6.3 HP StoreAll Storage File System User Guide (TA768-96093, June 2013)

The next command imports audit metadata exported by the ibrix_audit_reports command:
MDimport -f target -t audit -n
simple_report_for_source_at_1341513594723.csv
The ibrix_audit_reports command automatically generates the file name
simple_report_for_source_at_1341513594723.csv.
Metadata and continuous remote replication
When continuous remote replication (CRR) is configured for a file system, or run-once replication
is performed, metadata that is stored only in Express Query from the source cluster is not transferred
to the target cluster. System metadata stored in the inodes, such as file permissions, are transferred
intact by remote replication. Custom and audit history metadata, which are stored only in the
Express Query database, are not transferred, but can be exported manually and periodically to
files that CRR will replicate.
Express Query on the target system processes files replicated to its file system by CRR as new file
creations and modifications, with the dates and times the files were replicated. The audit history
contains the history of file replications from the source system, not the audit history of file accesses
on the source system. The target Express Query is not aware of any custom metadata applied to
files on the source system; however, you can add the source system’s audit history and custom
metadata to the target Express Query’s database by exporting and importing this metadata.
To include Express Query metadata in remote replication, you can create shell scripts (or scripts
in other languages) and set up periodic execution of those scripts by using standard Linux tools
such as “cron”. Such scripts must perform the steps of the export of custom metadata and audit
history described in “Saving and importing file system metadata” (page 219), with parameters
appropriate to your file systems. The exported files are copied to the target automatically by
continuous remote replication (CRR). Those files on the target can be imported into Express Query
running on the target at any time, if the target needs to become the active Express Query-enabled
file system.
The output file listed on the MDExport.pl command line must be in a directory that will be
replicated by CRR.
NOTE: The <mountpoint>/.archiving directory is excluded from replication. The
ibrix_audit_reports command creates its report output file in the <mountpoint>/
.archiving/reports subdirectory. Therefore, after issuing the ibrix_audit_reports
command, your script must move or copy the report output file to another directory on the file
system outside the .archiving tree for it to be replicated.
Metadata and synchronized server times
Metadata database updates often require coordinated actions from multiple StoreAll servers.
Therefore, it is important to keep your nodes as close together in time as possible, using NTP.
Modifications to the same file from different servers within the time difference between StoreAll
servers could cause inconsistencies between system metadata stored in the file inode and the
metadata stored in Express Query. To ensure consistency, keep servers synchronized with NTP
and avoid modifying a file from clients connected to different StoreAll servers at the same time. If
the server cannot be chosen by the client, such as when using external load balancers, then separate
your client operations on the same file beyond the maximum expected time difference between
any two StoreAll servers.
Metadata and synchronized server times 221