RDF System Management Manual for H-Series RVUs (RDF 1.8)

NOTE: You must be sure that volumes on the primary system containing alternate key files
and indexes are protected by RDF. It is not sufficient to protect just the associated data file or
table (particularly in the case of alternate keys). Likewise, if primary partitions reside on volumes
protected by RDF, you must ensure that the secondary partitions are also configured for protection.
File System Errors Involving Data Files
File system errors can occur when:
• A file is created.
• A file is opened.
• A modify operation is performed on the file. Modify operations are those that the updater
might perform on an open file, such as updating the file (logical REDO/UNDO) or altering
the owner or security after the replication of a file creation.
Errors encountered are reported in the EMS event log.
If an updater process encounters a file-system error, it responds in either of the following ways
(depending upon the type of error that occurred):
• Restarts and retries the operation again by reprocessing all database updates since the last
restart point. If the updater takes this course of action, it continues to do so until the
underlying problem goes away. This would be the action, for example, if an updater process
cannot create a data file on a backup volume because that volume is protected by the
Safeguard security management subsystem; in this case, the updater logs error message 739,
with an error 48, and restarts.
• Skips the operation. This would be the action, for example, in response to an error 10 (“record
already exists”).
Reading Image Data
Write operations to the various sorted image trails occur asynchronously to one another. To
ensure correct operation, the updaters cannot read to the end-of-file. Instead, they can only read
as far as the receiver allows (determined by receiver “save” points in the image trail). Thus, on
a finely tuned RDF backup node, the RDF time delay (RTD) for an updater can typically lag from
1 to 15 seconds behind TMF processing. This 15-second delay does not mean that 15 seconds are
needed for the updater to catch up, however; that operation typically requires less than a second.
Purger Process
The updaters apply all audit records to their data volumes regardless of whether the associated
transaction has committed, has aborted, or is still in progress.
The purging of redundant image trail files is based on transaction information. Specifically, the
receiver process maintains general information on what transactions might be in each image file.
This information is system-wide, not specific to any particular image trail. The reasons for this
pertain to performance.
First, if the receiver had to maintain specific information about what transactions were actually
represented in each image file on each image trail, the extractor-receiver performance rate would
be seriously degraded. Therefore, the receiver keeps general information about all transactions
it has seen across all trails.
Second, because considerable checking must be done across all trails to determine what files can
be purged based on what transactions might be represented in the various files on the various
image trails, the purger process performs this task.
The purger process is a restartable process pair that runs on the backup system (it is started
during START RDF and runs even when the updaters are stopped; image files are purged,
however, only when updating is enabled).
54 Introducing RDF