Important Information
for FDRPAS V5.4 level 16 customers:
In V5.4/16, Innovation added a change requested by IBM: after a swap,
FDRPAS calls an IBM service called IEEVARYD to do the equivalent of a
console VARY uuuu,ONLINE, UNCOND command to refresh control blocks related
to the target device. IBM has discovered a bug in that IEEVARYD service.
IBM APAR OW54976 has been opened to address this problem but as of today
(6/12/2002), no fixes are available yet.
The problem is that IBM may free the data areas and control blocks in
SQA that are in use by an active I/O. When the I/O completes, it may overlay
SQA storage that no longer belongs to it. If that storage has been acquired
by some other function, it may result in application or system failures.
This problem only occurs when the IEEVARYD function is abnormally terminated.
FDRPAS has a timer which terminates the function if it takes too long,
so if the I/Os issued by IEEVARYD take an excessive amount of time, the
problem may occur. To date, we have seen this problem only one time but
it did require a reIPL of a customer system.
When the PTFs for IBM APAR OW54976 become available, we recommend that
they be applied as soon as possible to eliminate the possibility of the
problem.
In the meantime, FDRPAS customers have two choices:
-
You can apply FDRPAS fix P-54.0216 (included below), which will change
the timeouts used by FDRPAS and greatly reduce the likelihood of the
problem occurring.
You can add the undocumented keyword ",VARYON=NOAFTER" to EVERY SWAP
and MONITOR statement to bypass issuing the IEEVARYD call. However,
this may result in the problem the call was added to fix, namely failures
using Concurrent Copy and Flashcopy after a swap.
***** ZAP-ID : P-54.0216
* DATE : 02.140
* PREREQ : V 5.4/16
* SYMPTOMS : FDRPAS:FDRPAS:MSG FDR260 VARY ONLINE FAILED
* CODE=0016 0032 0000. IT MAY RESULT IN A SYSTEM
* DUMP WHICH SHOWS ABEND U0107 AND/OR S13E.
* THE SWAP IS SUCCESSFUL BUT IT MAY RESULT IN A
* ESQA OVERLAY WITH UNPREDICTABLE EFFECTS
* OR SYSTEM FAILURES.
* PROBLEM : FDRPAS ISSUED A INTERNAL VARY ONLINE COMMAND WHICH
* ISSUED AN I/O WHICH HUNG FOR UNKNOWN REASONS.
* AFTER 20 SECONDS FDRPAS ISSUED A U0107 ABEND
* TO TERMINATE THE VARY ONLINE. DUE TO AN IBM BUG,
* THE HUNG I/O MAY READ INTO AN ESQA AREA WHICH HAS
* BEEN FREEMAINED BY THE VARY PROCESS. IBM APAR
* OW54976 HAS BEEN OPENED TO ADDRESS THIS PROBLEM.
* SOLUTION : CHANGE MIH VALUE TO 5 SECONDS ON ALL I/OS FROM
* VARY ONLINE. AFTER FIRST TIMEOUT (10 SECONDS)
* LOWER THE IOS LEVEL AND THEN WAIT AN ADDITIONAL
* 120 SECONDS BEFORE ABENDING WITH A U0107 INSTEAD
* OF THE 10 SECONDS WE NOW WAIT.
* NOTE : THIS CHANGE SHOULD SUBSTANTIALLY ELIMINATE THE
* OCCURRENCES OF THE IBM BUG, BUT CANNOT GUARANTEE
* IT WILL NOT OCCUR. TO INSURE THAT THE PROBLEM
* CANNOT OCCUR, ADD THIS OPERAND TO EVERY SWAP AND
* MONITOR STATEMENT: VARYON=NOAFTER
* HOWEVER, THIS MAY LEAVE YOU EXPOSED TO THE PROBLEM
* WHERE CERTAIN FUNCTIONS, SUCH AS CONCURRENT COPY
* AND FLASHCOPY, MAY NOT WORK AFTER A SWAP.
* ONCE PTFS BECOME AVAILABLE FOR IBM APAR OW54976,
* INNOVATION RECOMMENDS APPLYING THE APPROPRIATE PTF
* AS SOON AS POSSIBLE.
* MODULE(S) : FDRPAS
*
* THE FOLLOWING ZAP IS FOR LEVEL 16 ONLY
*-
NAME FDRPAS FDRPAS
IDRDATA P540216
VER 882A 47F0,B25C
VER 8B08 0A6B,47F0,B4C8
VER 8BB0 47F0,B564
VER 92EC BD00,BD02
REP 882A 47F0,BD00
REP 8B08 47F0,BD18,0700
REP 8BB0 47F0,BD0C
REP 92EC 41E0,00C8,50E0,B5F4,47F0,B25C,4100,0960
REP 92FC 5000,B5F4,47F0,B564,9101,5005,4780,BD28
REP 930C 94BF,5005,9640,5073,9205,507D,0A6B,47F0
REP 931C B4C8
CHECKSUM EC29EB94
*
***** END OF MODIFICATION
Recommended IBM and other maintenance
to be applied before running FDRPAS
REQUIRED FDRPAS MAINTENANCE:
If you have applied the PTF for IBM APAR OW53362,
and you are running FDRPAS V5.4/15 or 16, you must apply the FDRPAS fix
P-54.0215 to avoid a swap failures due to the change introduced by that
APAR. P54.0215 can be downloaded from the FDRPAS FTP site.
REQUIRED HDS (Hitachi Data Systems) MICROCODE UPDATE:
Customers swapping to a HDS 9xxx Lightning disk
subsystem must insure that the microcode level is 01-13-19/00 or higher.
Without this microcode, FDRPAS monitor tasks may not recognize that a
swap is starting.
REQUIRED AND RECOMMENDED IBM MAINTENANCE:
Please check this matrix against your operating
system level to see which IBM APARs should be applied (contact Innovation
if you are running an earlier level).
IBM |--------- OS/390 ---------| |----z/OS-----|
APAR 2.4 2.5 2.6 2.7 2.8 2.9 2.10 1.1 1.2 1.3 1.4
OW30926 R R
OW31942 C
OW41858 C C C C C
ow44548 R R R R R R
OW45683 R R R R R
OW46101 R R R R R R R R
OW46459 C C C C C
OW46936 R R R R R R R
OW48166 R R R R R
OW49672 C C
OW49783 R R R
OW51248 R
OW51840 C C C C C C
OW52127 R R R R
OW52422 C C C C C C
OW52631 C C C C C C
OW53222 R R R R
OW54976 C C C C C C C C C
C = Critical R = Recommended
Brief IBM APAR descriptions follow (consult
IBM for complete APAR text). Note that some APARs may not be required
in your environment; see the text.
OW54976: you MUST apply the PTF to avoid SQA overlays due to a problem
in the IBM service IEEVARYD. However, as of 6/12/02 the PTFs are not yet
ready. FDRPAS V5.4/16 customers should apply FDRPAS fix P-54.0216 to reduce
the likelihood of this problem occurring.
OW53222/OW52127: you may want to apply the PTFs to prevent accidentally
IPLing from the old versions of SYSRES and IODF volumes which have been
swapped. These fixes are optional but recommended.
OW52631: if you swap to or from devices with PAV (Parallel Access Volumes),
You MUST apply the PTF for APAR OW52631 to avoid a S0C4 abend after swapping
a non-PAV device to a PAV device or vice versa. The error may occur when
trying to use the non-PAV device after the swap.
OW52422: If you have applied the PTF for APAR OW51163 or are running z/OS
1.3, you should apply this PTF to avoid a S09A ABEND with reason code
CB01 in GRS after a swap. This will only occur if there is a RESERVE on
the volume at the point of the actual swap; FDRPAS will not complete the
swap until there are no outstanding RESERVE but a RESERVE may be issued
after we check. This problem has only been observed on a JES checkpoint
volume.
OW51840: if you have applied the PTF for IBM APAR OW48166 or one of the
catalog level set PTFs UW81063/64/65, you MUST apply this fix. This problem
causes a loop in the catalog address space at the end of a swap if you
are NOT using ECS (Enhanced Catalog Sharing).
OW49783/OW51248: you may want to apply the PTF if you plan to do dynamic
I/O configuration after a swap (before the next IPL).
OW49672: you MUST apply the PTF to avoid a hang when swapping a volume
containing a shared catalog. The APAR describes a catalog performance
problem, but it has resolved several hangs during swaps.
OW48166: if you are using ECS (Enhanced Catalog Sharing) in a parallel
sysplex, you should apply the PTF before swapping any volumes containing
catalogs. The fix will automatically remove a catalog from ECS if it is
on a volume that is swapped. You must also apply the PTF for APAR OW51840.
To determine if you are using ECS, issue this console command on any system:
F CATALOG,ECSHR(STATUS)
if all catalogs displayed have a status
of "inactive", ECS is not in use. Circumvention: If you have not applied
the PTF, or you wish to avoid the catalog messages, IBM's recommendation
is to remove catalogs from ECS before you swap the volumes on which those
catalogs reside. Read the IBM APAR text for details.
OW46936: you may want to apply the PTF to avoid an occasional ABEND0C4
during a swap. The ABEND0C4 is not harmful (the swap will complete successfully)
but it causes an unnecessary SVC DUMP.
OW46459: if you swap to or from devices with PAV (Parallel Access Volumes)
and are using WLM-managed dynamic aliases, you MUST apply the PTF to solve
problems with binding and unbinding aliases.
OW46101: you may want to apply the PTF to fix performance problems on
LLA-managed datasets after a swap.
OW45683: you may want to apply the PTF to fix performance problems after
swapping to a device with PAV (Parallel Access Volumes).
OW44548: If you have ever used FDR to convert DB2 or other linear VSAM
clusters from 3380 disks to 3390 disks, and you are now swapping those
clusters to IBM 2105 Sharks, you should apply the PTF to avoid I/O errors
when re-loading or extending those clusters after the swap. Circumvention:
delete/define and reload the clusters before or after the swap.
OW41858: you MUST apply the PTF before attempting to swap to or from a
device with PAV (Parallel Access Volumes). This PTF adds a PAV interface
routine that is invoked by FDRPAS.
OW31942: you MUST apply the PTF before attempting to swap a volume from
a device with a 3-digit device address to one with a 4-digit address.
OW30926: if you plan to swap volumes containing system couple datasets
(CDS) it is recommended, but not required, that you apply the PTF. This
fix allows the coupling facility to better tolerate short delays in I/O
to the datasets, which may occuring during the swap. Without the fix,
there is a possibility that a coupling facility failure may occur. See
Section 320.02 of the FDRPAS manual for more information.
|
