NEOHAPSIS - Peace of Mind Through Integrity and Insight

Neohapsis is currently accepting applications for employment. For more information, please visit our website www.neohapsis.com or email hr@neohapsis.com

Neohapsis > Archives > IBM AIX Digests> 2002-q1

From: AIX Service Mail Server (aixservaustin.ibm.com)
Date: Tue Feb 26 2002 - 02:31:47 CST

Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

APAR: IC32650 COMPID: 5724A6000 REL: 710
ABSTRACT: INTELLIGENT MINER SCORING PERFORMANCE PROBLEM ON DB2 UDB EEE

PROBLEM DESCRIPTION:
The customer has Intelligence Miner Scoring Services installed
and is running it against a DB2 database. The model, which is a
linear regression, resides on a single node with two partition.
One partition has the model and the table, while the other
partition just has the table. The customer has noted that the
part with the model and the table runs much better than the
other partition.
The partition with both the model and the table has inserted
about 10 million rows while the other partition has only
inserted about 300,000 rows.

LOCAL FIX:
To really apply the fixes in the fixpack she needs to modify her
SQL scripts as indicated in the instructions sent by email:
For each databases that were previously enabled for IM Scoring,
please do the following:
1. for a fenced database, do nothing,
2. for an unfenced database, connect to it and execute
"db2 -tf /usr/lpp/IMinerSc/bin/idmapplyupdt.db2".
In this fix, the 3 functions IDMMX.DM_applyClusModel,
IDMMX.DM_applyClasModel and IDMMX.DM_applyRegModel still
exist and have not been changed. The 3 new functions
IDMMX.DM_apply1ClusModel, IDMMX.DM_apply1ClasModel and
IDMMX.DM_apply1RegModel should solve the performance problem
in a parallel environment (in unfenced mode only).
Change your SQL scripts to use these new functions instead of
the original apply functions.

PROBLEM SUMMARY:
Performance problem on EEE at Bank of Montreal

------

APAR: IY18783 COMPID: 5765C3403 REL: 430
ABSTRACT: X.25 PERFORMANCE DEGRADATION

PROBLEM DESCRIPTION:
Running X.25 NPI test shows performance degradation. Also,
including workaround to connection sequence with X.25 CLE
(TELEMATICS ACP 70 model).

PROBLEM SUMMARY:
Addresses performance problem as well as problem with
TELEMATICS ACP 70 model on link bringup.

PROBLEM CONCLUSION:
Resolved performance and bringup problem with X.25 CLE.

------

APAR: IY21335 COMPID: 5765D5100 REL: 320
ABSTRACT: SWTADMD2 (SP SWITCH 2) DAEMON ERRORS LOGGED ALTHOUGH THIS

PROBLEM DESCRIPTION:
swtadmd2 (SP Switch 2) daemon are reported on a PSSP 32
(SP Switch 1) system.
following entry in /etc/inittab:
so by default: swt:2:once:/usr/bin/startsrc -g swt
if SP Switch 1 or SP Switch 2 is used.
the following errors are reported on a SP Switch 1 system
running with PSSP 3.2:
LABEL: CS_EMSTR_EXIT_ER
IDENTIFIER: FFDEEAA3

LOCAL FIX:
1/ start swtadmd and swtlog maually on a
   SP Switch_1 (with PSSP 3.2) with the startsrc -s
   command.
   and comment
   swt:2:once:/usr/bin/startsrc -g swt
2/ ignore those errors cause it's obvious that they
   occour if /usr/bin/startsrc -g swt is used.
1/ start swtadmd and swtlog maually on a
   SP Switch_1 (with PSSP 3.2) with the startsrc -s
   command.
   and comment
   swt:2:once:/usr/bin/startsrc -g swt

PROBLEM SUMMARY:
When the CWS is rebooted, the swt entry in /etc/inittab
attempts to start all the switch admin daemons. Those
that are not appropriate for the actual switch will not be
started but will cause SRC or CS_EMSTR_EXIT_ER error
report records to be generated. For example, on an
SP_Switch2, the swtadmd daemon is not valid; this daemon
will be inoperative, but there will be up to three SRC
error report records indicating swtadmd failed. On an
SP_Switch, the emasterd and swtadmd2 daemons are not
valid. These daemon will not be started but they
leave behind error records. The customer does not want
these error reports to be generated.

PROBLEM CONCLUSION:
The action for the swt entry in /etc/inittab is
changed from "startsrc -g swt" to
/usr/lpp/ssp/css/start_swt. This is a new script
which will start the appropriate swt daemons, e.g.
swtadmd and swtlog for the SP_Switch, and swtadmd,
emasterd and swtlog for the SP_Switch2. Note that
issuing the "startsrc -g swt" command, will still
cause the errpt records to be generated.

------

APAR: IY23554 COMPID: 5765D6600 REL: 201
ABSTRACT: CUMULATIVE MAINT #04 FOR MQSERIES INTEGRATOR FOR AIX V201

PROBLEM DESCRIPTION:
Cumulative Maint #04 for MQSeries Integrator for AIX V201

PROBLEM CONCLUSION:
This is the fourth PTF for IBM MQSeries Integrator Version 2.0.1
for AIX. It is delivered via PTF U480125 for APAR IY23554.
It contains fixes for ALL the problems listed below:
Name Abstract
IC29180 - THE INTERVAL DATETIME TYPE DOESN T WORK AS DOCUMENTED
IC29387 - MESSAGE TRANSLATION INTERFACE SEEMS TO HAVE A PROBLEM
IC29789 - IT APPEARS THAT A CDATA SECTION IS MARKED AS "SPECIAL"
IC29873 - NO ABILITY TO ASSIGN A VALUE OUTSIDE THE SPECIFIED RAN
IC30505 - ERROR HANDLING IN EXECUTION GROUP, CAUSES THE EXECUTIO
IC30970 - GLOBALLY CO-ORDINATED FLOW INVOKING NEON PARSER CAUSE
IC31150 - GETTING AN ERROR BIP1835E. THE CREATE RUNTIME MESSAGE
IC31766 - MQSIREPORTTRACE DOES NOT WORK CORRECTLY WITH USERNAMES
IC31802 - GETTING PARSING ERRORS WHILE USING REPEATING FIELDS WI
IC31806 - PROBLEM IMPORTING COBOL COPYBOOK. IMPORTER DOESN T LI
IC31825 - BOTHRESH AND BOQNAME NOT WORKING PROPERLY WHEN ERROR O
IC31916 - MESSAGE CONVERSTION TROUBLE (SI/SO HANDLING) WITH MQSI
IC32018 - WHEN A REFORMAT OPEARTION IN A NEON-MQSI-ENVIRONMENT D
IC32037 - MQSI ERROR RFH2 MESSAGE FAILS TO CONVERT. REASON 2116
IC32427 - V201 AND V202 TO V2.1 MRM MIGRATION AND COEXISTENCE
IY16261 - MQSI201 IGNORE NEON NODE. CSD1 FOR MQSI201 HAS NO MKE
IY22100 - CONTROL CENTER CACHE NOT REFRESHING PROPERLY.
IY22197 - ELEMENTS CONTAINING THE &APOS: ENTITY ARE NOT BEING HA
IY22757 - INCOMPLETE OR INVAVID SUBSCRIPTION DATA RETURNED DURIN
IY23076 - DEFINING BINARY ELEMENTS/FIELDS IN A MRM-MESSAGE SET W
IY23087 - EXCESSIVE CASHING PERFORMED BY MQSI CAN LEAD TO MEMORY
21983 - New Era renaming
16758 - Exponential growth of DeletePub messages in collective
16951.1 - Standard XML entities rejected when standalone="yes"
18105 - Mixed stress test fails with corupt core file
20404 - Multiple broker deploy fails
21173 - Using \u3044 causes filter error

------

APAR: IY24906 COMPID: 5765C3403 REL: 430
ABSTRACT: XSET +FP BADFONTDIR CAUSES SCALABLE FONT PBL

PROBLEM DESCRIPTION:
xset +fp badfontdir causes scalable font pbl.

LOCAL FIX:
xset fp rehash

PROBLEM SUMMARY:
xset +fp <invalid font directory> results in inability
to access scalable fonts

PROBLEM CONCLUSION:
Add a fix to dixfonts.c to rehash the fontpath after a
failure resulting from adding a non existing directory
to the fontpath.

------

APAR: IY25079 COMPID: 5765C3403 REL: 430
ABSTRACT: PROBLEMS WITH LDELETEPV AND DUMP HANGING

PROBLEM DESCRIPTION:
ldeletepv might cause problems on systems with
more than one disk on rootvg and depending on which
disk the dump device is residing.

LOCAL FIX:
Set your dump devices on hdisk0

PROBLEM SUMMARY:
Unable to remove pv from vg when dump device has
been changed.

PROBLEM CONCLUSION:
Fix dump device setup to ensure that only the
proper pv has dump_inited flag set.

TEMPORARY FIX:
Set dump device only as hdisk0; or reboot.

------

APAR: IY25152 COMPID: 5765D6100 REL: 220
ABSTRACT: LLSUMMARY CORE DUMPS DUE TO A NULL JOB_CREDENTIAL POINTER.

PROBLEM DESCRIPTION:
LLsummary core dumps with a segmentation violation when the
pointer for the job_crediential is NULL.

PROBLEM SUMMARY:
llsummary core dumps due to a null job_credential pointer

PROBLEM CONCLUSION:
The portion of code which reads the job credential
information should be changed to check for a NULL
pointer before attempting to extract data.

------

APAR: IY25987 COMPID: 5765C3403 REL: 433
ABSTRACT: NFS NOT FREEING PAGES WHILE REMOVING FILES

PROBLEM DESCRIPTION:
mount -o vers=3,proto=udp,biods=10 gingere:/test /test
dd if=/dev/zero of=/test/200m bs=1024k count=200
vmstat shows number of free pages going down to 183957 pages
time sync ==> sys 0m0.37s
time sync ==> sys 0m0.35s
rm /test/200m
time sync ==> sys 0m0.35s
time sync ==> sys 0m0.37s
vmstat still shows 183957 pages free

LOCAL FIX:
unmount the filesystem

PROBLEM SUMMARY:
Even after a file on an NFS filesystem has been deleted, the
segment is kept around for cache purposes. This is meant as
a performance enhancement when files are closed and reopened,
but if the file is deleted it can never be opened again.

PROBLEM CONCLUSION:
When a file is deleted, and link count reaches 0, then flush
the data out at the same time as it is deleted instead of
keeping it around.

------

APAR: IY26240 COMPID: 5765D5100 REL: 330
ABSTRACT: WINDOW SUSPEND FAILS BECAUSE KLAPI TIMER RETRY TOO FREQUENT

PROBLEM DESCRIPTION:
It's possible for VSD retry requests to interfere with the
switch device driver's attempts to suspend adapter windows
because the VSD retry is too frequent. The css adapter will be
taken off the switch and the css adapter.log file will show:
RESET suspend windows timed out
. . .
fs_daemon_fsm_adapter_thread_main: 2547-235 call to
QUERY_SUSPEND_WINDOW_COMPLETION ioctl failed, errno = 5

PROBLEM SUMMARY:
KLAPI can potentially spend too long in loop trying to
reregister a timer interrupt. If this happens, a recoverable
adapter failure may become an unrecoverable failure.

PROBLEM CONCLUSION:
The time KLAPI spends trying to reregister the timer is
reduced.

------

APAR: IY26258 COMPID: 5765D5100 REL: 320
ABSTRACT: FAILED RCP GETS RETURN CODE = 0

PROBLEM DESCRIPTION:
In pssp environment with kerberos enabled, rcp of non-existent
file returns rc=0. Module spk4rcp seems to be losing the correct
rcp return code.

PROBLEM SUMMARY:
When kerberos is active and if rcp is issued to copy a local
file to a remote system and the rcp fails for a reason
other than a kerberos error, an error message is displayed
but rcp exits with a return code of 0.

PROBLEM CONCLUSION:
The kerberos version of rcp has been modified to exit with
a non-zero return code if the rcp fails for any reason.

------

APAR: IY26309 COMPID: 5765D6100 REL: 220
ABSTRACT: NEGOTIATOR FAILS DURING RE-CONFIG

PROBLEM DESCRIPTION:
Negotiator crashes when re-config is being performed

PROBLEM SUMMARY:
The LoadLeveler negotiator daemon core dumps when handling a
reject transaction
at the same time as a reconfiguration is taking place. A
bug in locking between
the reconfiguration thread and the reject transaction thread
causes the reject
transaction thread to view data coming across the socket
incorrectly leading to
memory corruption.

PROBLEM CONCLUSION:
The solution is to adequately serialize the data reference
between
the reconfiguration thread and the reject transaction
thread.

------

APAR: IY26317 COMPID: 5765B9500 REL: 130
ABSTRACT: GPFS MOUNT OPERATION FAILED DURING INITILIZATION BECAUSE THE

PROBLEM DESCRIPTION:
GPFS mount operation failed with GPFS: 6027-592. Mount of <file
system> is waiting for the mount disposition. It appears that
gpfsready scripts in /var/mmfs/etc/ directory were clobbered/
changed after an upgrade to GPFS 1.4. Restoring the scripts
before the upgrade to GPFS 1.4 fixed the mount problem.

PROBLEM SUMMARY:
gpfsready script was clobbered during the
installation of GPFS 1.4.

PROBLEM CONCLUSION:
Upgrade the README for GPFS 1.3 and 1.4
which update these files to direct the administrator to save
the old versions and reimplement any user modifications to
gpfsready, mmfsdown and mmfsup.

------

APAR: IY26390 COMPID: 5765C3403 REL: 430
ABSTRACT: 'DD' BLOCKED ON TAPE DEVICE

PROBLEM DESCRIPTION:
'dd' blocked on tape device

PROBLEM SUMMARY:
dd command hangs. Cannot terminate dd process
on sending SIGINT.

PROBLEM CONCLUSION:
signal handler in dd code modified to
resolve the problem.

------

APAR: IY26461 COMPID: 5765D5100 REL: 320
ABSTRACT: PDF COMMAND RETURNES WRONG %FREE VALUES

PROBLEM DESCRIPTION:
pdf command returnes wrong %Free values
The %Free value is being formatted improperly and overlays the
Free-KBfield.
The problem is due to the formatting of the variable pfree in
the pdfpfck.cmds file, called by sysctl, which is called by pdf.

PROBLEM SUMMARY:
The output displayed by pdf was not readable. The fields
for Free-KB and %Free did not have a blank between them
and the %Free was being displayed as a real number, instead
of an integer.

PROBLEM CONCLUSION:
pdf was modified so that the %Free field is displayed as an
integer. This will then allow a space between the Free-KB
and %Free fields which make the output more readable.

------

APAR: IY26547 COMPID: 5765B9500 REL: 130
ABSTRACT: GPFS PROBLEMS CAUSED BY BAD SORT RESULTS

PROBLEM DESCRIPTION:
gpfs problems caused by bad sort results

PROBLEM SUMMARY:
Incorrect handling of sort results using
en_GB locale resulted in some nodes in an HACMP
environment not starting.

PROBLEM CONCLUSION:
Enforce LANG=C in GPFS to avoid problems
with sort results.

------

APAR: IY26551 COMPID: 5765B9501 REL: 320
ABSTRACT: IMPROVE PERFORMANCE OF RANDOM CONCURRANT READERS/WRITERS

PROBLEM DESCRIPTION:
improve performance of random concurrant readers/writers

PROBLEM SUMMARY:
performance improvement for randon I/O

------

APAR: IY26617 COMPID: 5765C3403 REL: 430
ABSTRACT: UNSUCCESSFUL_LOGIN_COUNT INCREMENTED TWICE IN LASTLOG

PROBLEM DESCRIPTION:
The unsuccessful_login_count value in /etc/security/lastlog is
incremented twice whenever loginfailed() is called due to
failure in the user login, instead of incrementing only once.
Due to this, the number of times user retries on login
failures will be incorrect. For example, if maxretries = 3 in
/etc/security/user, the user can unsuccessfully login 3 times
before being locked out. Now, with unsuccessful_login_count
incrementing twice, the user is locked out on 2 attempts.
This problem is seen on upgrade to bos.rte.libc 4.3.3.75 or
later (ML-09). This problem is seen with users locally
administered.

PROBLEM SUMMARY:
The unsuccessful_login_count value in /etc/security/lastlog
is incremented twice whenever loginfailed() is called due
to failure in the user login, instead of incrementing only
once. Due to this, the number of times user retries on login
failures will be incorrect. For example, if maxretries = 3
in /etc/security/user, the user can unsuccessfully login 3
times before being locked out. Now, with
unsuccessful_login_count incrementing twice, the user is
locked out on 2 attempts. This problem is seen on upgrade to
bos.rte.libc 4.3.3.75 or later (ML-09). This problem is seen
with users locally administered.

PROBLEM CONCLUSION:
loginfailed() function has a new if condition added to check
if the registry value is NOT NULL.

------

APAR: IY26771 COMPID: 5765E5400 REL: 440
ABSTRACT: IPAT FOR SP SWITCH FAILS ON NODE WITH NO BOOT ADDRESS CONFIGURED

PROBLEM DESCRIPTION:
The customer had a configuration with one node having IPAT
configured for his SP switch, but the takeover node did not
have IPAT configured for its switch so did not have a boot
address configured for the switch. After updating from
4.4.1.3 level of HAES to 4.4.1.4 level, a fallover to the
takeover node did not result in taking the failing node's
SP switch service address.

PROBLEM SUMMARY:
The customer had a configuration with one node having IPAT
configured for his SP switch, but the takeover node did not
have IPAT configured for its switch so did not have a boot
address configured for the switch. After updating from
4.4.1.3 level of HAES to 4.4.1.4 level, a fallover to the
takeover node did not result in taking the failing node's
SP switch service address.

PROBLEM CONCLUSION:
The acquire_takeover_addr script was changed to get the
network type from the address passed in to be taken over
rather than trying to get it later via a boot address which
may not exist.

------

APAR: IY26798 COMPID: 5765B9501 REL: 320
ABSTRACT: AIX "FIND" COMMAND FAILS TO FIND FILES IN GPFS FILE SYSTEM AS IT

PROBLEM DESCRIPTION:
The AIX "find" command does not find files in GPFS file system
as it does in NFS/JFS because GPFS ACLs are always enabled. The
"find" command assumes that it is dealing with JFS and will not
return the file name if it sees that there are ACLs.

LOCAL FIX:
The ACL bit will be turned off if the ACL consist only of the
three mandatory entries - user,group, and other.

PROBLEM SUMMARY:
The AIX "find" command does not find files in GPFS file
system as it does in NFS/JFS because GPFS ACLs are always
enabled. The "find" command assumes that it is dealing with
JFS and will not return the file name if it sees that there
are ACLS.

PROBLEM CONCLUSION:
The ACL bit will be turned off if the ACL consist only of
the three mandatory entries - user,group, and other.

------

APAR: IY26867 COMPID: 5765B9501 REL: 320
ABSTRACT: ASSERT ON STRUCTURE ERROR

PROBLEM DESCRIPTION:
assert on structure error

PROBLEM SUMMARY:
When SFSDoDeferredDeletions checks the inode map status to
decide whether to broadcast a "queryBeingCreated" message,
it must do so after acquiring the file lock, because the
inode map status could change while waiting for the lock.

PROBLEM CONCLUSION:
When SFSDoDeferredDeletions checks the inode map status to
decide whether to broadcast a "queryBeingCreated" message,
it must do so after acquiring the file lock, because the
inode map status could change while waiting for the lock.

------

APAR: IY26868 COMPID: 5765B9501 REL: 320
ABSTRACT: IMPROVE MULTITHREAD RANDOM IO PERFORMANCE

PROBLEM DESCRIPTION:
improve multithread random io performance

PROBLEM SUMMARY:
Random I/O performance improvement

PROBLEM CONCLUSION:
Drop vinfo lock during fetch for random reads/writes.
Improves parallelism when multiple threads share the same
file handle.

------

APAR: IY26869 COMPID: 5765D5100 REL: 320
ABSTRACT: CORSAIR: UTILITIES SHOULD CHECK THE ADAPTER TYPE

PROBLEM DESCRIPTION:
corsair: utilities should check the adapter type

PROBLEM SUMMARY:
The col_dump and cor_udump should check the adapter
type before reading from the card (ie: Today, I can do
col_dump on a Corsair node). Obviously, the data that
the command gives back is garbage, but it may confuse
our customers/service teams.

PROBLEM CONCLUSION:
The ddriver does provide an ioctl (QUERY_ADAPTER) that gives
a user adapter information which can be used to check
adapter type.

------

APAR: IY26874 COMPID: 5765D5100 REL: 330
ABSTRACT: A FIX TO CHECK THE STATUS OF THE PORT ASSOCIATED WITH DMA HANDLE

PROBLEM DESCRIPTION:
As part of the performance optimization, KLAPI wasallowed to
do a KHAL dma setup prior to when the KHAL dma handle was
required and later use that dma handle when needed. In this case
the DMA handle was obtained for KHAL port 1. sometime afterwards
the KHAL port 1 was closed, perhaps as a result of an adapter
faiure. Finally a zero copy rendezvous request was received on
b port 0. The KHAL dma handle for this request was no longer
valid since the port was closed and the associated structures
were cleaned up.

LOCAL FIX:
The fix is to check thestatus of the port associated with the
dma handle if the handle doesnot match the current handle.

PROBLEM SUMMARY:
When a single adapter on a double-single system fails while
using VSD/KLAPI, DMA structures associated with that adapter
are freed. However, the surviving adapter may still
reference these structures and cause the system to crash
when freed memory is assign to other kernel use, but is
still modified by KLAPI.

PROBLEM CONCLUSION:
The DMA data structures associated with one adapter used by
KLAPI will not be freed until the use of both adapters by
KLAPI has completed.

------

APAR: IY26888 COMPID: 5765C3403 REL: 430
ABSTRACT: MKCD WANTS LARGE CD_IMAGES SPACE WITH MULTIVOLUME BACKUP

PROBLEM DESCRIPTION:
mkcd wants large cd_images space with multivolume backup

PROBLEM SUMMARY:
When executing a mkcd backup that requires multiple
volumes, it would require that the temporary space
to hold the cd image (default /mkcd/cd_images) have
enough free space to hold all volumes even if the
user was burning the CDs and did not indicate to
keep the images. In this case, mkcd should only want
this space to be able to hold a single volume because
it removes the image after burning it and then
proceeds to process the next volume.

PROBLEM CONCLUSION:
Need to require this space only when the flags
indicating to save the cd images are used.

------

APAR: IY26891 COMPID: 5765C3403 REL: 430
ABSTRACT: DISPLAY OF IMBRICATE FUNCTIONS BY TYPESET -F IS INCORRECT

PROBLEM DESCRIPTION:
Imbricate functions are displayed incorrectly by 'typeset -f'
builtin.

PROBLEM SUMMARY:
Imbricate functions are displayed incorrectly by 'typeset -f'
builtin. The shell recognizes the beginning of the function,
but the display goes beyond the end of the function.

PROBLEM CONCLUSION:
'typeset -f' displays imbricate functions incorrectly.
Code changes are made so as to display imbricate functions
properly.

------

APAR: IY26900 COMPID: 5765B9501 REL: 320
ABSTRACT: ASSERT: !IS ACTIVE

PROBLEM DESCRIPTION:
assert: !isactive

PROBLEM SUMMARY:
Kernel assert !isActive when doing small random read after a
multi-block random read. Reset prefetch strategy to preNone
after large random operation completes.

PROBLEM CONCLUSION:
Kernel assert !isActive when doing small random read after a
multi-block random read. Reset prefetch strategy to preNone
after large random operation completes.

------

APAR: IY26914 COMPID: 5765C3403 REL: 430
ABSTRACT: CATIA CORE DUMPS W/ GXT2000P

PROBLEM DESCRIPTION:
Getting Catia abends that are only happening with 2000p
graphics adapter. A core file is created. Problem is
very reproducible with a specific Catia Model with
GXT2000P only.

PROBLEM SUMMARY:
Getting Catia abends that are only happening with 2000p
graphics adapter. A core file is created. Problem is
very reproducible with a specific Catia Model with
GXT2000P only.

PROBLEM CONCLUSION:
Correct logic error in the GXT2000P graPHIGS raster logic
releted to the breakup of a special line with many vertices.
A calculation with unsigned integers is corrected to avoid
an overflow.

------

APAR: IY26917 COMPID: 5765C3403 REL: 430
ABSTRACT: OSLEVEL CACHE FILE CAN ME CORRUPTED BY USERS

PROBLEM DESCRIPTION:
oslevel cache file can me corrupted by users

LOCAL FIX:
run "oslevel -f" to rebuild the cache

PROBLEM SUMMARY:
Implemented better and more secure locking mechanism for
oslevel.

PROBLEM CONCLUSION:
Implemented better locking mechanism for oslevel.

------

APAR: IY26918 COMPID: 5765B9501 REL: 320
ABSTRACT: MANUALLY UNMOUNTING AUTOMOUNTED GPFS DIR HANGS AUTOFS

PROBLEM DESCRIPTION:
umount an automounted GPFS filesystem manually hangs. The mount
commands shows the filesystem no longer mounted. mmfsmount and
mmfsmnthelper are hung.

PROBLEM SUMMARY:
A manual unmount of a file system mounted using the
automounter hangs.

PROBLEM CONCLUSION:
Remove unnecessary VNOP_RELE of underlying mountpoint when
releasing the last hold on the root vnode. If autofs was the
fstype of the underlying mountpoint, the vnode lock would
hang the thread since the same thread had locked it ub
uvmount before calling GPFS.

------

APAR: IY26930 COMPID: 5765C3403 REL: 430
ABSTRACT: /ETC/INITTAB FILE OWNERSHIP IS NOT PRESERVED

PROBLEM DESCRIPTION:
The ownership of the /etc/inittab file is not preserved
when there is more than one root user with different
group IDs.

PROBLEM SUMMARY:
penable, pdisable, pstart, phold, pdelay, pshare, chitab,
rmitab, and mkitab commands do not ensure that the group
ownership of the /etc/inittab is preserved. If these
applications are called by a user with required privilege,
the group ownership of the /etc/inittab file will be
changed to that users primary group.

PROBLEM CONCLUSION:
The code of the commands penable, pdisable, pstart, phold,
pdelay, pshare, chitab, rmitab, and mkitab is changed to
preserve the group ownership of the /etc/inittab file.

------

APAR: IY26941 COMPID: 5765B9501 REL: 320
ABSTRACT: ASSERT IN BRC.C LINE 811

PROBLEM DESCRIPTION:
assert in brc.c line 811

PROBLEM SUMMARY:
GPFS self check logic detected an invalid condition during
GPFS termination on a node reporting an error at
BRC.C line 811

PROBLEM CONCLUSION:
brLock not setting err variable if daemon already shutting
down, so may accidentally return E_OK resulting in unusual
results that think they have the byte-range lock. Assert in
brUnlock resulted when it tried to use an garbage lHandle
variable as a pointer.

------

APAR: IY26951 COMPID: 5765B9501 REL: 320
ABSTRACT: HEAVY I/O RUN HUNG WITH LONG WAITERS

PROBLEM DESCRIPTION:
heavy i/o run hung withlong waiters

PROBLEM SUMMARY:
Deadlock under extreme I/O loads running mmchmgr.

PROBLEM CONCLUSION:
Deadlock in token acquire/revoke where the sequence number
on the acquire request to the TM is equal to the old
sequence number. If the request gets held up by another
acquire in progress, and that request send a revoke to the
first node, the revoke will hang in deadlock with the
acquire.

------

APAR: IY27038 COMPID: 5765C3403 REL: 433
ABSTRACT: ADD IBM TUNNEL SUPPORT OVER ALIAS IP ADDR

PROBLEM DESCRIPTION:
IBM tunnel doesn't work over alias IP addresses.

PROBLEM CONCLUSION:
Add the alias IP address support in IBM tunnel daemon (skeyd).

------

APAR: IY27077 COMPID: 5765C3403 REL: 430
ABSTRACT: CRFS AND CHFS LOG AND LOGNAME INCONSISTENT

PROBLEM DESCRIPTION:
crfs and chfs are inconsisent in the use of log and logname.
Both crfs and chfs should accept both log and logname to
specify the log device.

LOCAL FIX:
Use the regular way with "-a logname"

PROBLEM CONCLUSION:
crfs will now accept both log and logname as valid attributes.
LOG_str2 has been defined in the header file as "log", to
check when parsing for attribute value.

------

APAR: IY27116 COMPID: 5765E6100 REL: 110
ABSTRACT: CHILD PROCESS OF IBM.SERVICERMDGOES DEFUNCT

PROBLEM DESCRIPTION:
On AIX-5.1.C 32 bit one child process of IBM.ServiceRMd gets a d
efunct process

PROBLEM SUMMARY:
ServiceRMd was forking and execing diagService
Event, but upon termination the process was not being cleaned
up. It was thus left as a defunct process in the process
table. The child process will not be cleaned up until its
termination signal is processed.

PROBLEM CONCLUSION:
The code has been modified to process the
child termination signal. This results in the process being
cleaned up, which includes removing it from the process table.

------

APAR: IY27136 COMPID: 5765D5100 REL: 320
ABSTRACT: ABENDING US JOB LEADS TO BOOKKEEPING PROBLEMS

PROBLEM DESCRIPTION:
abending us job leads to bookkeeping problems

PROBLEM SUMMARY:
memory use not being tracked correctly in abnormal
termination cases

PROBLEM CONCLUSION:
correct memory use tracking

------

APAR: IY27175 COMPID: 5765C3403 REL: 430
ABSTRACT: FLOPPY DRIVE CONFIG METHOD IMPROVEMENT

PROBLEM DESCRIPTION:
The special files for /dev/fd0l and /dev/fd0h may have
different major numbers than other floppy drive special files.

PROBLEM CONCLUSION:
At configuration time remove the following special files
before linking them.
/dev/fd0l
/dev/fd0h

------

APAR: IY27215 COMPID: 5765C3403 REL: 430
ABSTRACT: WEBSM FAILS TO CORRECTLY DISPLAY LATEST FILESET INSTALLED

PROBLEM DESCRIPTION:
When displaying installed software using websm, applied updates
with a fix level of two digits for which the first digit is a
number that is lower than other applied or committed levels
will cause the wrong fileset to be listed as the highest level
installed.

LOCAL FIX:
Use lslpp to display information about installed software or
collapse the software trees in websm to reveal all the
different levels for a particular fileset. When the problem is
present the fileset with the highest modification level will
show after collapsing the tree.

PROBLEM SUMMARY:
The function inst_query_utils is not sorting filesets
by their correct level (v.r.m.f). This problem causes
the highest level of a fileset to not show as such.

PROBLEM CONCLUSION:
After changing the sorting algorithm on inst_query_utils,
now the highest level of a fileset will be displayed first,
always.

TEMPORARY FIX:
Use lslpp to list filesets.

------

APAR: IY27219 COMPID: 5765E5400 REL: 440
ABSTRACT: SMIT CHANGE_SHOW_TS_GS DOES NOT WORK - HAES

PROBLEM DESCRIPTION:
The customer attempt to use the smit change_show_ts_gs menu
to change the size of the topology services or group services
log files and the command returned error message:
odmchange: invalid numeric value, stanza line: 2

PROBLEM CONCLUSION:
The heartbeat interval and fibrillate count were previously
removed from the smit panel, but the cmd_to_exec had not
been changed to disregard these fields. The cmd_to_exec
for this was now changed to remove these fields.

------

APAR: IY27298 COMPID: 5765C3403 REL: 430
ABSTRACT: XSERVER HANG WITH GXT4000P/GXT2000P ON LOTUS NOTES DATABASE

PROBLEM DESCRIPTION:
Xserver hang problem with GXT4000P. This happens while opening
a lotus notes database within Citrix Metaframe and a particular
lotus notes database. Customer also experiences an X Server
Hang on GXT200P with Citrix Metaframe and lotus notes, while
scrolling windows in lotus notes database. This problem is
specific to the GXT4000P, GXT2000P and accessing Citrix via RSH
and the Citrix Unix Integration package for Windows 2000.

PROBLEM SUMMARY:
Xserver hang problem with GXT4000P. This happens while opening
a lotus notes database within Citrix Metaframe and a particular
lotus notes database. Customer also experiences an X Server
Hang on GXT200P with Citrix Metaframe and lotus notes, while
scrolling windows in lotus notes database. This problem is
specific to the GXT4000P, GXT2000P and accessing Citrix via RSH
and the Citrix Unix Integration package for Windows 2000.

PROBLEM CONCLUSION:
Check for a pixmap image whose width of length exceeds that
supported by the GXT2000P/GXT4000P/GXT6000P adapters, and if
exceeded, use the machine independent path of the X Server
to break up the pixmap and process it in tiles.

------

APAR: IY27313 COMPID: 5765D6100 REL: 220
ABSTRACT: BAD LLQ PERFORMANCE W/ 6000+ JOBS IN QUEUE

PROBLEM DESCRIPTION:
llq is takeing close to a minute to run when it is run alone
with around 6000 jobs in the queue. It is not uncommon for
llq's to timeout when a couple invocations are made in a short
space of time.

PROBLEM SUMMARY:
LoadL 2.2 takes approximately 3 times as long to produce llq
output
when there are 6000 jobs in the queue as LoadL 1.3 did.

PROBLEM CONCLUSION:
LoadL 2.2 needs a fast path in the job query code for plain
llq's
that will skip data manipulation which is only needed for
more
detailed output. This fix in itself will not make the
performance
equal to LoadL 1.3, but will make the performance resonably
close.
Note1: because of a major design difference between 1.3 and
2.2,
  the 2.2 time from when llq is invoked until the first job
is
  displayed on the screen is still significantly longer than
it was
  on 1.3, but the last job output is reasonably similar.
Note2: only a plain llq will be enhanced. llq performance
with any
  options will remain unchanged.

------

APAR: IY27331 COMPID: 5765E5400 REL: 440
ABSTRACT: HAES: SMITTY SWAP ADAPTER AFT IPAT (NO STBY) GIVES PING REQUEST

PROBLEM DESCRIPTION:
Execute smit swap adapter with 2 pairs of svc/standby to swap
gives a "ping request timed out" error if an IPAT has occurred
and there exists 2 service addresses.

PROBLEM CONCLUSION:
Remove faulty algorithim that determines the local cluster
node and replace with the correct one.

------

APAR: IY27332 COMPID: 5765E5400 REL: 440
ABSTRACT: HAES:AEN ERROR NOTIFY METHOD DOES NOT RUN AFTER ERROR EMULATION

PROBLEM DESCRIPTION:
AEN: Error notification method does not run after error
emulation. AEN methods are generated in a cluster by running
the SMIT sequence RAS Support- Error Notification-
Configure Automatic Error Notification-
Add Error Notify Methods for Cluster Resources and verified
with odmget errnotify.
However, running an emulated error against a resource in a
notification method does not result in that method executing.
The emulated error does appear in the system's errpt.

PROBLEM CONCLUSION:
The difference in the errnotify structure between AIX 4.x and
AIX 5.x causes a problem. New fields have been inserted into
the errnotify structure in AIX 5.x. As a result, code compiled
under AIX 4.3 causes the wrong fields to be filled in. The
solution is to use a temporary file to create the ODM, and
use system(odmadd <temp_file>). This will ensure any extra
fields will be taken care of by AIX.

------

APAR: IY27337 COMPID: 5765C3403 REL: 430
ABSTRACT: FIND RETURNS INCORRECT RESULTS WITH -XDEV -DEPTH

PROBLEM DESCRIPTION:
"find / -xdev -depth -ls", produces an empty file
as the last entry in the output.

PROBLEM SUMMARY:
find / -xdev -depth -ls -print
displays the last line of the output without file name.

PROBLEM CONCLUSION:
Proper code changes were made to fix the problem.

------

APAR: IY27351 COMPID: 5765D5100 REL: 320
ABSTRACT: UPDAUTHFILES RECREATES /DEV/NULL AS NORMAL CHAR FILE

PROBLEM DESCRIPTION:
In PTF set 15, APAR IY20005 (PTF U481355) introduced a bug in
/usr/lpp/ssp/bin/updauthfiles. A typo caused output redirection
to /dev/null to instead be a deletion & recreation of /dev/null
as a normal file. The lines involved were:
line 571: `/bin/rm $SP_RHOSTS 2>&1 /dev/null`;
line 1157: `/bin/rm $SP_K5LOGIN 2>&1 /dev/null`;
(Note the redirect missing before /dev/null.)
updauthfiles is run from spsetauth, spdelnode, spauthconfig and
setup_CWS. It will run the offending lines if the .spgen_rhosts
or .spgen_k5login files already exist in root's home dir
(usually "/"). These files should exist as long as a .rhosts or
.k5login file is in place (again, in root's home dir).
The result will be a /dev/null file that looks like this:
-rw-r--r-- 1 root system 234 Jan 18 13:53 /dev/null
This can result in failed commands if the noclobber option is
set (/dev/null will not be overwritten) or the filling up of
/dev if large output is redirected to /dev/null.

LOCAL FIX:
1) In updauthfiles, change the lines to be:
   line 571: `/bin/rm $SP_RHOSTS 2>&1 >/dev/null`;
   line 1157: `/bin/rm $SP_K5LOGIN 2>&1 >/dev/null`;
2) If the bad /dev/null file has already been created, do:
   rm /dev/null;mknod /dev/null c 2 2
   Those commands are stacked because mknod will fail if
   /dev/null already exists, and a bad /dev/null will be
   recreated every time anything on the system does a redirect
   to /dev/null while the special file does not exist
The good /dev/null file should look like this:
crw-r--r-- 1 root system 2, 2 Jan 18 13:55 /dev/null

PROBLEM SUMMARY:
***********************************************************
* USERS AFFECTED: Users with ssp.basic 3.2.0.15 *
***********************************************************
* PROBLEM DESCRIPTION: *
* Running /usr/lpp/ssp/bin/updauthfiles may change *
* /dev/null from a special character file to a normal *
* data file. *
***********************************************************
* RECOMMENDATION: *
* Install APAR IY27351, currently targeted for *
* ssp.basic 3.2.0.16 on PTF Set 16, when available. *
* *
* Until APAR IY27351 is available, after applying U481355 *
* you should edit the following lines of the updauthfiles *
* script: *
* *
* line 571: `/bin/rm $SP_RHOSTS 2>&1 /dev/null`; *
* to: `/bin/rm $SP_RHOSTS 2>&1 >/dev/null`; *
* *
* line 1157: `/bin/rm $SP_K5LOGIN 2>&1 /dev/null`; *
* to: `/bin/rm $SP_K5LOGIN 2>&1 >/dev/null`; *
* *
* (Note the ">" being added before /dev/null) *
* *
* If /dev/null has been affected, restore it using: *
* rm /dev/null;mknod /dev/null c 2 2 *
* chmod 666 /dev/null *
***********************************************************

------

APAR: IY27365 COMPID: 5765E8301 REL: 320
ABSTRACT: DCED CORE DUMP IN VFPRINTF

PROBLEM DESCRIPTION:
On Solaris, dced is core dumping with the following stack:
current thread: t61
=> 1 strlen() at 0xff136f54
   2 _doprnt() at 0xff1818f8
   3 vfprintf() at 0xff183720
   4 dce_svc__vfprintf() at 0xfeeb9f30
   5 dce_svc_printf_withfile() at 0xfeeba 8e0
   6 rca_site_resolve_update() at 0xfee48580
   7 _dce_exJACrdTWT() at 0xfee4a268
   8 _dce_RBcVlAWouE() at 0x77abc
   9 send_as_request() at 0xfed8601c
   10 _dce_aieYQgADmG() at 0xfed869c4
   11 _dce_FjYIdZiHTt() at 0xfee9c6fc
   12 sec_login_validate_identity() at 0xfee8c384
   13 sec_login_valid_from_keytable() at 0xfee8cde8
   14 keymgmt_handler() at 0x30fa8
   15 dce_pthread_start() at 0xfee276d0

PROBLEM SUMMARY:
On Solaris, dced is core dumping with the following stack:
(l61) terminated by signal SEGV (no mapping at the
                                  fault address)
(/opt/SUNWspro/bin/dbx) where
current thread: t61
=> 1 strlen(0x0, 0x0, 0x0, 0x7efefeff, 0x81010100,
         0xdb72c), at 0xff136f54
   2 _doprnt(0x39ee796, 0x0, 0x0, 0x0, 0x0, 0x0),
         at 0xff1818f8
   3 vfprintf(0xff1b9d24, 0x39ee778, 0xfdcf0620,
         0xff1bd5d4, 0xff1b9708, 0xff02d17c), at 0xff183720
   4 dce_svc__vfprintf(0x0, 0xa, 0x70, 0xfdcf0620,
         0xfef8770c, 0xff1b9d24), at 0xfeeb9f30
   5 dce_svc_printf_withfile(0x1, 0x84, 0x1, 0xff045904,
         0x1, 0x40), at 0xfeeba 8e0
   6 rca_site_resolve_update(0x3a26d70, 0x11ed18,
         0xfdcf0e80, 0xfdcf0f88, 0x0, 0x3507130),
         at 0xfee48580
   7 _dce_exJACrdTWT(0xfdcf0f88, 0x10d, 0xd0a550, 0x2000,
         0xfdcf0ffc, 0x3b14b78 ), at 0xfee4a268
   8 _dce_RBcVlAWouE(0x38a3120, 0xb90a8, 0x3a40930,
         0xfdcf10dc, 0xfee96e30, 0xf dcf13f0), at 0x77abc
   9 send_as_request(0xfdcf10dc, 0xfdcf10dc, 0x6a, 0x0,
         0x3bb94d4d, 0xfdcf10d4) , at 0xfed8601c
   10 _dce_aieYQgADmG(0x3a1ae44, 0xfdcf10dc, 0x0, 0x80ad4,
         0x39510c0, 0x3a1ae44 ), at 0xfed869c4
   11 _dce_FjYIdZiHTt(0x3ab41f0, 0x37616e0, 0x37616e0,
         0xfdcf1be8, 0x3ab41f0, 0 x0), at 0xfee9c6fc
   12 sec_login_validate_identity(0x3ab41f0, 0x3ab41f0,
         0x3ab41f0, 0x0, 0x0, 0x fdcf1bd4), at 0xfee8c384
   13 sec_login_valid_from_keytable(0x37616e0, 0x3ab41f0,
         0xfdcf1be8, 0xfdcf1be 8, 0xfdcf1cf4, 0xfdcf1bd8),
         at 0xfee8cde8
   14 keymgmt_handler(0xc11b0, 0x0, 0x6, 0xff24c9ac, 0x4,
         0x1fce40), at 0x30fa8
   15 dce_pthread_start(0x1fce40, 0xfefc2274, 0xfef8770c,
         0x1fce40, 0x1, 0xff24 c9ac), at 0xfee276d0

PROBLEM CONCLUSION:
This is a problem with NULL string pointers being passed
into vfprintf. Apparently AIX is more forgiving about
this than Solaris.
Within rca_site_resolve_update():
...
if (BAD_STATUS(status)) {
rpc_binding_to_string_binding(tmp_rpch, &rpch_string, &xst);
dce_svc_printf(SEC_RCA_OP_STATUS_MSG ,
                rpch_string,
                msgp=dce_sprintf(*status));
...
Both rpc_binding_to_string_binding and dce_sprintf involve
memory allocation to which rpch_string and msgp will point.
However, if for some reason that fails, or some other error
within rpc_binding_to_string_binding leaves
rpch_string NULL, we'll end up calling dce_svc_printf with
a NULL string pointer (which eventually cores in strlen).
The fix is to pass in a null string if either of these
pointers is NULL:
  msgp=dce_sprintf(*status);
  dce_svc_printf(SEC_RCA_OP_STATUS_MSG ,
  (rpch_string!=NULL)?rpch_string:(unsigned char *)"",
  (msgp!=NULL)?msgp:(unsigned char *)"");

------

APAR: IY27367 COMPID: 5765E8300 REL: 320
ABSTRACT: LDAP: CREATE THREAD TO UNBIND LDAP HANDLES WHEN OUT OF USE

PROBLEM DESCRIPTION:
If an LDAP_SERVER_DOWN message is received, it causes
SEGV's within the LDAP client library.

PROBLEM CONCLUSION:
pthread is createad to our new unbind thread.

------

APAR: IY27368 COMPID: 5765E8301 REL: 320
ABSTRACT: LDAP:RGY_LDAP_GET_DELETETYPE SEARCH KRBREALM INSTEAD DCEREALM

PROBLEM DESCRIPTION:
The ldapdeletetype for the ldapized security server may not
be set during initial migration.

PROBLEM CONCLUSION:
Look for the correct objectclass for modifying the
attribute.

------

APAR: IY27371 COMPID: 5765E8301 REL: 320
ABSTRACT: SVT, LDAP: LDAP_SERVER_DOWN ERROR DURING LDAP_INIT_ATTR_CURSOR

PROBLEM DESCRIPTION:
Customer will see multiple lines with the following error
text in the DCE error log (var/svc/error.log)
LDAP_SERVER_DOWN during ldap_init_attr_cursor

PROBLEM CONCLUSION:
Fixed a problem in the security server code that was doing
a re-connect to the LDAP server.

------

APAR: IY27380 COMPID: 5765E8300 REL: 320
ABSTRACT: LDAP:ERROR LOST WHEN NO LDAP HANDLES AVAILABLE

PROBLEM DESCRIPTION:
An invalid errno is written to the error log when trying
to log an LDAP error from an LDAP function call.

PROBLEM CONCLUSION:
Called the logging routine with the correct errno

------

APAR: IY27386 COMPID: 5765E8301 REL: 320
ABSTRACT: CONFIG FAILS WHEN THERE IS NO /ETC/ENVIRONMENT FILE

PROBLEM DESCRIPTION:
config.dce will fail on a full or local configuration if
there is not /etc/environment (AIX) or /etc/default/init
(Solaris) file.

PROBLEM CONCLUSION:
The routine that adds DCE environment variables will create
this file if it doesn't exist, or append entries to it if
it does exist.

------

APAR: IY27449 COMPID: 5765C3403 REL: 430
ABSTRACT: INCORRECT AIXTERM FONT ALIAS

PROBLEM DESCRIPTION:
Some entries in the font.alias file are incorrect.

PROBLEM SUMMARY:
when trying to use ibm-1252 reduced fonts, i.e. extra small
and extra extra small ibm-1252 fonts, an error message will
be displayed

PROBLEM CONCLUSION:
set the extra small(xs) and extra extra small(xxs) ibm-1252
font alias to existing fonts

------

APAR: IY27452 COMPID: 5765C3403 REL: 430
ABSTRACT: ERROR WHILE CHANGING THE PASSWORD

PROBLEM DESCRIPTION:
Users will not be able to change passwords either at
login time or with the 'passwd' command. The error that
occurs is '3004-619 Security method <some_message_or_blank>
could not be loaded'.

PROBLEM CONCLUSION:
The memory allocation from a call to get the registry or other
attributes is done in a cache which is de-allocated after a
call to enduserdb(). Previously, some of the allocations
were not cached and not deallocated explicilty resulting in
a memory leak.

------

APAR: IY27480 COMPID: 5765C3403 REL: 430
ABSTRACT: CUT COMMAND IGNORES LINES WITHOUT NEWLINE CHARACTERS

PROBLEM DESCRIPTION:
Lines without Newline characters are ignored when the cut
command processes its input.

LOCAL FIX:
Lines without Newline characters are ignored when the cut
command processes its input.

PROBLEM CONCLUSION:
The cut command is changed to processes lines that don't
end with a newline character also.

------

APAR: IY27481 COMPID: 5765C3403 REL: 430
ABSTRACT: FIND GETS A SEGMENTATION VIOLATION ON A LARGE DIRECTORY

PROBLEM DESCRIPTION:
find command is not able to descend to arbitrary depths in a
file hierarchy and dumps core if the total path length exceeds
PATH_MAX.

PROBLEM CONCLUSION:
The find command is modified to allocate memory dynamically
for the path variable and is made to decend to any depths in
the file hierarchy.

------

APAR: IY27491 COMPID: 5765C3403 REL: 430
ABSTRACT: GXT2000P/GXT3000P GRAPHIGS APP GETS WARNING: NEED TO CODE COLOR

PROBLEM DESCRIPTION:
Catia application recieves the following message on GXT2000P or
GXT3000P: WARNING: Need to code color compare enable clean up
NOTE that there is no functional impact to the application, it
operates correctly. This message is mearly an annoyance.

PROBLEM SUMMARY:
Catia application recieves the following message on GXT2000P or
GXT3000P: WARNING: Need to code color compare enable clean up
NOTE that there is no functional impact to the application, it
operates correctly. This message is mearly an annoyance.

PROBLEM CONCLUSION:
This message was used during development, and should have been
removed. Remove this message.

------

APAR: IY27497 COMPID: 5765B9501 REL: 320
ABSTRACT: EXECUTABLE COMPLETELY MEMORY MAPPED WON'E BE TERMINATED ON UNMOU

PROBLEM DESCRIPTION:
If a program is completely mapped in memory and the filesystem
unmounts, when the filesystem comes back and you try to run the
program again it dies with Bus error (coredump).

LOCAL FIX:
make a copy of the executable and run that instead

PROBLEM SUMMARY:
A program which is mapped from GPFS core dumps when trying
to reexecute the program after a forced file system unmount
while it was mapped.

PROBLEM CONCLUSION:
Invalidate memory mapped pages after remount.

------

APAR: IY27505 COMPID: 5765B9501 REL: 320
ABSTRACT: GPFS NOATIME IS NOT SAVED IN SDR AND /ETC/FILESYSTEMS

PROBLEM DESCRIPTION:
gpfs noatime is not saved in sdr and /etc/filesystems

PROBLEM SUMMARY:
some options on mmchfs not being correctly
passed to the mount command.

PROBLEM CONCLUSION:
parse the options string from getefoptions
first, then parse the options string passed in on the mount
command for possible overrides.

------

APAR: IY27506 COMPID: 5765C3403 REL: 430
ABSTRACT: BI_MAIN NEED TO HAVE RMINSTAL RUN WHEN CLONING

PROBLEM DESCRIPTION:
Users cloning or reinstalling a mksysb during which newer
versions of installed device drivers are being installed
during the Base Installation, can have duplicate entries
added to the Pd* databases, resulting in a machine that fails
to reboot.

LOCAL FIX:
Use a hand-edited bi_main to turn off INUBOSTYPE.

PROBLEM CONCLUSION:
When installing from mksysb media, turn off the variable,
INUBOSTYPE, that causes rminstal not to run during fileset
installation.

------

APAR: IY27536 COMPID: 5765C3403 REL: 430
ABSTRACT: MSMIT LOCKS UP IF YOU CANCEL FROM A CONFIRMATION DIALOG

PROBLEM DESCRIPTION:
In msmit only, if you can cancel from a confirmation dialog
that came from a ghost cmd_hdr, msmit will lock up.

LOCAL FIX:
Use ascii smit (smitty) instead

PROBLEM SUMMARY:
In msmit only, if you cancel from a confirmation dialog
that came from a ghost cmd_hdr, msmit will lock up.

PROBLEM CONCLUSION:
Cleanup dialogs from ghost cmd_hdrs.
Use smitty or ascii smit instead.

TEMPORARY FIX:
Use ascii smit

------

APAR: IY27552 COMPID: 5765D5101 REL: 121
ABSTRACT: 128WAY:HATS_NIM CORE DUMPED

PROBLEM DESCRIPTION:
128way: hats_nim core dumped

PROBLEM SUMMARY:
The RSCT Topology Services daemon uses auxiliary
processes, called "Network Interface Modules" (NIMs), to
monitor the state of each local network interface. The
communication between the daemon and the NIMs is done via
a Unix-Domain socket (UDS) connection.
There have been cases where the UDS connection was broken
and only the daemon noticed the breakage. This led to the
daemon attempting to dispatch a new instance of the NIM
process, but the attempt was not successful since the
previous instance was still running. The end result was
the daemon declaring the corresponding adapter as down.
The previous instance of the NIM would eventually
terminate with an assert() statement and create a core
file.
Because the local adapter was declared to be down, some
applications like GPFS may terminate on the node.

PROBLEM CONCLUSION:
Fixes were introduced into both the RSCT Topology Services
daemon and the Network Interface Module (NIM). The fixes
should prevent the NIM from not detecting a broken
Unix-Domain Socket (UDS) connection. The fix also extends
the time that the daemon allows for the NIM to terminate
before declaring the local adapter as down.
With the fix, if the UDS connection of the daemon with the
NIM breaks then the NIM process should terminate quickly,
thus allowing the daemon to start a new instance. The
local adapter will then not be declared as down.

------

APAR: IY27555 COMPID: 5765E6110 REL: 220
ABSTRACT: 128WAY:HATS_NIM CORE DUMPED

PROBLEM DESCRIPTION:
128way: hats_nim core dumped

------

APAR: IY27556 COMPID: 5765E6110 REL: 220
ABSTRACT: EXTRANEOUS MESSAGE FROM HAGS ON TERMINATION

PROBLEM DESCRIPTION:
extraneous messge from hags on termination

PROBLEM SUMMARY:
When read() call fails, Group Services library
prints messages to stderr, which indicating the failure.
However, if the client uses multiple HAGS dispatch
threads and one other thread terminates HAGS connection
using (ha_gs_quit), those dispatch threads will
see the read() failures because the socket is now closed.
With this, the library will check whether the socket
is closed or not before it prints the messages.
Thus, the spuriously messages will not be printed.

PROBLEM CONCLUSION:
Group Services library spuriously prints out some messages
even after the clients terminates HAGS connection.
It does not necessarily indicate the problem, but
confuse the users. So, the fix will eliminate
those confusing messages.

------

APAR: IY27559 COMPID: 5765E6110 REL: 220
ABSTRACT: LSRESPONSE -A TRUNCATES ACTIONSCRIPT

PROBLEM DESCRIPTION:
lsresponse -a truncates actionscript

PROBLEM SUMMARY:
The lsresponse command incorrectly displays the RSCT
monitoring response definition if the action script (the -s
flag operand) contains a colon. A typical scenario would be
to use the mkresponse command to define a monitoring
response. When defining the action script for a response
action, a colon is used in the -s flag operand text. Then,
when attempting to display this response definition using
lsresponse, incorrect data is displayed.

PROBLEM CONCLUSION:
The lsresponse command was slightly modified to parse the
response definition data differently. When the definition
was retrieved from RMC, a colon was used as a field
separator. This is what caused the incorrect output. Now, a
more random string is used as a field separator. The
parsing of the display fields yields the correct results.

------

APAR: IY27565 COMPID: 5765E6110 REL: 220
ABSTRACT: NAS:HA_GS_INIT() LEAKS THE FILE DESCRIPTORS

PROBLEM DESCRIPTION:
nas:ha_gs_init() leaks the file descriptors

PROBLEM SUMMARY:
Group Services library contains a debug statement
debug(3, "read_sock(%d,%d) ==> (rc=%d, errno=%d).
%s \n",
fd, size-bytes_rcvd, rc, errcode, msgbuf);
which may print unrecognizable messages in certain cases,
where msgbuf is not completed filled due to rc==-1.
The fix must check rc whether msgbuf is valid or not.

PROBLEM CONCLUSION:
The unexpected debug messages from Group Services API
will be prevented.

------

APAR: IY27566 COMPID: 5765E6110 REL: 220
ABSTRACT: CM&CSSMEMBERSHIP GAVE WRONG STATUS

PROBLEM DESCRIPTION:
cm & cssmembership gave wrong status

PROBLEM SUMMARY:
There could be two reasons of the problem;
1) Due to a configuration problem, somehow
  the minor numbers of css0 and css1 are swapped
  after the adapter move.
2) If an IP address of a switch device is
  changed to the other IP address.
Both cases, Group Services Switch support daemon
(HAGSGLSM) does not properly report the status
of the switch memberships because
a) ioctl(,POLL_SW_STAT) does not give the
  correct value for the case 1), or
b) HAGSGLSM does not correctly update the
  change of IP address for the case 2).
The fix will resolve those situations.

PROBLEM CONCLUSION:
Even if a switch device (css0 or css1) is reconfigured
for a new IP address or a different minor number while
Group Services is running, the status of switch memberships
(e.g., cssMembership, css1Membership, or ml0Membership)
should be correct as to what Topology Services tells.

------

APAR: IY27572 COMPID: 5765C3403 REL: 430
ABSTRACT: ADD LINK POLLING FUNCTIONALITY.

PROBLEM DESCRIPTION:
10/100 Mbps PCI Ethernet Adapter II does not provide
support for Etherchannel and Network Interface backup.

PROBLEM CONCLUSION:
Introduced code in the driver to allow these features to
be used with this adapter.

------

APAR: IY27600 COMPID: 5765C3403 REL: 430
ABSTRACT: NETGROUP ROOT MNT OPTION DOES NOT FUNCTION

PROBLEM DESCRIPTION:
Netgroups specified in root option for exports are not
used.

PROBLEM SUMMARY:
User is unable to use netgroups in root export option. User
can add them to the root export option, but this will not
function properly.

PROBLEM CONCLUSION:
Modification of mountd was made to parse netgroups from
/etc/xtab.

------

APAR: IY27614 COMPID: 5765C3403 REL: 430
ABSTRACT: COREDUMP IN WAITPID() CALLED FROM SIGNALHANDLER

PROBLEM DESCRIPTION:
If a signal is received while a pthread is detached from its
kernel thread than the errno value of that pthread will
not be accessible. This will cause an access of a NULL pointer
and as a results will coredump.Even if the program is
multithreaded there could be instances where the pthreads
information is not accessible.

PROBLEM CONCLUSION:
Check the value of Get_Error_Ref which is the pointer to the
pthread's errno. If NULL because pthread is
unavailable then return the global errno otherwise
return the pthread's errno value.

------

APAR: IY27619 COMPID: 5765C3403 REL: 430
ABSTRACT: MKCD TO CD FAILS IF LARGER THAN 2 GB

PROBLEM DESCRIPTION:
Doesn't allow the customer to create a backup image using
mkcd greater than 2GB.

PROBLEM SUMMARY:
When using mkcd to create backups, if the
backup will be larger than 2 GB, it will
fail if attempting to do this to CDs
because it does not create its temporary
file system /mkcd/mksysb_image as a large
file enabled filesystem.

PROBLEM CONCLUSION:
Need to have mkcd create file system as a
large file enabled file system when the
backup will be larger than 2 GB.

TEMPORARY FIX:
Manually create this temporary file system
and create it as a large file enabled
file system.

------

APAR: IY27668 COMPID: 5765E6110 REL: 220
ABSTRACT: CT: AUTHENTICATION OK FOR INCORRECT CONTECT

PROBLEM DESCRIPTION:
ct: authentication ok for incorrect context

PROBLEM SUMMARY:
security issue

------

APAR: IY27689 COMPID: 5765C3403 REL: 430
ABSTRACT: POLL LINK EVEN IF IN DISABLE_STATE.

PROBLEM DESCRIPTION:
The driver does not log any errors in a netif_backup
setup, if the adapter is the backup adapter.

PROBLEM CONCLUSION:
Removed code in the driver that was not polling the adapter
for link status if it was in the DISABLE_STATE.

------

APAR: IY27694 COMPID: 5765C3403 REL: 433
ABSTRACT: XTAPPADDINPUT NOT WORKING CORRECTLY WITH XTAPPPEEKEVENT

PROBLEM DESCRIPTION:
Applications using XtAppPeekEvent can hang, or pause, or
just not process data as expected with XtAppAddInput.

PROBLEM CONCLUSION:
Merge in some code changes from the latest X.Org source
(X11R6.6).

------

APAR: IY27703 COMPID: 5765C3403 REL: 430
ABSTRACT: CRON FILLS /VAR FILESYSTEM WHEN THE NICE VALUE FOR BATCH JOBS

PROBLEM DESCRIPTION:
cron fills the /var filesystem, due to the continous
error messages addded to the cron log file, when the
nice value for batch jobs is given as 2.

PROBLEM SUMMARY:
cron logs the same error message continously
into the log file which fills the /var file
system, when the nice value for batch jobs is
given as 2 in the queue definition.

PROBLEM CONCLUSION:
The cron code analysed and proper code
changes made to resolve the problem.

------

APAR: IY27704 COMPID: 5765E6110 REL: 220
ABSTRACT: 2523-055 NODE NUMBER DUPLICATED: 0 LANG =

PROBLEM DESCRIPTION:
Hats didn't start and following error messages were logged.
read_machine hatsd 0 : 2523-055 Node number duplicated:0 LANG =
read_machines hatsd 0 : 2523-056 Gethostbyname = Failed,errno= 1
This problem had occurred for a customer who installed the AIX
as Japanese (LANG=Ja_JP) and /etc/environment had
"LC_MESSAGES=Clft". If it was defined, perl command display
warning message and change the locale to C, But the messages
were put into hats.machines.lst and caused this problem.

PROBLEM SUMMARY:
At rsct level 2.2.0.20 and AIX 5.1.0.15,
hats will not start when /etc/environment
contains the following statement:
LC_MESSAGES=Clft
Perl warning msgs are printed to the hats.machine.lst
preventing hats from starting up correctly.

PROBLEM CONCLUSION:
When a customer has LC_MESSAGES=Clft in their
/etc/environment - hats will not start due to
perl warning msgs that were logged in the
hats.machines.lst.
LC_MESSAGES=Clft is a work around written by AIX
to overcome a foreign message display problem at
boot time.
LC_MESSAGES=Clft is considered the same as LC_MESSAGES=C
for AIX applications. However, some applications, such as
PERL, do not consider "C=lft" the same as "C" - thus error
msgs such as the one below get printed:
perl: warning: Setting locale failed
The error msgs can be avoided by putting LC_ALL=C
in front of the perl commands in the hats script.

------

APAR: IY27711 COMPID: 5765C3403 REL: 430
ABSTRACT: BAD CX ADAPTER CAUSES MEMORY LEAK

PROBLEM DESCRIPTION:
When a cx adapter is marked as bad, it can cause memory leaks
when messages sent to the ports on the adapter are not
handled properly.

PROBLEM CONCLUSION:
The solution is to add in checks for the bad adapter in the
open, close, add, and put routines and handle the situation
more gracefully.

------

APAR: IY27721 COMPID: 5765B9501 REL: 320
ABSTRACT: LONG WAITERS: WAITING FOR RPC REPLIES FOR TMMSGTELLSERVER

PROBLEM DESCRIPTION:
Long Waiters: waiting for RPC replies for tmMsgTellServer

PROBLEM SUMMARY:
Deadlock due to lock handling on some recent
high SMP processors.

PROBLEM CONCLUSION:
Missing isync in a locking routine caused
incorrect wait counts in processors which handle instructions
in parallel.

------

APAR: IY27728 COMPID: 5765C3403 REL: 430
ABSTRACT: CRFS ON JFS TAKES A LONG TIME TO FIND A LOG DEVICE

PROBLEM DESCRIPTION:
crfs takes a long to find a jfslog if the underlying
volume group has may logical volumes defined in it.

PROBLEM CONCLUSION:
querylog() first needs to build a list of LVs that have a
type of jfslog, then read thru the list to find a LV that
is formatted as a jfslog.

------

APAR: IY27734 COMPID: 5765C3403 REL: 430
ABSTRACT: GETHOSTBYNAME FUNCTION IS VERY SLOW USING HOSTS FILE

PROBLEM DESCRIPTION:
When trying to resolve a host name from /etc/hosts file.
From the performance group ctrace tool, it shows that
bzero() a 8K buffer takes the most of CPU cycles every time
before reading a line from /etc/hosts file.

PROBLEM CONCLUSION:
We only bzero() the 8K buffer once. Then for each line
we read from /etc/hosts, we keep a counter on the length
of the string read. Before we read the next line from
/etc/hosts file, we only bzero() the number of bytes read
from the last line. It will improves the performance.

------

APAR: IY27737 COMPID: 5765C3403 REL: 430
ABSTRACT: SX25 - CIO RETURNS OPTION 0 =45 ON CIO_START_DONE FOR PVC

PROBLEM DESCRIPTION:
PVC will not start and QLLC trace shows that CIO returns
option 0 =45 on CIO_START_DONE. The X25 adapter setup will
also have PVC - lowest logical channel number non-zero.

PROBLEM CONCLUSION:
Correct the manipulation of the QLLC raddr_name (PVC channel
number) when attempting to insert the adapter number.

------

APAR: IY27738 COMPID: 5765C3403 REL: 430
ABSTRACT: REVOKING X SECURITY AUTHORIZATION CORES X SERVER

PROBLEM DESCRIPTION:
X server core dumps when client revokes authorization. Note
that the client is not suppose to revoke authorization.

PROBLEM CONCLUSION:
Correctly handle closing of client on X server.

------

APAR: IY27739 COMPID: 5765C3403 REL: 430
ABSTRACT: SERIALIZATION PROBLEM ON STRIPED MIRRORED LVS

PROBLEM DESCRIPTION:
There is a serialization problem with striped mirrored LVs
when there is a sync operation for a physical address and
a normal striped request on a logical address where it is
possible to allow both requests at the same time when one
should be held while the other is being processed.

PROBLEM CONCLUSION:
Correct the macro that deterines overlap of a physical address
of a sync request with the logical striped address of a
normal striped request.

------

APAR: IY27740 COMPID: 5765C3403 REL: 430
ABSTRACT: BUFFER OVERFLOW VULNERABILITIES IN NON-ROOT DIAGNOSTIC

PROBLEM DESCRIPTION:
Buffer overflow can occur in the diagnostics library. This
can cause data and instructions to be overwritten.

PROBLEM CONCLUSION:
Truncate the contents of the DIAGNOSTICS environment variable
before copying it into a buffer to prevent buffer overflow.

------

APAR: IY27741 COMPID: 5765C3403 REL: 430
ABSTRACT: APPARENT MEMORY LEAK WITH SECURITY LDAP PLUG-IN

PROBLEM DESCRIPTION:
Probably none, except if the slapd is running for a long time
it may result in system crash.

PROBLEM CONCLUSION:
Free the unused memory

------

APAR: IY27742 COMPID: 5765C3403 REL: 430
ABSTRACT: IPREPORT DECODE SUPPORT FOR FIELDS IN UNI ADD PARTY MESSAGES.

PROBLEM DESCRIPTION:
When receiving an Add Party message ipreport decodes the
CALL_STATE, ENDPOINT_REFERENCE, and ENDPOINT_STATE as an
ATM_TRAFFIC_DESC.

PROBLEM CONCLUSION:
Add support to ipreport to decode these messages correctly.

------

APAR: IY27743 COMPID: 5765C3403 REL: 430
ABSTRACT: GXT4500/6500 DIAGNOSTIC ENHANCEMENTS

PROBLEM DESCRIPTION:
Some of the resolution/refresh rate combinations would not be
available for the diagnostics to run under such as 1600x1200
on a digital monitor or 2048x1536.

PROBLEM CONCLUSION:
Provide support for these new resolution/refresh rate
combinations as well as allowing the use of the device driver
for set up of them (both options will be available in code,
though the device driver will be doing it in all code the
customer sees). Also provide similar support for the 0x1090
and 0x1094 registers whose required values keep changing: let
the device driver value be the default or allow a new value
to be written in including the user-requested value (again, a
real customer would not get these options.

------

APAR: IY27744 COMPID: 5765C3403 REL: 430
ABSTRACT: INCREASE TIME RGB COLOR BARS ARE DISPLAYED

PROBLEM DESCRIPTION:
This defect is caused by the pattern test not waiting long
enough on older displays.

PROBLEM CONCLUSION:
Increase the time where the pattern is displayed, allowing
older displays to refresh, and the pattern to be checked by
the user.

------

APAR: IY27745 COMPID: 5765C3403 REL: 430
ABSTRACT: DATABASE FAILED WHEN ASYNC IO RETURNED EAGAIN

PROBLEM DESCRIPTION:
Oracle returning from aio_nwait with bad argument EINVAL

PROBLEM CONCLUSION:
Discovered 2 problems.
On early returns from lio_listio the aio control blocks were
not updated to reflect errno causing early return. This was
not however causing the problem.
A race condition occurred with the update of aio_pending and
aio_done which could cause aio_nwait to return EINVAL
Code was changed to serialize updates to avoid race condition.
The problem was not seen after this change.

------

APAR: IY27746 COMPID: 5765C3403 REL: 430
ABSTRACT: AIONWAIT RETURNS EARLY

PROBLEM DESCRIPTION:
aionwait is called and returns a smaller amount of aio control
blocks done then it should return.

PROBLEM CONCLUSION:
fixed serialization problem of updating aiodone & aiopending
fields on filesystem path. This has already been done on fast
path by IY24838.

------

APAR: IY27750 COMPID: 5765C3403 REL: 430
ABSTRACT: MTU NOT SET WHEN JUMBO FRAMES ENABLED ON GIGIBIT ETHERNET

PROBLEM DESCRIPTION:
Customer's enX mtu size may not automatically be set to
9000 when jumbo frames are enabled on gigabit ethernet.

PROBLEM CONCLUSION:
Enhance cfggxent to change the enX mtu attribute
automatically.

------

APAR: IY27751 COMPID: 5765C3403 REL: 430
ABSTRACT: 10/100 ETHERNET NOT COUNTING INTERRUPTS CORRECTLY

PROBLEM DESCRIPTION:
The 10/100 Ethernet driver is not counting the number of
receive interrupts correctly.

PROBLEM CONCLUSION:
Introduce a new flag inside the slih routin to indicate
whether the interrupt has been counted.

------

APAR: IY27752 COMPID: 5765C3403 REL: 430
ABSTRACT: LSVG CAN CAUSE ODM CORRUPTION

PROBLEM DESCRIPTION:
The LVM commands lsvg, lsvg, and lspv may corrupt the
ODM database, causing other commands to start failing,
and requiring the system to be rebooted to rebuilt the
ODM database.

PROBLEM CONCLUSION:
Obtain an ODM database lock before writing info into the
database.

------

APAR: IY27756 COMPID: 5765C3403 REL: 430
ABSTRACT: DSI IN ARMCM_FREE_CALL()

PROBLEM DESCRIPTION:
System with ATM adapter crashes with data storage interrupt
with the following traceback.
01E25988 atmcm_free_call+000054
01E2143C atmcm_svc_cleanup+0001D0
01E249FC atmcm_close+000394

PROBLEM CONCLUSION:
Add an else statement to avoid calling atmcm_free_call.

------

APAR: IY27757 COMPID: 5765C3403 REL: 430
ABSTRACT: DSI IN THE ADAPTER DRIVERS IODONE ROUTINE

PROBLEM DESCRIPTION:
DSI in the p8xx_iodone() routine of the adapter driver.

PROBLEM CONCLUSION:
The code will be modified to wait for iodone() to
complete before freeing it in the p8xx_iodone()
routine.

------

APAR: IY27758 COMPID: 5765C3403 REL: 430
ABSTRACT: ENTSTAT DOES NOT DISPLAY EACH ADAPTER'S NAME ON ETHERCHANNEL

PROBLEM DESCRIPTION:
When using the "entstat" command with an EtherChannel adapter,
the names of each adapter contained therein will not be
displayed: instead, the name of the EtherChannel adapter will
be displayed along with each adapter's statistics.

PROBLEM CONCLUSION:
The "ethchan_all_stats" structure was augmented to contain
an array of strings to hold the names of the adapters in the
EtherChannel. The code that retrieves the channel's statistics
was changed so that it would fill out this array with the
names of all the adapters contained therein. Finally, the
"entstat" command was changed so that it would retrieve the
names of the adapters from this array and display them
correctly.

------

APAR: IY27759 COMPID: 5765C3403 REL: 430
ABSTRACT: ADD XINERAMA EXTENSION TEXT TO X11/README

PROBLEM DESCRIPTION:
Documentation was missing for X Server Xinerama extension.

PROBLEM CONCLUSION:
Insert missing information in README file.

------

APAR: IY27760 COMPID: 5765C3403 REL: 430
ABSTRACT: WRONG AM_PM CHARACTER FOUND IN ZH_TW AND ZH_TW LOCALES

PROBLEM DESCRIPTION:
The user has a wrong Chinese character which indicates am_pm
when he(she) is in Zh_TW or ZH_TW locale and uses "date +%p"
command.

PROBLEM CONCLUSION:
Need to modify zh_TW.src

------

APAR: IY27761 COMPID: 5765C3403 REL: 430
ABSTRACT: WSM ALT_DISK_INSTALL: -B FLAG ALWAYS SET

PROBLEM DESCRIPTION:
The "boot from target disk on next reboot" option will be set
always even if not selected in wsm alt_disk_install dialog.

PROBLEM CONCLUSION:
Checking for value of set_bootlist flag instead of just
checking for its existence before adding it to command string.

------

APAR: IY27762 COMPID: 5765C3403 REL: 430
ABSTRACT: NETIF_BACKUP DOES NOT LOG ANY ERRORS.

PROBLEM DESCRIPTION:
If 'network interface backup' feature is enabled, the
driver does not log error log entries anytime the
cable is plugged/unplugged.

PROBLEM CONCLUSION:
Modified the driver so that these error messages are correctly
logged.

------

APAR: IY27764 COMPID: 5765C3403 REL: 430
ABSTRACT: ATMSVCD CORE DUMPS WHEN IT RECEIVES AN INVALID REQUEST ID

PROBLEM DESCRIPTION:
The Request ID should not be more than 4 octets. There is
a switch with a bug that is sending a Request ID with 5 octets
where the 5th high order octet is zero. This is causing
atmsvcd to coredump.

PROBLEM CONCLUSION:
Change atmsvcd so that it does not core dump. If the high
order bytes passed 4 are zero accept the packet and strip
off the zero padding. If the high order bytes are not zero
log an error in the internal trace table and treat the packet
as invalid.

------

APAR: IY27765 COMPID: 5765C3403 REL: 430
ABSTRACT: LARGE DUMPS TO FC FAIL

PROBLEM DESCRIPTION:
Customer will experience I/O errors when trying to dump to
FC devices.

PROBLEM CONCLUSION:
Corrected a DMA mapping error inside of the dump_write path
of the adapter driver that was manifesting itself as an I/O
error.

------

APAR: IY27773 COMPID: 5765C3403 REL: 430
ABSTRACT: SECURITY: BUFFER OVERFLOW IN /USR/BIN/CU

PROBLEM DESCRIPTION:
Unknow, since it is a possible overflow vulnerability

PROBLEM CONCLUSION:
Replace strcpy with a strncpy, to prohibit overflow

------

APAR: IY27774 COMPID: 5765C3403 REL: 430
ABSTRACT: STDOUT INFORMATION FROM ATMSVCD CAUSES CFGMGR TO REPORT ERRORS

PROBLEM DESCRIPTION:
cfgmgr reports the following errors when configuring the
atm adapters:
cfgmgr: 0514-604 Cannot access the CuDv object class in the
device configuration database.

PROBLEM CONCLUSION:
Remove the line that is printing debug information to stdout.

------

APAR: IY27777 COMPID: 5765C3403 REL: 430
ABSTRACT: SYSTEM CRASH AT BOOT WITH ATM ADAPTER

PROBLEM DESCRIPTION:
Machine crashing during boot due to ATM adapter

PROBLEM CONCLUSION:
Device driver was changed to correct the problem.

------

APAR: IY27778 COMPID: 5765C3403 REL: 430
ABSTRACT: SECURITY: ENVIRONMENT VARIABLE NAME STRING TOO SHORT.

PROBLEM DESCRIPTION:
Login program may core dump when specifying 100+ environment
variables after the user name.

PROBLEM CONCLUSION:
Change the length of the string allocated for the environment
variable name to accomodate 3 digit numbers.

------

APAR: IY27779 COMPID: 5765C3403 REL: 430
ABSTRACT: DATA SCRUBBING NOT ENABLED AFTER MICROCODE DOWNLOAD

PROBLEM DESCRIPTION:
Downloading microcode to the 3 or 4 channel PCI SCSI RAID
adapters causes data scrubbing to be disabled.

PROBLEM CONCLUSION:
After downloading microcode, the user must delete any arrays
attached to the adapter, enable data scrubbing, then recreate
the arrays.

------

APAR: IY27783 COMPID: 5765C3403 REL: 430
ABSTRACT: GETADDRINFO DOESN'T RETURN CANONICAL HOSTNAME IF AI_CANONNAME

PROBLEM DESCRIPTION:
getaddrinfo() doesn't return canonical host name
if AI_CANONNAME is set. Insteas it returns the 1st alias
name on the host alias list.

PROBLEM CONCLUSION:
If AI_CANONNAME flag is set, we will return canonical
host name instead of the alias host name.

------

APAR: IY27788 COMPID: 5765C3403 REL: 430
ABSTRACT: FAST-T 500 EXTENDED FUNCTIONALITY CHANGES

PROBLEM DESCRIPTION:
Reservations may not be properly retained, multiple partitions
are not supported, problems with UTM LUNs. Fast-T 200 not
included in ODM.

PROBLEM CONCLUSION:
Add support for Fast-T extended features.

------

APAR: IY27792 COMPID: 5765C3403 REL: 430
ABSTRACT: ALLOCP COREDUMPS FROM EXTENDLV

PROBLEM DESCRIPTION:
extendlv command fails due to allocp core dump.

PROBLEM SUMMARY:
extendlv causes allocp to coredump given invalid
upperbound lv attribute.

PROBLEM CONCLUSION:
extendlv checks validity of upperbound value before
calling allocp.

TEMPORARY FIX:
Ensure value of upperbound is valid given number
of PVs in VG, and number of copies of LV.

------

APAR: IY27794 COMPID: 5765C3403 REL: 430
ABSTRACT: CORE DUMP WITH ATM DEVICE DRIVER

PROBLEM DESCRIPTION:
While configuring ATM interface, core file is getting
generated.

PROBLEM CONCLUSION:
Initializing the variable mtu_size while calling
tcp_getattr_isno().

------

APAR: IY27799 COMPID: 5765B8100 REL: 230
ABSTRACT: MAKECALL FAILURES USING EURO-ISDN.

PROBLEM DESCRIPTION:
MakeCall fails for certain switches when using Euro-ISDN. An
ISDN_MONITOR trace shows the switch sending a DATE/TIME IE
(Information Element) in the connect message - DirectTalk
responds with a Connect Acknowledge message containing a CAUSE
IE stating that DATE/TIME is not supported. The CAUSE IE is
actually not valid for the Connect Ack.

PROBLEM SUMMARY:
MAKECALL FAILURES USING EURO-ISDN

PROBLEM CONCLUSION:
Changed ISDN so that when a connect msg is
received with an unsupported IE eg, date+time, then the IE is no
w ignored and the connect ack msg that is sent does not
contain a cause code.

------

APAR: IY27801 COMPID: 5765C3403 REL: 430
ABSTRACT: XINPUTMETHOD HAVING TROUBLE WITH XKB EXTENSION

PROBLEM DESCRIPTION:
When XKeyBoard extension is in use, some applications may not
be able to process keyboard input correctly.

PROBLEM CONCLUSION:
When XKB is enabled use the function XkbLookupKeySym to do
KeySym lookup inside the input method.

------

APAR: IY27803 COMPID: 5765C3403 REL: 430
ABSTRACT: PROBLEMS WITH KSH BUILTIN TYPESET

PROBLEM DESCRIPTION:
ksh builtin typeset -f <function name> returns improper
output for imbricate functions.

PROBLEM CONCLUSION:
Suitable code changes were made to fix the above reported
problem.

------

APAR: IY27815 COMPID: 5765C3403 REL: 433
ABSTRACT: ATM LANE - VCS HANG ON ADD FILTER FAILURES

PROBLEM DESCRIPTION:
Config Direct VC's are not removed on add filter failures,
causing adapter VC's to become increasingly unavailable as
recovery attempts continue.

PROBLEM CONCLUSION:
Restructure the recovery procedure.

------

APAR: IY27818 COMPID: 5765C3403 REL: 430
ABSTRACT: DEVICES.PCI.4F111100.COM PTF NEEDS REBOOT FLAG

PROBLEM DESCRIPTION:
The recent PTF's for devices.pci.4f111100.com need to
inform the customer that a reboot is necessary before
the changes will take effect.

PROBLEM CONCLUSION:
Create a cfginfo file for the fileset which will cause the
install and update process to inform the user that a reboot
is necessary before the changes applied by the new PTF will
take effect.

------

APAR: IY27822 COMPID: 5765C3403 REL: 430
ABSTRACT: DIGITAL SUPPORT FOR 3840X240016HZ ON NEW GRAPHICS ADAPTERS

PROBLEM DESCRIPTION:
Adding 3840x240016Hz support to GXT4500P and GXT6500P for
very large digital TFTs. Also ensuring that no digital TFT
will be given a resolution with more pixels than are supported

PROBLEM CONCLUSION:
Add 3840x240016Hz timings to ODM and device driver. Add new
checks for maximum pixels on digital displays to central
config method.

------

APAR: IY27825 COMPID: 5765C3403 REL: 430
ABSTRACT: FASTT CONFIG AND RESERVATION CHANGES

PROBLEM DESCRIPTION:
Problems configuring FastT in several different manners.
LUN-Level reset does not work properly.

PROBLEM CONCLUSION:
Fix SCIOLCMD2 logic to always respect LUN choice. Fix
fdarcfgrule to handle dacs greater than 9. Increase adapter
limit. Fix SC_FORCED_OPEN and SC_FORCED_OPEN_LUN issues.

------

APAR: IY27830 COMPID: 5648C9802 REL: 430
ABSTRACT: SDK 1.3.0 PTF 9A : CA130-20020124

PROBLEM DESCRIPTION:
Fixes since PTF 9 (ca130-20020117) :
(Note: The descriptions here have been truncated.)
+----------+--------+-------+---------------------------------+
|20020124 |39013 | |space in JScrollPane not repainte|
+----------+--------+-------+---------------------------------+
|20020124 |39468 | |JNI Compatibility section of Read|
+----------+--------+-------+---------------------------------+
|20020124 |39473 | |Cannot navigate the print dialog |
+----------+--------+-------+---------------------------------+
|20020124 |39752 | |Path problem. |
+----------+--------+-------+---------------------------------+
|20020124 |39891 |PQ56430|Problem with java.net.URLEncoder/|
+----------+--------+-------+---------------------------------+
|20020124 |40008 |IY27249|JNI->ExceptionDescribe can print |
+----------+--------+-------+---------------------------------+
|20020124 |40083 | |awt.h is broken with DEBUG flag |
+----------+--------+-------+---------------------------------+
|20020124 |40094 | |AIXPPC32 Unexport remote object->|
+----------+--------+-------+---------------------------------+
|20020124 |40218 | |SEGV in libawt.so with TL6.5 |
+----------+--------+-------+---------------------------------+

------

APAR: IY27836 COMPID: 5765C3403 REL: 430
ABSTRACT: IMPROVE THREADSAFETY OF MALLOC

PROBLEM DESCRIPTION:
When executing under high-stress loads, it is possible for
malloc to call abort() and core dump the process

PROBLEM CONCLUSION:
removed optimistic malloc algorithm which is vulnerable to a
catastrophic race condition.

------

APAR: IY27837 COMPID: 5765C3403 REL: 430
ABSTRACT: LOG LINK ERROR EVERYTIME THE LINK GOES DOWN.

PROBLEM DESCRIPTION:
In a network interface backup setup, the device driver
logs LINK error only once.

PROBLEM CONCLUSION:
Changed the driver to log LINK errors everytime the link
changes on the backup card.

------

APAR: IY27855 COMPID: 5765C3403 REL: 430
ABSTRACT: BUFFER OVERFLOW IN LSCFG

PROBLEM DESCRIPTION:
Buffer overflow vulnerabilities in lscfg

PROBLEM CONCLUSION:
Limit the length of input string

------

APAR: IY27856 COMPID: 5765C3403 REL: 430
ABSTRACT: DIAG APPLICATION SHOULD IDENTIFY ERROR LOG CAUSING SRN

PROBLEM DESCRIPTION:
New diagnostics functionality for cryptographic adapter.

------

APAR: IY27857 COMPID: 5765C3403 REL: 430
ABSTRACT: CANNOT UPDATE SYSTEM FIRMWARE IN NIM ENVIRONMENT

PROBLEM DESCRIPTION:
Attempting to update system or service processor flash over
the network will fail on Common Hardware Reference Platform
systems when using diagnostics.

PROBLEM CONCLUSION:
Prevent the shutdown command from attempting to log the time
of reboot. Create new menus for use when updating system or
service processor flash over the network.

------

APAR: IY27858 COMPID: 5765C3403 REL: 430
ABSTRACT: DSI ON UNPIN DURING BACKUP FROM DB2

PROBLEM DESCRIPTION:
System crashes with DSI while restoring database from backup
Tape.

PROBLEM CONCLUSION:
The leaf routine in read call of tape driver modifies the
aspace_id in xmem descriptor, which shouldnot be, since
no data xfer is required in leaf routine. Instruction that
modifies aspace-id in leaf routine str_space is removed.

------

APAR: IY27883 COMPID: 5765D5101 REL: 121
ABSTRACT: CM&CSSMEMBERSHIP GAVE WRONG STATUS

PROBLEM DESCRIPTION:
cm & cssmembership gave wrong status

------

APAR: IY27886 COMPID: 5765C3403 REL: 430
ABSTRACT: SRN 803-7C1 RUNNING DIAG ON PAUD0

PROBLEM DESCRIPTION:
Customer will see wrong menues which don't make sense.

PROBLEM CONCLUSION:
New entries added to the message file should be appended at
the end of the set, not inserted in the middle.

------

APAR: IY27887 COMPID: 5765C3403 REL: 430
ABSTRACT: ATMSVCD CORES WHEN RELEASE IS RECEIVED BEFORE SETUP COMPLETES

PROBLEM DESCRIPTION:
atmsvcd can core dump when it receives a Release while it is
still processes the Setup that is being released.

PROBLEM CONCLUSION:
Don't delete object until outstanding commands for the VC
are received from Call Manager.

------

APAR: IY27889 COMPID: 5765C3403 REL: 430
ABSTRACT: IPREPORT DOES NOT DECODE UNI DROP PARTY ACK MESSAGES.

PROBLEM DESCRIPTION:
Ipreport does not decode UNI Drop Party Ack messages.

PROBLEM CONCLUSION:
DROP_PARTY_ACK is defined as 0x86 when it should be 0x84.
Fix the define so it is correct.

------

APAR: IY27895 COMPID: 5765E8301 REL: 320
ABSTRACT: LDAP:PROBLEMS IN LDAP_SEARCH_REVISED AND LDAP_RESULT_REVISED

PROBLEM DESCRIPTION:
dcecp commands may fail with LDAP_SERVER_DOWN logged to
the warning log.

PROBLEM CONCLUSION:
coded error recovery for asynchronous searches correctly

------

APAR: IY27897 COMPID: 5765E8301 REL: 320
ABSTRACT: LDAP: UNBIND HANDLE IMMEDIATELY IN GET_HANDLE() IF POSSIBLE

PROBLEM DESCRIPTION:
LDAP connections are not unbound when they are no longer
in use

PROBLEM CONCLUSION:
unbind the LDAP connections when we are sure they are no
longer being used

------

APAR: IY27898 COMPID: 5765E8301 REL: 320
ABSTRACT: LDAP: FAILURE TO CHANGE PASSWORD: RECURRING

PROBLEM DESCRIPTION:
Some ldap errors put out a hex number rather than the
real error

PROBLEM CONCLUSION:
return the correct ldap error when received

------

APAR: IY27903 COMPID: 5765C3403 REL: 430
ABSTRACT: DUMP ADAPTER'S EXTENDED STATISTICS COUNTERS.

PROBLEM DESCRIPTION:
Driver does not display all the adapter's statistics.

PROBLEM CONCLUSION:
Modified the driver to make the adapter dump the extended
statistical counters.

------

APAR: IY27907 COMPID: 5765C3403 REL: 430
ABSTRACT: SVMON DISPLAYS REDUDANT SEGMENT ENTRIES

PROBLEM DESCRIPTION:
svmon -U, -C shows high and inaccurate Inuse values.
svmon -U, -C shows duplicate entries for the same vsid.

PROBLEM CONCLUSION:
Fixed svmon code to discard the duplicate entries and
take only unique entries for the Inuse calculations.

------

APAR: IY27908 COMPID: 5765B9501 REL: 320
ABSTRACT: THIRD PARTY SOFTWARE (CLEARCASE) NON-STANDARD USE OF THE VNOP-

PROBLEM DESCRIPTION:
A Third party program called VNOP_OPEN without passing a valid
pointer (vinfo). This led to a node crash due to GPFS kernel
panic in GPFS_V_OPEN.

LOCAL FIX:
Allow open without a vinfo pointer.

PROBLEM SUMMARY:
A third party distributed file system called a GPFS entry
point with an invalid pointer resulting in a GPFS panic.

PROBLEM CONCLUSION:
Allow open without a vinfo pointer. If this occurs, close
the file after successfully opening it.

------

APAR: IY27909 COMPID: 5765C3403 REL: 430
ABSTRACT: SYSINTR ERRORS IN ERROR LOG

PROBLEM DESCRIPTION:
The follow error log entries in the error log where the
interrupt level matches the level of the 2-port HDLC adapter:
    LABEL: INTRPPC_ERR
    IDENTIFIER: DADF69E4
    Date/Time: Fri May 18 15:28:50
    Sequence Number: 29777
    Machine Id: 000B089D4C00
    Node Id: ibm
    Class: H
    Type: UNKN
    Resource Name: SYSINTR
    Resource Class: NONE
    Resource Type: NONE
    Location: NONE
    Description
    UNDETERMINED ERROR
    Probable Causes
    SYSTEM I/O BUS
    SOFTWARE PROGRAM
    ADAPTER
    DEVICE
            Recommended Actions
            PERFORM PROBLEM DETERMINATION PROCEDURES
    Detail Data
    BUS NUMBER
    0000 00C0
    INTERRUPT LEVEL
    0000 0016

PROBLEM CONCLUSION:
These entries are due to a pending transmit interrupt being
handled in the write entry point on a subsequent transmit as
the adapteris posting an interrupt to the driver's SLIH for
the previous one. The SLIH, once called, then does not see an
outstanding interrupt from the adapter and returns INTR_FAIL
which causes the error log.

------

APAR: IY27910 COMPID: 5765C3403 REL: 430
ABSTRACT: PCI ERROR RECOVERY NEEDS TO RESTORE PCI CONFIG SPACE REGISTERS

PROBLEM DESCRIPTION:
During PCI error recovery, the initial settings of the PCI
config space registers are reset to the default values.

PROBLEM CONCLUSION:
Reinitialize the PCI config space register settings following
a PCI error during recovery.

------

APAR: IY27911 COMPID: 5765C3403 REL: 430
ABSTRACT: MODIFY HERMOSA ASIC PROBLEM WORKAROUND

PROBLEM DESCRIPTION:
This is a potential that if the latency between accesses to
the 2-port PCI adapter's internal I/O registers is not long
enough that system could crash.

PROBLEM CONCLUSION:
It is recommended to access adapter DRAM at offset 0 prior to
accessing the adapter's internal I/O registers.

------

APAR: IY27925 COMPID: 5765C3403 REL: 430
ABSTRACT: SYSTEM CRASH (ITI) AFTER LSRESOURCE -AL

PROBLEM DESCRIPTION:
If the first adapter that is configured fails, then the code
is not unpinned and the system will panic when the config
method attempts to unload the code.

PROBLEM CONCLUSION:
The solution is to call last_adapter_del to release all global
resources of the driver, remove itself from the device switch
table, and unpin the code.

------

APAR: IY27926 COMPID: 5765B8100 REL: 230
ABSTRACT: HOMOLAGATION IMPROVEMENTS

PROBLEM DESCRIPTION:
Homolagation improvements

PROBLEM CONCLUSION:
Homolagation improvements added

------

APAR: IY27956 COMPID: 5765C3403 REL: 430
ABSTRACT: ENTSTATS.RESERVED1 NOT SET TO CORRECT DEFAULT

PROBLEM DESCRIPTION:
Non-optimal network performance when running with Gigabit
ethernet adapters.

PROBLEM CONCLUSION:
Report default connect speed of 1000Mbit/sec.

------

APAR: IY27969 COMPID: 5765B8100 REL: 230
ABSTRACT: ISDN HOMOLAGATION IMPROVEMENTS

PROBLEM DESCRIPTION:
ISDN Homolagation improvements

PROBLEM CONCLUSION:
ISDN Homolagation improvements added

------

APAR: IY27982 COMPID: 5765C3403 REL: 430
ABSTRACT: CTSDD CAN DO UNNEEDED UNPINCODE

PROBLEM DESCRIPTION:
If ctsconfig gets an error during config, it can do an
unpincode which is not needed. This will cause the system to
panic.

PROBLEM SUMMARY:
If ctsconfig gets an error during config, it can do an
unpincode which is not needed. This will cause the system to
panic.

PROBLEM CONCLUSION:
Add flags in the code to remember if the code has been pinned
and then unpin only if necessary.

------

APAR: IY27984 COMPID: 5765C3403 REL: 430
ABSTRACT: SPMI GENERATES OVERFLOW PEAK VALUES

PROBLEM DESCRIPTION:
Plot data graphs have high Y-axis label values making the plot
unreadable.

PROBLEM CONCLUSION:
Code adjusted to handle Long counter value overflow.

------

APAR: IY28029 COMPID: 5765D5101 REL: 121
ABSTRACT: TOPSVCS WITH LARGE TUNABLES

PROBLEM DESCRIPTION:
topsvcs with large tunables

PROBLEM SUMMARY:
RSCT Topology Services, when working to support HACMP/ES,
allows different networks to have different network
tunables: the "heartbeat rate", which is the period with
which heartbeats are sent and monitored, and the "cycle",
which is the number of missed heartbeats before an adapter
is declared as down. These tunables control how long it
takes from a failure in an adapter to be detected.
A problem has been found when very long adapter
detection times are specified. After an HACMP resource is
added to or removed from the configuration dynamically
(which may force adapters to move from or to service
addresses), it is possible for Topology Services'
adapter membership groups to take too long to form. In
particular, it could take longer than the network's grace
period value. In a configuration without working non-IP
adapters, this may result in a perceived temporary
network partition. When the partition is finally healed,
the result will be one of the nodes being powered down due
to RSCT Group Services (and consequently the HACMP/ES
Cluster Manager) exiting.

PROBLEM CONCLUSION:
Logic has been introduced into the RSCT Topology
Services daemon to speed up the detection of remote
adapters that leave an adapter membership the group
and try to rejoin it. This logic has been enhanced to take
adapters moving between boot and service addresses (and
vice-versa) into account.
As a result of the fix, the adapter membership group for
the network affected by the resource change can be
reinstated well before the network grace period is over,
and no perceived network partition should occur.

------

APAR: IY28031 COMPID: 5765E6110 REL: 220
ABSTRACT: TOPSVCS WITH LARGE TUNABLES

PROBLEM DESCRIPTION:
topsvcs with large tunables

------

APAR: IY28048 COMPID: 5765C3403 REL: 430
ABSTRACT: TOPAS DOESN'T SHOW ALL OF LONG DEV NAMES

PROBLEM DESCRIPTION:
Customer unable to identify devices with long names in topas.

PROBLEM CONCLUSION:
Since space is limited in the curses based topas and the
device fields are only 8 char long, A changed was made to
show the last eight characters of the name.

------

APAR: IY28051 COMPID: 5765B8100 REL: 220
ABSTRACT: ERROR ID 1 GENERATED WHEN CALLER HANGS UP BEFORE ANSWER

PROBLEM DESCRIPTION:
Error ID 1 is generated when the caller hangs up before the
State Table answers the call. This is a similar symptom to
APAR IY19102 but is cause by a diferent internal error
(EINTR).

PROBLEM SUMMARY:
Error ID 1 is generated when the caller hangs
up before the State Table answers the call.

PROBLEM CONCLUSION:
Added EINTR case for return from IOCTL
ANSWER_CALL in vp_answer_call.

------

APAR: IY28057 COMPID: 5765C3403 REL: 430
ABSTRACT: BORDER LINE INCONSISTENTLY DRAWN

PROBLEM DESCRIPTION:
The GXT120P, GXT130P and GXT135P graphic adapters may hang
or render lines or rectangles incorrectly.

PROBLEM CONCLUSION:
Place an __iospace_eieio before polling the status register in
the DDX (Device Dependent X) code. The eieio will make sure
all commands have been sent to the graphics adapter before
polling the status register.

------

APAR: IY28072 COMPID: 5765C3403 REL: 430
ABSTRACT: DATA SCRUB ENABLE

PROBLEM DESCRIPTION:
FC 2494 and 2498 PCI SCSI adapters will have data scrubbing
disabled by default.

PROBLEM CONCLUSION:
To enable data scrubbing, all arrays on the adapter must be
deleted. SMIT may now be used to enable data scrubbing. Any
new arrays created on the adapter will now have data scrubbing
enabled.

------

APAR: IY28149 COMPID: 5765C3403 REL: 430
ABSTRACT: FIX DATABASE NOT SHIPPED WITH 9/2001 4.3.3 MEDIA

PROBLEM DESCRIPTION:
Systems installed from the 9/2001 AIX 4.3.3 media do not have
swvpd necessary to determine if APARs included in this level
are installed.

PROBLEM CONCLUSION:
Add fixdata to bos.rte.install for all APARs included in the
9/2001 AIX 4.3.3 media.

------

APAR: IY28180 COMPID: 5765B8100 REL: 230
ABSTRACT: DIRECTTALK BEANS SUPPORT FOR WVS 2.0

PROBLEM DESCRIPTION:
Additional support for DirectTalk BEANS (DTBE) for WVS 2.0

PROBLEM SUMMARY:
Additional support for WVS 2.0

------

APAR: IY28183 COMPID: 5765D5100 REL: 320
ABSTRACT: LATEST PSSP 3.2.0 FIXES AS OF FEBRUARY 2002

PROBLEM DESCRIPTION:
This is the latest PSSP ptf as of February 2002.
Order this apar to get all of the ptfs as of February 2002.

PROBLEM SUMMARY:
This is a packaging apar for PSSP 3.2.0 fixes
as of February 2002

PROBLEM CONCLUSION:
This is a packaging apar for PSSP 3.2.0
fixes as of February 2002

------

APAR: IY28184 COMPID: 5765B8100 REL: 230
ABSTRACT: ERROR ID 1 GENERATED WHEN CALLER HANGS UP BEFORE ANSWER

PROBLEM SUMMARY:
Error ID 1 is generated when the caller hangs
up before the State Table answers the call.

PROBLEM CONCLUSION:
Added EINTR case for return from IOCTL
ANSWER_CALL in vp_answer_call.

------

APAR: IY28187 COMPID: 5765D5101 REL: 121
ABSTRACT: DEATH IN FAMILY MESSAGE WAS NOT SENT AS DESIGNED

PROBLEM DESCRIPTION:
death in family message was not sent as designed

PROBLEM SUMMARY:
Logic has been introduced into the RSCT Topology Services
daemon so that on daemon exit, the remote peers are
immediately notified, therefore allowing for a quicker
detection that the local node is "down".
A problem was found, however, in the situation where both
RSCT Group Services and Topology Services are shut down at
approximately the same time: when the Group Services
daemon terminates, the Topology Services daemon notices
the break in the Unix-Domain Socket connection and
activates the "Simulate Death" mechanism. A problem with
the mechanism is that it is not notifying the remote
daemons. Moreover, it was preventing the subsequent daemon
termination logic from sending the notification.

PROBLEM CONCLUSION:
A fix was introduced into the RSCT Topology Services
daemon's "Simulate Death" logic. When "simulating
death" (a reaction to the termination of the Group
Services daemon), Topology Services now sends
a notification to remote peers about the need to expel
the local node from the remote nodes' adapter membership
groups.
With the fix, remote Topology Services daemons will be
notified of the termination of the local daemon, even
when Topology Services and Group Services are terminated
at approximately the same time.

------

APAR: IY28189 COMPID: 5765E6110 REL: 220
ABSTRACT: DEATH IN FAMILY MESSAGE WAS NOT SENT AS DESIGNED

PROBLEM DESCRIPTION:
death in family message was not sent as designed

------

APAR: IY28196 COMPID: 5765C3403 REL: 430
ABSTRACT: BOS.NET.SNAPP 4.3.3.75 DOES NOT INSTALL PROPERLY

PROBLEM DESCRIPTION:
SNAPP does not work on systems installed from the
9/2001 4.3.3 install media.
The bos.net install image on the 9/2001 4.3.3 install
media does not contain files for bos.net.snapp.

PROBLEM CONCLUSION:
This APAR ships a bos.net.snapp install image to
resolve this problem.

------

APAR: IY28225 COMPID: 5765E6110 REL: 110
ABSTRACT: SECURITY ISSUE

PROBLEM DESCRIPTION:
security issue

------

APAR: IY28236 COMPID: 5765C3403 REL: 430
ABSTRACT: PCI ETHERNET II ADAPTERS FAIL TO CONFIGURE

PROBLEM DESCRIPTION:
The limited size of xmalloc memory introduces a
limit on the number of PCI Ethernet II adapters that
may be configured at the same time.

PROBLEM CONCLUSION:
Changed the device driver to allocate it's control blocks
from the net_malloc memory instead.

------

APAR: IY28244 COMPID: 5765C3403 REL: 430
ABSTRACT: STRIPES IN GXT6X00P AND GXT4X00P SERVICE AIDS

PROBLEM DESCRIPTION:
The service aids for the GXT6x00P and GXT4x00P have stripes
on screen

PROBLEM CONCLUSION:
Code needs to check new flags in service aids as well as
normal diag sequence selections

------

APAR: IY28252 COMPID: 5765B8100 REL: 220
ABSTRACT: MAKECALL FAILURES USING EURO-ISDN.

PROBLEM SUMMARY:
MAKECALL FAILURES USING EURO-ISDN

------

APAR: IY28270 COMPID: 5765D5100 REL: 330
ABSTRACT: REQUIRED UPDATES FOR PSSP R3.3.0

PROBLEM DESCRIPTION:
required updates for PSSP R3.3.0

PROBLEM SUMMARY:
required updates for pssp R3.3.0

------

APAR: IY28301 COMPID: 5765C3403 REL: 430
ABSTRACT: GXT4500P HANGS WITH MODEL ROTATION

PROBLEM DESCRIPTION:
The adapter could hang running in various situations such
as animating a CATIA model for an extended amount of time at
a high resolution.

PROBLEM CONCLUSION:
Added an INI_NOP and an INI_FLUSH before every FRG_FUNCTEN
call. Added an INI_NOP before FRG_ANCWE and FRG_FBWE that
immediately followed any DRW commands.

------

APAR: IY28303 COMPID: 5765C3403 REL: 430
ABSTRACT: PCI LATENCY TIMER SETTING CAUSES DEGRADATION ON ADSM BACKUPS

PROBLEM DESCRIPTION:
For customers using the JBOD SCSI 825A and 875 adapters, when
moving from AIX 4.2 to AIX 4.3.3 or AIX version 5, throughput
degradation may be noticed on the following systems: F50, H50,
150 and 260.

PROBLEM CONCLUSION:
Created a special configuration attribute to work around the
throughput degradation. This change will not have any affect
on systems other than F50, H50, 150 or 260. Moreover, the
attribute is only for the JBOD SCSI 825A amd 875 adapters.

------

APAR: IY28317 COMPID: 5765E8300 REL: 320
ABSTRACT: DCE320 MAINTENANCE PACKAGE FOR PTF SET 1

PROBLEM DESCRIPTION:
THIS APAR CREATED FOR MAINTENANCE PACKAGE OF PTF SET 1
dce320 Maintenance Package for PTF SET 1

PROBLEM SUMMARY:
AIXDCE320 MAINTENANCE PACKAGE FOR PTF SET 1

------

APAR: IY28352 COMPID: 5765D5100 REL: 330
ABSTRACT: LATEST PSSP 3.3.0 FIXES AS OF FEBRUARY 2002

PROBLEM DESCRIPTION:
This is the latest PSSP ptf as of February 2002.
Order this apar to get all of the ptfs as of February 2002.

PROBLEM SUMMARY:
This is a packaging apar for PSSP 3.3.0 fixes
as of February 2002

PROBLEM CONCLUSION:
This is a packaging apar for PSSP 3.3.0
fixes as of February 2002

------

Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]