To confirm that your systems are recording the correct SMF data
and that other environmental factors will support the use of ICFRU
in an actual recovery situation, we recommend that you execute ICFRU,
using your own data as input. In general, the process will
"simulate" a
real recovery procedure (see
Executing Catalog Recovery),
with these exceptions:
- The actual recovery (that is, the IMPORT) need not be done.
- A later EXPORT should be created to compare with
the new EXPORT created by CRURRAP.
In time sequence, the data should look as shown in
Figure 1.
Figure 1. Time Sequence for SMF Data Collection.
Both
of the two EXPORTs must be inside the interval spanned by the SMF
data. ---
- First EXPORT |
|
|
| SMF Data
|
|
|
- Second EXPORT |
---
In other words, the SMF data should start before the first EXPORT
and continue after the second EXPORT, completely spanning the period
between the EXPORTs.
To confirm installation readiness, use this somewhat simplified
version of the general recovery procedure in Executing Catalog Recovery:
- Do two EXPORTs. The amount of time between them is at your discretion.
Bear in mind that in a real situation you are unlikely to want to
create backups for catalogs less often than once a day, unless the
catalog has low activity. Large quantities of SMF data can be time-consuming
to process and difficult to manage.
- Take note of the date and time information from the top of the
SYSPRINT output for the EXPORTs (message IDC0594I).
- Ensure that you have all the SMF data covering
the intervening period. It may be necessary to switch and dump SMF
data sets to accomplish this, depending on your installation procedures
for handling SMF data. ICFRU cannot handle SMF data directly from
the VSAM SMF data sets, nor can it handle concatenation of unlike
devices, so all the SMF data must be gathered together onto similar
media or batched and then concatenated after multiple passes with
CRURRSV.
- Run the CRURRSV, SORT and CRURRAP steps as shown
in Figure 1, Figure 2 and Figure 3. The specified start time
should be the time of the first EXPORT; the specified stop time should
be the time of the second EXPORT.
- Compare the EXPORT data set created by ICFRU processing with the
second IDCAMS EXPORT data set you created at the simulated failure
point. See Figure 2 for suggested JCL.
Figure 2. Comparing
the Results of CRURRAP with an IDCAMS EXPORT.
Use this
JCL (for tests or rehearsals) to compare the output of ICFRU with
an EXPORT copy of the catalog created at the simulated point of failure.//********************************************************
//* INTEGRATED CATALOG FORWARD RECOVERY UTILITY *
//* JCL EXAMPLE - *
//* THIS JCL EXECUTES IEBCOMPR TO COMPARE THE NEW EXPORT *
//* PRODUCED BY CRURRAP WITH A SECOND EXPORT TAKEN AT *
//* THE RECOVERY STOP TIME. *
//* *
//* THIS COMPARISON IS TO BE USED ONLY FOR TESTING OR *
//* REHEARSAL SITUATIONS SINCE A SECOND EXPORT WILL *
//* NOT EXIST IN A REAL RECOVERY SITUATION. *
//* *
//* THIS STEP SHOULD ALSO BE EXECUTED AS A PART OF THE *
//* INSTALLATION READINESS CONFIRMATION. *
//* MODIFY THE SYSUT1 DD STATEMENT TO POINT TO *
//* THE FIRST (OLDER) EXPORT DATA SET. *
//* MODIFY THE SYSUT2 DD STATEMENT TO POINT TO *
//* THE SECOND (NEWER) EXPORT DATA SET. *
//********************************************************
//* COMPARE OUTPUT FROM CRURRAP WITH 'REAL' EXPORT.
//********************************************************
//COMP EXEC PGM=IEBCOMPR
//SYSPRINT DD SYSOUT=*
//SYSUT1 DD DSN=Baplicat.NEW.CATALOG.EXPORT,DISP=SHR
//SYSUT2 DD DSN=Baplicat.SECOND.CATALOG.EXPORT,DISP=SHR
//SYSIN DD DUMMY
- Examine the output from the program runs including any miscompares
from IEBCOMPR.
In the IVP, the second EXPORT data
set should compare exactly with the EXPORT data set created by CRURRAP. This
will not be true for all cases where this technique of comparing two
EXPORTs is used. The control records in the EXPORT data sets contain
values representing the status of the catalog at the time of EXPORT.
Since the EXPORTs are from different times, certain values may miscompare
in records 3, 5, and 7. This is of no concern since the catalog will
be redefined during the actual IMPORT and these values will not be
used.
There should be no miscompares after the last EXPORT
control record (currently record number 8) except when VVR relative
byte addresses (RBAs) have changed in the ecord. Again, this is of
no concern and is expected as catalog management will reestablish
the correct values when the catalog entry is first accessed.
See Sample Reports from ICFRU for examples of correct output
from ICFRU. Other sections in this appendix describe the logs, messages,
and reports, including interpretation and suggested responses.
- If you find other errors during program execution, or other miscompares,
check the following:
- SMFPRMxx member in SYS1.PARMLIB - Are you recording record types
61/65/66 for all jobs, including subsystems and system tasks?
- SMF data - Are you sure you have all the SMF data covering the
period between the two EXPORTs?
- Execution parameters - Are you certain that the start and stop
times correspond to the times of the two EXPORTs (see the SYSPRINT
output from EXPORT)?
- RBA values - Did the VVR RBA values in the record change? This
is to be expected. This is not an error. If the record appears to
be otherwise intact, do not be concerned.