover of the Metro Mirror or the Global Copy relationship, you must verify the state of the pair relationships. For a planned failover, the status of the Metro Mirror between H1 and H2 must be Full Duplex mode. when the status is still Copy Pending, the data has not been completely copied to the secondary volumes.
It is important to understand the status of the entire Metro/Global Mirror environment before you act. Otherwise, your actions could result in a situation where either the Metro Mirror or the Global Copy environments cannot be recovered to their original state. In this case, the relationships might need to be completely re-created, requiring a full copy of the data.
The following examples show the status of the Metro/Global Mirror environment during normal operations.
In Example 33-1, all volumes of the Metro Mirror are in Full Duplex mode.
Example 33-1 Metro Mirror Full Duplex for H1 to H2
dscli>lspprc -remotedev IBM.2107-75ABTV1 6000-6003
ID State Reason Type SourceLSS Timeout (secs) Critical Mode First Pass Status
==========================================================================================================
6000:6200 Full Duplex - Metro Mirror IBM.2107-7520781/60 unknown Disabled Invalid
6001:6201 Full Duplex - Metro Mirror IBM.2107-7520781/60 unknown Disabled Invalid
6002:6202 Full Duplex - Metro Mirror IBM.2107-7520781/60 unknown Disabled Invalid
6003:6203 Full Duplex - Metro Mirror IBM.2107-7520781/60 unknown Disabled Invalid
The Global Copy pairs are always Copy Pending. To determine that all tracks are copied to the secondary site, you must inspect the Out of Sync Tracks value. The Global Copy environment is synchronized when the Out of Sync Tracks value is zero for all volumes, as shown in Example 33-2.
 
Note: The output data from the lspprc command is lengthy, so parts have been edited to better show the fields of interest.
Example 33-2 Synchronized Global Copy relationship
dscli> lspprc -remotedev IBM.2107-75ABTV1 -l 6400-6403
ID State Reason Type Out Of Sync Tracks ... Critical Mode First Pass Status
=================================================================================================
6400:6200 Copy Pending - Global Copy 0 ... Disabled True
6401:6201 Copy Pending - Global Copy 0 ... Disabled True
6402:6202 Copy Pending - Global Copy 0 ... Disabled True
6403:6203 Copy Pending - Global Copy 0 ... Disabled True
 
In a Metro/Global Mirror environment, the volumes at H2 are in a different state than in a conventional Metro Mirror or Global Mirror environment. Because the Global Mirror environment is cascaded to the Metro Mirror environment, the volumes at H2 are both a secondary and primary at the same time. Thus, the lspprc command that is run at H2 shows both the Metro Mirror and the Global Mirror environments, as shown in Example 33-3 on page 378.
Example 33-3 Status of the H2 volumes in a Metro/Global Mirror
dscli> lspprc -remotedev IBM.2107-7520781 6200-6203
ID State Reason Type SourceLSS Timeout (secs) Critical Mode First Pass Status
==========================================================================================================
6000:6200 Target Full Duplex - Metro Mirror 60 unknown Disabled Invalid
6001:6201 Target Full Duplex - Metro Mirror 60 unknown Disabled Invalid
6002:6202 Target Full Duplex - Metro Mirror 60 unknown Disabled Invalid
6003:6203 Target Full Duplex - Metro Mirror 60 unknown Disabled Invalid
6200:6400 Copy Pending - Global Copy 62 unknown Disabled True
6201:6401 Copy Pending - Global Copy 62 unknown Disabled True
6202:6402 Copy Pending - Global Copy 62 unknown Disabled True
6203:6403 Copy Pending - Global Copy 62 unknown Disabled True
33.3 Freezing and unfreezing Metro Mirror volumes
The freezepprc and unfreezepprc commands work in a Metro Mirror environment when the Peer-to-Peer Remote Copy (PPRC) paths are created with the -consistencygrp option. A freezepprc command does two things:
•It sets the extended long busy condition on all the primary volumes of the LSS. This condition blocks the I/O to these volumes. All I/O from the hosts waits until the long busy condition is removed by running unfreezepprc or when the long busy timeout value is exceeded. The default timeout value is 60 seconds for fixed-block architecture (FB) volumes and 120 seconds for count key data (CKD) volumes. Because the long busy timeout value is a value that is related to the LSS, you can change this value with either the chlss -extlongbusy command, for FB volumes, or the chlcu -extlongbusy command, for count key data (CKD) volumes.
•It fails the paths between the primary site and the secondary site.
For a detailed description about how consistency is provided in a Metro Mirror environment, see 16.6, “Metro Mirror Data consistency” on page 125.
As shown in Example 33-4, the pair status at the primary site after freezepprc runs is Suspended with the reason Freeze (for all volumes in the LSS). At the secondary site, the pair status is still in Target Full Duplex mode.
Example 33-4 Example of a freezepprc and the pair status at the primary site
# At H1:
#
 
dscli> freezepprc -remotedev IBM.2107-75ABTV1 68:62
CMUC00161W freezepprc: Remote Mirror and Copy consistency group 68:62 successfully created.
dscli> lspprc -remotedev IBM.2107-75ABTV1 6800-6803
ID State Reason Type SourceLSS Timeout (secs) Critical Mode First Pass Status
================================================================================================
6800:6200 Suspended Freeze Metro Mirror 68 unknown Disabled Invalid
6801:6201 Suspended Freeze Metro Mirror 68 unknown Disabled Invalid
6802:6202 Suspended Freeze Metro Mirror 68 unknown Disabled Invalid
6803:6203 Suspended Freeze Metro Mirror 68 unknown Disabled Invalid
 
# At H2:
#
 
dscli> lspprc -remotedev IBM.2107-7503461 6200-6203
ID State Reason Type SourceLSS Timeout (secs) Critical Mode First Pass Status
==========================================================================================================
6800:6200 Target Full Duplex - Metro Mirror 68 unknown Disabled Invalid
6801:6201 Target Full Duplex - Metro Mirror 68 unknown Disabled Invalid
6802:6202 Target Full Duplex - Metro Mirror 68 unknown Disabled Invalid
6803:6203 Target Full Duplex - Metro Mirror 68 unknown Disabled Invalid
The unfreezepprc command removes the long busy status from the primary volumes and I/O continues. The pair status of the primary volumes is still Suspended, as shown in Example 33-5.
Example 33-5 unfreezepprc after the freezepprc
dscli> unfreezepprc -remotedev IBM.2107-75ABTV1 68:62
CMUC00198I unfreezepprc: Remote Mirror and Copy pair 68:62 successfully thawed.
dscli> lspprc -remotedev IBM.2107-75ABTV1 6800-6803
ID State Reason Type SourceLSS Timeout (secs) Critical Mode First Pass Status
================================================================================================
6800:6200 Suspended Freeze Metro Mirror 68 unknown Disabled Invalid
6801:6201 Suspended Freeze Metro Mirror 68 unknown Disabled Invalid
6802:6202 Suspended Freeze Metro Mirror 68 unknown Disabled Invalid
6803:6203 Suspended Freeze Metro Mirror 68 unknown Disabled Invalid
A freezepprc command can be issued against the volumes of a running application, although it has an impact on the application: after the freezepprc command runs, the applications wait to continue with the I/O because the freeze command puts the volumes into the extended long busy state. This freeze is ended by the unfreezepprc command or after the consistency group timeout value is exceeded. This method is used to provide a consistent copy of the data at the secondary site when the applications at the primary site are not stopped.
33.4 Checking consistency at H3
The Metro Mirror paths from H1 to H2 ensure consistency through the use of the freeze and unfreeze functions. The paths must be created with the -consistgrp option. Consistency is provided to the Metro Mirror environment by an extended long busy condition (set at the primary volumes) when the last Metro Mirror physical link fails and no logical path is available. When I/O to the primary volumes can resume, an unfreezepprc command is issued to the H1 LSS’s to remove the extended long busy condition.
The Global Mirror process runs at H2 and uses FlashCopy at H3 to preserve consistency groups. When more than one storage system is used at H2, one of these systems is the master, where the Global Mirror environment is (see 32.1.2, “Metro/Global Mirror with multiple storage systems” on page 363). The other storage systems are the subordinates. The Global Mirror environment coordinates the consistency formation for all volumes of the master and the subordinates, which are joined into a Global Mirror session. The consistency is formed in the following steps:
1. The master coordinates with all the subordinates to stop I/O for 3 - 5 ms to all primary volumes that form a consistency group.
2. The consistency group is transmitted to the secondary volumes.
3. FlashCopy copies all the tracks that changed since the last write operation to the secondary volumes.
For a detailed description of consistency formation in a Global Mirror environment, see 19.2.2, “Consistency group formation” on page 208.
Section 21.2.2, “Creating FlashCopy relationships: H2 to J2 volumes” on page 235 describes how to verify the FlashCopy status and gives examples on how to perform the revertflash and reverseflash operations to return the H3 volumes to the last successful consistency group.
33.5 Setting up an additional Global Mirror from H3
When a planned failover of the production to H3 is done, depending on the reason for the failover, it might be that the production environment remains at H3 for an extended period. During this period, H3 volumes are not protected against a disaster. If the H2 site is still available, it is possible to create an additional Global Mirror session from H3 to H2 to protect the data against a possible disaster at H3.
To accomplish this task, more volumes are required at H2 for setting up the a FlashCopy relationships as a new Global Mirror secondary. In addition, H2 must be prepared for these Global Copy relationships from H3. This task is necessary because the H2 volumes are still secondaries for the Metro Mirror environment and a volume cannot be a secondary of two different relandscapes at the same time. Finally, a session with the related volumes must be created and the Global Mirror must be started at H3.
Figure 33-2 shows the steps to create the additional Global Mirror relationship.
Figure 33-2 Create an additional Global Mirror from H3
The creation of the additional Global Mirror relationship consists of the following steps:
1. Clean up Metro Mirror relationships
2. Create or failback from H3 to H2
3. Establish a FlashCopy path to the additional volumes at H2
4. Create a session and start the Global Mirror environment
33.5.1 Clean up the remaining Metro Mirror relationships
To reverse the Global Copy direction from H3 to H2, the remaining parts of the Metro Mirror environment at H2 must be removed. If the local storage device is still available, the Metro Mirror environment can be removed by running rmpprc. But when H1 is gone, you failover the Metro Mirror environment at H2 before the Metro Mirror can be removed in the opposite direction.
33.5.2 Create Global Copy pairs from H3 to H2
The volumes at H3 (see Figure 33-2 on page 380) are a primary for the Global Mirror environment, which is established from H3 to H2. The first step is to establish the Global Copy pairs from the remote H3 volumes to the intermediate H2 volumes.
Example 33-6 shows how to set up the Global Copy environment. Because the data at H3 is the same as it is at H2, the Global Copy environment can be established with the -mode nocp option to avoid background copy.
Example 33-6 Establish Global Copy from H2 to H3
dscli> mkpprc -remotedev IBM.2107-75ABTV1 -type gcp -mode nocp 6400-6403:6200-6203
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 6400:6200 successfully created.
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 6401:6201 successfully created.
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 6402:6202 successfully created.
CMUC00153I mkpprc: Remote Mirror and Copy volume pair relationship 6403:6203 successfully created.
dscli>
dscli>
dscli>
dscli> lspprc -remotedev IBM.2107-75ABTV1 -l 6400-6403
ID State Reason Type Out Of Sync Tracks Tgt Read Src Cascade SourceLSS Critical Mode First Pass Status
================================================================================= =========== ================================
6400:6200 Copy Pending - Global Copy 0 Disabled Enabled 64 Disabled True
6401:6201 Copy Pending - Global Copy 0 Disabled Enabled 64 Disabled True
6402:6202 Copy Pending - Global Copy 0 Disabled Enabled 64 Disabled True
6403:6203 Copy Pending - Global Copy 0 Disabled Enabled 64 Disabled True
33.5.3 Create FlashCopy relationships at H2
The Global Mirror environment at H3 requires FlashCopy relationships at the intermediate site from the H2 volumes to a new set of volumes named J2 (see Figure 33-2 on page 380).
Example 33-7 shows how to set establish the FlashCopy relationships.
Example 33-7 Establish FlashCopy at H2
dscli> mkflash -tgtinhibit -record -persist -nocp 6200-6203:6300-6303
CMUC00137I mkflash: FlashCopy pair 6200:6300 successfully created.
CMUC00137I mkflash: FlashCopy pair 6201:6301 successfully created.
CMUC00137I mkflash: FlashCopy pair 6202:6302 successfully created.
CMUC00137I mkflash: FlashCopy pair 6203:6303 successfully created.
33.5.4 Create Global Mirror at H3
To complete the setup of the Global Mirror, create a session at H3 and add the Global Copy primary volumes in to the session. Finally, start the Global Mirror environment.
Example 33-8 shows the setup of the session and the Global Mirror environment and how to check whether the Global Mirror environment is running properly.
Example 33-8 Create Global Mirror at H3
dscli> mksession -lss 64 1
dscli> chsession -lss 64 -action add -volpair 6400-6403:6200-6203 1
CMUC00145I mksession: Session 1 opened successfully.
CMUC00147I chsession: Session 1 successfully modified.
dscli> lssession 64
LSS ID Session Status Volume VolumeStatus PrimaryStatus SecondaryStatus FirstPassComplete AllowCascading
=========================================================================================================================
64 01 CG In Progress 6400 Active Primary Copy Pending Secondary Simplex True Enable
64 01 CG In Progress 6401 Active Primary Copy Pending Secondary Simplex True Enable
64 01 CG In Progress 6402 Active Primary Copy Pending Secondary Simplex True Enable
64 01 CG In Progress 6403 Active Primary Copy Pending Secondary Simplex True Enable
dscli> mkgmir -lss 64 -session 1
CMUC00162I mkgmir: Global Mirror for session 1 successfully started.
dscli> showgmir 64
ID IBM.2107-75ABTV2/64
Master Count 1
Master Session ID 0x01
Copy State Running
Fatal Reason Not Fatal
CG Interval (seconds) 0
XDC Interval(milliseconds) 50
CG Drain Time (seconds) 30
Current Time 11/14/20nn 10:38:21 CET
CG Time 11/14/20nn 10:38:21 CET
Successful CG Percentage 10
FlashCopy Sequence Number 0x43785B0D
Master ID IBM.2107-75ABTV2
Subordinate Count 0
Master/Subordinate Assoc -
 
..................Content has been hidden....................

You can't read the all page of ebook, please click here login for view all page.
Reset
3.16.130.201