It is important to prevent the workspace data sets from filling
up. If they do fill up, requests might be rejected and database inconsistencies
might occur. The procedure shown here can be used to increase the
size of the workspace data sets before a problem occurs. The procedure
can also be used to recover after the data sets have filled up.
Each workspace data set deals only with the connection between
the local node and a single target node. During this procedure, any
work originating at the local node to be sent to the remote node will
be lost. Work which was previously saved in the workspace data sets
will not be lost. Work originating at the remote node destined for
the local node will not be lost.
In this example, assume that you want to increase the size of node
ATLANTA's INMSG workspace data set used in communicating with node
RALEIGH. The name of the workspace data set depends on the name of
the RACF® subsystem, the PREFIX
information specified on the TARGET command, and the LU names of the
local and remote nodes. You could determine the name of this workspace
data set using the TARGET NODE(RALEIGH) LIST command. In this example,
assume that the name of this workspace data set is RRSF.WORK.ATLLU.RALLU.INMSG.
- On the local node, issue a TARGET command to make the connection
with the remote node dormant. This insures that work destined for
the local node from the remote node is queued at the remote node.
For this example, from ATLANTA issue:
TARGET NODE(RALEIGH) DORMANT
- The workspace data sets are VSAM files. From TSO, create
a new VSAM file using the IDCAMS DEFINE CLUSTER command. Use the MODEL
parameter to insure that the new VSAM file has the same properties
as the original VSAM file for the workspace data set. Use the RECORDS()
parameter to override the insufficient size of the original VSAM file.
Give the file a temporary name. For example:
DEFINE CLUSTER(
NAME('TEMP.INMSG')
MODEL('RRSF.WORK.ATLLU.RALLU.INMSG')
RECORDS(1000 750))
SYS1.SAMPLIB member IRRSRRSF contains
a sample member RRSFALOC with sample JCL to define the VSAM workspace
data sets.
- Issue a TARGET DELETE command to delete the remote node. This
command causes RACF to deallocate
the original VSAM file and release its control over the file. For
example, from ATLANTA issue:
TARGET NODE(RALEIGH) DELETE
- Copy the original VSAM file into the new VSAM file, using the
TSO REPRO command. For example:
REPRO IDS('RRSF.WORK.ATLLU.RALLU.INMSG') ODS('TEMP.INMSG')
- Delete the original VSAM file.
DELETE 'RRSF.WORK.ATLLU.RALLU.INMSG'
- Rename the new VSAM file to the name of the original VSAM file:
ALTER 'TEMP.INMSG' NEWNAME('RRSF.WORK.ATLLU.RALLU.INMSG')
ALTER 'TEMP.INMSG.*' NEWNAME('RRSF.WORK.ATLLU.RALLU.INMSG.*')
An
alternative is to create a new RRSF.WORK.ATLLU.RALLU.INMSG file, as
shown in step 2, specifying TEMP.INMSG
on the MODEL keyword, copy the information stored in TEMP.INMSG into
it, and then delete TEMP.INMSG.
- Update your RACF parameter
library to reflect the new file size on TARGET commands. The new file
size is the first number on the RECORDS keyword in the DEFINE CLUSTER
command, shown in step 2. For example,
if a member of the RACF parameter
library on ATLANTA contains the following command:
TARGET NODE(RALEIGH) PREFIX(RRSF.WORK) FILESIZE(500)
change
it to: TARGET NODE(RALEIGH) PREFIX(RRSF.WORK) FILESIZE(1000)
If
you fail to update the RACF parameter
library, and the node is ever deleted while the VSAM file is empty, RACF deletes the VSAM file. Then,
the next time the parameter library is processed, RACF will re-create the file using the smaller
file size.
- Issue the TARGET command to redefine the connection to the remote
node, specifying the appropriate configuration, protocol, and workspace
information. (Depending on how you have set up your RACF parameter library, you might have already
defined a parameter library member containing a TARGET command that
does this, and updated the TARGET command in the preceding step. If
so, you can issue a SET INCLUDE command to execute that member.)
Note: - After you perform this procedure, the INMSG file for the RALEIGH
node might be a different size than the OUTMSG file, and could be
on a different volume. If you do a TARGET NODE(RALEIGH) LIST command,
the output will reflect what was entered on the TARGET command that
established the connection, and this might not be the values currently
in effect.
- If an INMSG file becomes full, RACF deallocates
it, yet might still process the work (based on in-storage queues).
However, when each piece of work is done, RACF cannot delete the INMSG file record for
it while the INMSG file is deallocated. Consequently, if the INMSG
file, or a copy of it, is made operative again, the work might be
executed again.