Disaster Recovery
Advanced Recovery Tasks
Chapter 10492
Merging P1S files
of all nodes for
EADR
Another step is required for EADR after backup has been performed.
Information on shared cluster volumes in P1S files for all nodes in the
MSCS has to be merged so that P1S file of each node contains
information on shared cluster volumes configuration. It is namely
impossible to collect enough information to configure the disk in Phase 1
for shared disk volumes that are locked by another node at backup.
Merging P1S files is necessary to enable restore of all cluster shared
disks if all nodes in the cluster have experienced a disaster. To merge the
P1S files of all nodes, execute the mmerge.cmd command from the
<Data_Protector_home>\bin\drim\bin:
mmerge p1sA_path ... p1sX_path
Where p1sA is the full path of the first node’s P1S file and p1sX is the full
path of the P1S file of the last node in the MSCS. Merged P1S files will
be saved in the same directory as the source P1S files with the .merged
appended to their filename (for example,
computer.company.com.merged). Rename the merged P1S files back to
the original name (delete the .merged extension).
The mmerge.cmd command works only on Windows systems with Data
Protector Automatic Disaster Recovery component installed. If you are
using an UNIX Cell Manager, copy the P1S files to a Windows client
which has Automatic Disaster Recovery component installed and merge
the files. Rename the merged P1S files back to the original name and
copy them back to the Cell Manager.
Example for merging P1S files for MSCS with 2 nodes: mmerge
<Data_Protector_home>\config\dr\p1s\node1.company.com
<Data_Protector_home>\config\dr\p1s\node2.company.com. You
have to enclose the path in quotes on Windows if the path contains a
space character. The merged files will be node1.company.com.merged
and node2.company.com.merged. Rename the files back to their original
names (you will have to rename the source P1S files first):
node1.company.com and node2.company.com.
You can avoid merging P1S files after backup by moving all shared
cluster volumes temporarily to the node which you are going to back up.
In this case all required information about all shared cluster volumes
can be collected. In this case only that node can be the primary node.
The following must be true in addition to other prerequisites for disaster
recovery:
• the primary node must have write access to the quorum disk (the
quorum disk must not be locked)