HP XP P9500 Storage - Manual (Page 105)

You detect and re-create lost updates by checking database log files and other current information
at the primary site. Journal log file entries for most DBMS have the same clock information as the
TOD clock, which is used for I/O time-stamps (when Timer Type = system). Cnt Ac-J Z group
consistency time can be extremely useful when performing this detection and retrieval. You can
also use the sense information with system time stamp that is transferred via ERC.
Design your recovery plan so that detection and retrieval of lost updates is performed after the
application is started at the secondary site. The detection and retrieval process can take some time.
Prepare for file and database recovery using files for file recovery (for example, database log files
that have been verified as current).
Remote copy and disaster recovery procedures are inherently complex. Consult your HP account
team about recovery procedures.
Switching host operations to the secondary site
These procedures apply to single Cnt Ac-J Z pairs and Cnt Ac-J Z systems with multiple primary
and/or secondary P9500 systems.
The basic procedures for switching operations to the remote site are as follows.
NOTE: When a failure occurs in a primary or secondary system when multiple P9500 systems
are used, all journals in the extended consistency group are suspended and remote copy operations
are suspended. The consistency of data update sequence is maintained if a failure occurs in a
primary system, but the consistency is not maintained if a failure occurs in a secondary system.
1. If you need to perform an IPL of the secondary host system, delete SIMs and then perform the
IPL. When the IPL finishes, the S-VOLs are usually varied online. If the secondary volume is not
varied online, vary the volumes online.
If you do not need to perform an IPL of the secondary host system, ensure that the S-VOLs are
varied online before continuing.
2. Check whether the restore journal includes an S-VOL with a pair status of Pending duplex or
Suspend (equivalent to SUSPOP in Business Continuity Manager).
1.
If such a pair exists, consistency in the S-VOL is suspect. Recovery with guaranteed
consistency is impossible. In this case, to use the S-VOL, you must release the pair.
2. If such a pair does not exist, run the YKSUSPND REVERSE option on the restore journal
to split the pairs. Doing this ensures consistency in the S-VOLs and thus makes them usable.
If an error occurs, consistency in the S-VOL is dubious, and recovery with guaranteed
consistency is impossible. In this case, to use the S-VOL, you must release the pair.
If no error occurs, wait until the split operation finishes.
3. Run the YKRESYNC REVERSE option on the restore journal (YKRESYNC is a command for
restoring a pair). This option attempts to restore the pair and reverse the primary/secondary
relationship.
4. Check whether there is a pair whose pair status of the restore journal is Suspend (equivalent
to SWAPPING in Business Continuity Manager).
If such a pair does not exist, the pair is successfully restored and the copy direction is reversed.
Copying of data from the secondary site to the primary site starts.
See the Business Continuity Manager User Guide for more information.
Reversing copy direction from secondary to primary sites
When host operations have been switched to the secondary site, restore the primary site and
re-establish the Cnt Ac-J Z system from secondary to primary sites.
1.
Restore the primary system and data paths.
2. Bring up the host servers at the primary site.
General recovery procedures 105