APAR: IY17530 COMPID: 5765C3403 REL: 430
ABSTRACT: SOCKET STAY OPEN AFTER LSSRC

PROBLEM DESCRIPTION:
socket stay open after lssrc

PROBLEM SUMMARY:
Socket opened for using lssrc command is not getting freed.

PROBLEM CONCLUSION:
Proper check is being given so that the previous user
did read the messages of the socket until cont=END.

------

APAR: IY20098 COMPID: 5765C3403 REL: 430
ABSTRACT: NET STOP HANGS

PROBLEM DESCRIPTION:
net stop hangs

PROBLEM CONCLUSION:
break from the recvmsg call

------

APAR: IY20101 COMPID: 5765C3403 REL: 433
ABSTRACT: SYSLOGD TURNS ALL DBCS CHARACTERS INTO GARBAGE

PROBLEM DESCRIPTION:
syslogd turns all DBCS characters to garbage.

PROBLEM SUMMARY:
Customer will see control charaters in the log file when he is
trying to log Japanese characters

PROBLEM CONCLUSION:
This problem occurs when LS_MESSAGES=JA_JP or Ja_Jp
and TERM=Clft in Japanese lang environments
the problem is becuase the setlocale doesnt happen
as Japanese when it sees a Clft. When we check whether it
is a multi-byte code set in code we get a 1 instead of a 2
because of the wrong set locale. The solution is to check if
LC_MESSAGES=Ja_Jp and TERM=Clft and if so unset TERM before
the setlocale call in syslogd.c

------

APAR: IY20175 COMPID: 5765C3403 REL: 433
ABSTRACT: CANNOT SEE THE CONTENTS OF THE BODY OF A MAIL

PROBLEM DESCRIPTION:
In the output of iptrace during the connection of
imap server and client the client requests the
body by "BODY " but the server answers by
"BODY .MIME ". So the client cannot see the
content of body of mail.

PROBLEM SUMMARY:
Changed the line in the imapd.c file
        snprintf(tmp,sizeof(tmp),"BODY %s.MIME",ta->section);
to
        snprintf(tmp,sizeof(tmp),"BODY %s ",ta->section);

------

APAR: IY20384 COMPID: 5765C3403 REL: 430
ABSTRACT: DUMP ESTIMATES CAN BE LOW

PROBLEM DESCRIPTION:
Dump estimates can be really low depending
on processes running.

LOCAL FIX:
Consider the processes for estimating the dump
size.

PROBLEM CONCLUSION:
Consider the processes and their threads for dump estimates.

------

APAR: IY21079 COMPID: 5765C3403 REL: 430
ABSTRACT: INIT DISPLAYS IRRELEVANT ERROR MESSAGE AND ERRNO

PROBLEM DESCRIPTION:
This apar addresses two problems.
1. init displays bogus errno value if inittab
   is trunacted.
2. The error message displayed is irrelevant in
   this case.

PROBLEM SUMMARY:
"init" does not show proper error message
and errno value when "inittab" file is
truncated to zero size.

PROBLEM CONCLUSION:
The problem is addressed through proper
code changes.

------

APAR: IY21234 COMPID: 5765C3403 REL: 430
ABSTRACT: DAYLIGHT SAVINGS TIME PROBLEM

PROBLEM DESCRIPTION:
Files will have wrong timestamps

PROBLEM CONCLUSION:
Adding and subtracting 1 hour appropriately should solve this
problem

------

APAR: IY21502 COMPID: 5765C3403 REL: 430
ABSTRACT: NET VIEW FROM CLIENT RETURNS ERROR WITH MORE SHARES

PROBLEM DESCRIPTION:
"net view \\server" command on client
returns error when server has lot of
shares defined.

LOCAL FIX:
server needs to return correct return code.

PROBLEM SUMMARY:
"net view \server" command on client
returns error when server has lot of
shares defined.

PROBLEM CONCLUSION:
server needs to send a correct return code.

------

APAR: IY21604 COMPID: 5765C3403 REL: 430
ABSTRACT: CANNOT MODIFY UNSUCCESSFUL_LOGIN_COUNT ATTRIBUTE WITH -R OPTION

PROBLEM DESCRIPTION:
The attribute "unsuccessful_login_count" cannot be changed with
-R option from chuser. This attribute was supported from chsec
command as it is located in /etc/security/passwd file. But to
administer attributes in loadmodules like LDAP etc. chuser is
provided with -R option. This attribute cannot be changed with
chuser -R.

PROBLEM CONCLUSION:
/etc/security/lastlog attributes attributes added to the
array "chusatab" of type "struct chusattr".

------

APAR: IY21774 COMPID: 5765C3403 REL: 430
ABSTRACT: PROBLEM WITH MOVING FILES WITHIN SAME SHARE

PROBLEM DESCRIPTION:
Problem with moving files within same share

PROBLEM CONCLUSION:
Sending a different return code from Rename function will
solve this problem

------

APAR: IY21889 COMPID: 5765C3403 REL: 430
ABSTRACT: CPIO CORRUPTS FILES WITH LINK IF NEWER FILE (LINK) IS COPIED

PROBLEM DESCRIPTION:
==> on system A (ux48)
==> file pp created Jul 5
==> link ppl from pp using ln -s created Jul 5
[ux48:root]/tmp/ssh-klu15/ww: ls -l
total 8
-rw-rw-r-- 1 root system 28 Jul 05 08:55 pp
lrwxrwxrwx 1 root system 20 Jul 05 09:00 ppl -> /tmp/ssh-kl
u15/ww/pp
[ux48:root]/tmp/ssh-klu15/ww: cat pp
qdfnhjwiov
evgfjweriopvgjko
[ux48:root]/tmp/ssh-klu15/ww: cat ppl
qdfnhjwiov
evgfjweriopvgjko
 ==> on system B (ux40)
 ==> created file pp at Jul 6
 ==> created link ppl using ln -s at Jul 6
==> use cpio to copy files from system B (ux40) to system A (ux
48)
[ux40:root]/tmp/ssh-klu15/ww: ls -l
total 8
-rw-rw-r-- 1 root system 16 Jul 06 09:09 pp
lrwxrwxrwx 1 root system 20 Jul 06 09:10 ppl -> /tmp/ssh-klu
15/ww/pp
[ux40:root]/tmp/ssh-klu15/ww: cat pp
adsfhqwe
asdfsa
[ux40:root]/tmp/ssh-klu15/ww: cat ppl
adsfhqwe
asdfsa
[ux40:root]/tmp/ssh-klu15/ww: find /tmp/ssh-klu15/ww -xdev -dep
th -print | cpio -odm |remsh deziru48 cpio -idm
1 blocks
1 blocks
==> on system A (ux48)
==> file pp is corrupted. it contains now path of itself
[ux48:root]/tmp/ssh-klu15/ww: ls -l
total 16
-rw-rw-r-- 1 root system 20 Jul 06 09:14 pp
lrwxrwxrwx 1 root system 20 Jul 06 09:14 ppl -> /tmp/ssh-kl
u15/ww/pp
[ux48:root]/tmp/ssh-klu15/ww: cat ppl
/tmp/ssh-klu15/ww/pp[ux48:root]/tmp/ssh-klu15/ww:
[ux48:root]/tmp/ssh-klu15/ww: cat pp
 /tmp/ssh-klu15/ww/pp[ux48:root]/tmp/ssh-klu15/ww:

LOCAL FIX:
use tar or backup command

PROBLEM SUMMARY:
When cpio command is passed a link file as the
parameter, the source to which the link file
points gets corrupted.

PROBLEM CONCLUSION:
Suitable code changes are done so that the
source file to which the link points does
not get corrupted.

------

APAR: IY21904 COMPID: 5765C3403 REL: 430
ABSTRACT: HD_TERMINATE(STRIPE)

PROBLEM DESCRIPTION:
A system experiencing this problem would crash at the
assert in hd_terminate, and analysis of the dump would
indicate that the logical buffer being terminated is
actually a striped physical buffer.

PROBLEM SUMMARY:
System asserts in hd_terminate

PROBLEM CONCLUSION:
Fix internal logic for striped and mirrored LVs

------

APAR: IY22030 COMPID: 5765C3403 REL: 430
ABSTRACT: PROBLEMS WITH GETUTLINE() AND TTYSLOT() SUBROUTINES

PROBLEM DESCRIPTION:
The getutline and ttyslot subroutines returns processes
which are DEAD instead of the current one, which is running.

PROBLEM SUMMARY:
The getutline and ttyslot subroutines returns pids/slots of
processes which are DEAD instead of the current LOGIN process.

------

APAR: IY22050 COMPID: 5765D5101 REL: 121
ABSTRACT: MISSING MESSAGE TEMPLATES

PROBLEM DESCRIPTION:
RSCT 1.2.1 is missing some error templates and many messages
show up in errpt with LABEL: NONE and missing detail data.

PROBLEM SUMMARY:
In rsct.basic.rte 1.2.1 the error template errid's have
changed. When this release is migrated to, the new error
templates are missing and any of the rsct errpt entries
will not have detailed data.

PROBLEM CONCLUSION:
Modified the packaging files so that the updated
error templates for rsct.basic.rte will be applied
as part of the service stream.

------

APAR: IY22062 COMPID: 5765C3403 REL: 430
ABSTRACT: INVALID DIR USING A : CAN TRAP NT.

PROBLEM DESCRIPTION:
Some NT-clients can occasionally crash
if they receive AIX-filenames containing
a ':' (colon) -- to fix this, FastConnect
will make these files invisible to PCclients.

PROBLEM SUMMARY:
AIX filenames containing ":" can sometimes crash NT-clients,
due to flaws in some NT-client software.
This can be significant for customers upgrading from ACONN
to FastConnect, because ACONN created these filenames as
"invisible" files.

PROBLEM CONCLUSION:
Allow FastConnect to filter-out these filenames as "invisible"
files, like ACONN did.

------

APAR: IY22113 COMPID: 5765C3403 REL: 430
ABSTRACT: NISPLUS FIXES FOR 64-BIT APPLICATION ENVIRONMENT

PROBLEM DESCRIPTION:
Currently 64-bit applications cannot use NISPLUS-based
lookups for functions such as getpwnam, getpwent, gethostent
etc. This APAR addresses this issue and contains 64-bit cleanup
of parts of the libnsl library.

LOCAL FIX:
Recompile the application as 32-bit.

PROBLEM CONCLUSION:
There's a number of places where long and int types and
pointers are used intermixed. This causes problems in 64-bit
environments when they are of different size. This is
particularily a problem with the XDR functions used by RPC.
This APAR cleans up some of this type confusion, to make it
work better under 64-bit applications.

------

APAR: IY22203 COMPID: 5765D5101 REL: 121
ABSTRACT: PERSPECTIVES ERROR MSGS WITH V2.1 LIBDCE.A IN NON-DCE AUTHEN ENV

PROBLEM DESCRIPTION:
If a customer has V2.1 /usr/lib/libdce.a present,
with PSSP 3.2 (PTF 10) and AIX 4.33, using NON-
DCE Authentication,the invocation of either
Perspectives or Sphardware will result in the
following error messages:
0509-150 Dependent module /usr/lpp/ssp/bin
could not be loaded.
exec(): 0509-036 Cannot load program spsec_ldmod
because of the following errors:
0509-022 Cannot load module
                    /usr/lpp/ssp/bin/spsec_ldmod.
0509-150 Dependent module libdcepthreads.a
         (dcepthreads_shr.o) could not be
         loaded.
These were similar, but not the same errors
observed before IY17070 was applied in PTF
10 of PSSP 3.2. IY17070 corrected errors
with SDRChangeAttrValues when V2.1 libdce.a
was present in a non-DCE authentication
environment. Authough V2.1 libdce.a is not
"supported" with AIX 4.33, we have already
committed to its coexistence with PSSP 3.2
in a non-DCE environment with APAR IY17070.
The problem with Perspectives / Sphardware
is independent of the problem that was
fixed in APAR IY17070.

LOCAL FIX:
The workaround is to remove or rename
/usr/lib/libdce.a; however, there are
two independent customers who insist
that V2.1 libdce.a should be allowed
to coexist with PSSP 3.2 / AIX 4.33
in a non-DCE authentication environment.

PROBLEM SUMMARY:
Issuing sphardware on a CWS in a non-DCE environment will
result in error messages being issued if a level of DCE
prior to 3.1 exists on the system. The following messages
will be issued several times:
exec(): 0509-036 Cannot load program spsec_ldmod because
               of the following errors:
       0509-022 Cannot load module
               /usr/lpp/ssp/bin/spsec_ldmod.
       0509-150 Dependent module libdcepthreads.a
               (dcepthreads_shr.o) could not be loaded.
       0509-022 Cannot load module libdcepthreads.a
               (dcepthreads_shr.o).
       0509-026 System error: A file or directory in the
               path name does not exist.
       0509-022 Cannot load module /usr/lpp/ssp/bin.
       0509-150 Dependent module /usr/lpp/ssp/bin could
               could not be loaded.
The routines were trying to access a module that does
not exist in /usr/lib/libdce.a in the earlier version
of DCE.

PROBLEM CONCLUSION:
Modified code in perspectives and Event Management to
first verify that DCE is being used on the system,
prior to attempting to load the DCE libraries.
This will allow sphardware to be run in a non-DCE
environment when a level of DCE prior to 3.1 exists
on the system.

------

APAR: IY22308 COMPID: 5765C3403 REL: 430
ABSTRACT: MEMORY LEAK IN FCLOSE() SUBROUTINE

PROBLEM DESCRIPTION:
applications doing many file opens and closes will see an
increase in memory use and a decrease in performance

PROBLEM SUMMARY:
Memory leak in fclose()

PROBLEM CONCLUSION:
Fixed file threadsafe macros

------

APAR: IY22354 COMPID: 576565400 REL: 220
ABSTRACT: TPROF PRODUCES WRONG SOURCE LINE INFO ON -E PRE-PROCESSED FILES

PROBLEM DESCRIPTION:
tprof produces wrong source line information while
microprofiling a program which is compiled from
-E generated source code.

PROBLEM CONCLUSION:
Modified stripnm to display appropriate source file names
when a object is generated from the -E preprocessed file.

------

APAR: IY22374 COMPID: 5765C3403 REL: 430
ABSTRACT: GREP -W $ HANGS INFINITELY

PROBLEM DESCRIPTION:
"grep -w $" hangs infinitely causing increased CPU usage.

PROBLEM SUMMARY:
"grep -w $" hangs indefinitely resulting in
increased CPU usage.

PROBLEM CONCLUSION:
"grep -w $" hangs because of an infinite loop
thereby causing increased CPU usage.

------

APAR: IY22421 COMPID: 5765C3403 REL: 430
ABSTRACT: ETHERCHANNEL COMING UP IN DEFINED STATE FOLLOWING SYSTEM REBOOT

PROBLEM DESCRIPTION:
Etherchannel is coming up in defined state after system system
reboot

LOCAL FIX:
Phase value is changed back to 2 in cfginet.c

PROBLEM SUMMARY:
Customer is not able to use Etherchannel following system
reboot.cute cfcgech() and bring up etherchannel to
available state during system reboot.

PROBLEM CONCLUSION:
Code is modified to call cfgech with value 2 so that it
can execute cfgech() and bring up etherchannel to
available state during system reboot.

------

APAR: IY22423 COMPID: 5765C3403 REL: 430
ABSTRACT: SECURE NFS CREDENTIALS NEVER TIME OUT

PROBLEM DESCRIPTION:
In a secure NFS setting, the credentials that are cached in
the client never expire. Since failed key establishments are
cached as well, this means that on a client that is not used
heavily the cached entry will never expire. This can be a
problem, since keylogout essentially has no effect.

LOCAL FIX:
None really. A more busy client won't really see this since
the cache entries will be replaced by other entries as needed.

PROBLEM CONCLUSION:
Add a timeout field in each des cache entry in the secure
KRPC implementation. Also add an nfso option to tune the
timeout of DES entries, defaulting to 0 (no timeout) to
preserve current behaviour.

------

APAR: IY22432 COMPID: 5765C3403 REL: 430
ABSTRACT: DETECT LIGHT LOST ON STARTUP

PROBLEM DESCRIPTION:
If cable is not plugged in, the driver does not
go into limbo state to test for reconnect.

LOCAL FIX:
Run cfgmgr after reconnection of cables

PROBLEM SUMMARY:
Adapter at startup did not fail on light lost.

PROBLEM CONCLUSION:
Adapter now goes to limbo if light lost at startup

TEMPORARY FIX:
running cfgmgr after reconnect sometimes works

------

APAR: IY22438 COMPID: 5765C3403 REL: 430
ABSTRACT: FOLD -W <ODD> OUTPUT EXCEEDS THE COLUMN WIDTH SPECIFIED

PROBLEM DESCRIPTION:
"fold -w <integer>" displays a maximum of <integer>
bytes for a line. But, in a multibyte locale,
"fold -w <odd>" output exceeds the specified
column width.

PROBLEM SUMMARY:
"fold -w <odd>" displays output beyond the
specified column width in multibyte locales.

PROBLEM CONCLUSION:
In a multibyte locale, say Ja_JP, each Ja_JP
character occupies 2 bytes. A "fold -w <odd>"
displays one Ja_JP character in excess exceeding
the specified column width.

------

APAR: IY22469 COMPID: 5765C3403 REL: 430
ABSTRACT: CIFSSNAP: UPDATE TO COLLECT MORE DATA

PROBLEM DESCRIPTION:
Added support for the ps commands lsvg and netstat -rn

------

APAR: IY22480 COMPID: 5765C3403 REL: 430
ABSTRACT: MEMORY LEAK IN /USR/BIN/EXPR

PROBLEM DESCRIPTION:
Execution of JAVA application shows memory leak in AIX 4330-08.
ZeroFault obtained that /usr/bin/expr caused memory leak.

PROBLEM SUMMARY:
Memory leak observed in /usr/bin/expr command
when tested through ZeroFault tool.

PROBLEM CONCLUSION:
Code has been dropped to plug the UNREF blocks(mem leak).

------

APAR: IY22549 COMPID: 5765C3403 REL: 433
ABSTRACT: JAVA PROGRAM HANG

PROBLEM DESCRIPTION:
Programs using pthread_suspend_np() interfaces can hang,
if one of the threads doing a short sleep.
Since Java garbage collection reiles on pthread_suspend_np()
interface, Java programs are more likely to experience this
problem. Programs written in other languauges using this
interface is also at risk.

PROBLEM SUMMARY:
 Programs using pthread_suspend_np() interfaces can hang.

PROBLEM CONCLUSION:
 Programs using pthread_suspend_np() interfaces can hang.

------

APAR: IY22679 COMPID: 5765C3403 REL: 430
ABSTRACT: DTFILE REPORTS INCORRECT VALUES FOR FILES LARGER THAN 2 GB

PROBLEM DESCRIPTION:
With the By Name, date, size ... viewing option set, File
Manager displays the date, size, permission fields incorrectly
for large files whose size is more than or equal to 2
Gigabytes.

PROBLEM SUMMARY:
With the By Name, date, size ... viewing option set, File
Manager displays the date, size, permission fields incorrectly
for large files whose size is more than or equal to 2 GB.

PROBLEM CONCLUSION:
All the stat(),lstat() system calls were replaced by stat64()
and lstat64() system calls respectively.

------

APAR: IY22688 COMPID: 5765C3403 REL: 430
ABSTRACT: SYSTEM CRASH VMS_DELETE/SHMFREEWS IN CRASH STACK

PROBLEM DESCRIPTION:
System Crash following reboot. Analysis of the generated dump
via the crash tools gives a stack trace like this:
 0041E77C vms_delete+000070 (C0972FF4 ?? )
 004750C4 shmfreews+0000B4 () 0005AE84 freeuspace+000010 ()
 000599E0 kexitx+000768 (??) 00058E8C kexit+000078 ()

PROBLEM SUMMARY:
System Crash following reboot. Analysis of the generated dump
via the crash tools gives a stack trace like this:
0041E77C vms_delete+000070 (C0972FF4 ?? )
004750C4 shmfreews+0000B4 () 0005AE84 freeuspace+000010 ()
000599E0 kexitx+000768 (??) 00058E8C kexit+000078 ()

PROBLEM CONCLUSION:
Correct logic error in smconnect.c to ensure memory
area being freed is a Valid Shared Memory area.

------

APAR: IY22845 COMPID: 5765D5101 REL: 121
ABSTRACT: YP BLOCKING NETMONADAPTERHEALTH

PROBLEM DESCRIPTION:
netmonAdapterHealth hungs causing a false down of an adapter
which leads to multiple failovers in an hacmpes environment
that using YP.

PROBLEM SUMMARY:
When running Network Information Service (NIS),
it has been noticed that calls to getprotobyname()
could block for 30 seconds or more. Since such
calls are issued by the netmonAdapterHealth
program -- which is used by RSCT Topology Services
to determine the status of the local adapters --
the result was the program being blocked. In an
HACMP/ES environment, the blockage may cause the
local adapters to be flagged as down incorrectly,
provoking spurious "swap_adapter" events.

PROBLEM CONCLUSION:
To work around the problem of getprotobyname() blocking
in the presence of Network Information Service (NIS),
the code in RSCT Topology Services is being changed
to remove such call. With the fix, the
netmonAdapterHealth program should no longer
block when NIS is running, and therefore false
adapter down indications or "swap_adapter" events
in HACMP/ES should not occur.

------

APAR: IY22870 COMPID: 5765C3403 REL: 430
ABSTRACT: CDE HANGS WITH GXT2000

PROBLEM DESCRIPTION:
CDE hangs with GXT2000

PROBLEM SUMMARY:
DCE and FVWM hang on GXT2000

PROBLEM CONCLUSION:
Fix serialization problem in
GXT2000 loadddx

------

APAR: IY22944 COMPID: 5765C3403 REL: 430
ABSTRACT: JFS CORRUPTION CHECK FAILS DUE TO SERIALIZATION ISSUE

PROBLEM DESCRIPTION:
Customers with multi-threaded file-intensive processes may see
their system crash on something that looks like FS corruption.

PROBLEM CONCLUSION:
Fixed serialization problem that allowed a second thread to
try to delete a fragment the first thread was already deleting.

------

APAR: IY22953 COMPID: 5765C3403 REL: 433
ABSTRACT: TFTPD DIES WHEN SENT A PACKET WITH NO DATA

PROBLEM DESCRIPTION:
Tftpd dies when sent a packet with no data.

PROBLEM CONCLUSION:
Check for packet with no data.

------

APAR: IY23022 COMPID: 5765C3403 REL: 430
ABSTRACT: CHGRP FAILS TO CHANGE GROUPS FOR A LDAP USER

PROBLEM DESCRIPTION:
The chgrp fails to change a file's group
to other groups that the ldap user belongs
to. Even the commands like id, groups do
not show the groups that the ldap user
belongs to.

PROBLEM CONCLUSION:
Fix a bug in ldap_getentry() so that it copy out the right
group data.

------

APAR: IY23037 COMPID: 5765C3403 REL: 430
ABSTRACT: SECURITY: BUFFER OVERFLOWS IN LPD

PROBLEM DESCRIPTION:
Incorrect buffer handling in lpd can cause a security problem.

PROBLEM CONCLUSION:
Buffer handling has been corrected and lpd security has
been tightened up.

------

APAR: IY23038 COMPID: 5765C3403 REL: 430
ABSTRACT: DEVICES.IDE.CDROM NOT INSTALLED FROM GENERIC CD

PROBLEM DESCRIPTION:
If a generic mkcd (mksysb) CD is used in an IDE CDROM drive
to install a system, the IDE device drivers are not installed
on the system so the CD drive is not available once the
system is up and running.

PROBLEM SUMMARY:
If a generic mkcd (mksysb) CD is used in an IDE CDROM drive
to install a system, the IDE device drivers are not installed
on the system so the CD drive is not available once the
system is up and running even though it was used to install
the system. This is only a problem if the mksysb image used
to create the generic CD did not have the IDE drivers
installed.

PROBLEM CONCLUSION:
The install code thinks that everything is already there
since the drive is showing available because it is
currently in use.
Therefore, it does not attempt to install the software
for the device. Adding the files for the IDE drive to
the cdfs.optional.list file causes the install of the
IDE drive to correct this situation.

TEMPORARY FIX:
Install the devices.pci.ad100501.rte fileset on the system
that will be used to create the mksysb image for the generic
CD prior to creating the mksysb image.

------

APAR: IY23040 COMPID: 5765C3403 REL: 430
ABSTRACT: CPLV WORKS ON OPEN LVS

PROBLEM DESCRIPTION:
If a customer attempts to run cplv against an open logical
volume, cplv will not return an error as it should. This
may cause filesystem corruption and/or system crashes.

PROBLEM SUMMARY:
cplv works with open LV

PROBLEM CONCLUSION:
Fix check for open LV in cplv

------

APAR: IY23052 COMPID: 5765C3403 REL: 430
ABSTRACT: SNMPD CORE DUMPING IN AIX 4.3.3

PROBLEM DESCRIPTION:
snmpd is dying with errors. It gets
s/w permanent src error 139 and -9017

PROBLEM SUMMARY:
snmpd daemon core dumps _only_ when
SysKonnect PCI FDDI Adapter (48110040) is configured.

PROBLEM CONCLUSION:
bcopy() routine, the number of the source buffer copied to
destination is bigger than the destination buffer size.
It just writes outside the boundray of the destination buffer.
In file: src/tcpip/usr/sbin/snmpd/fddi.c,
*) In routine get_frames():
  replace:
      bcopy(buffer, frames *numframes .un.buffer,
        frames *numframes .buflen);
  with:
      bcopy(buffer, frames *numframes .un.buffer,
        MIN(frames *numframes .buflen, sizeof(buffer)));
*) In routine parse_frame()
   replace:
     bcopy((char *)(&(frame -> un.buffer idx + 4 )),
      varptr -> data, varptr -> len);
   with:
     bcopy((char *)(&(frame -> un.buffer idx + 4 )),
       varptr -> data,
                MIN(varptr -> len, sizeof(varptr -> data)));

------

APAR: IY23071 COMPID: 5765C3403 REL: 430
ABSTRACT: FTP CLIENT NOT LOGGING HOST PART IN SYSLOG OUTPUT

PROBLEM DESCRIPTION:
The information logged by syslogd while ftp
client is running in debug mode, doesn't
include the information about the remote
host that the user is connected to. It
includes the information about the user.
remote file name and local file name, but
no information about the hostname.

PROBLEM SUMMARY:
ftp client does not log the host information
in syslog output.

PROBLEM CONCLUSION:
audit_hostent->h_name is being replaced by
hostname in ftp.c file.

------

APAR: IY23072 COMPID: 5765C3403 REL: 430
ABSTRACT: VMSTAT REPORTS INCORRECT VALUES FOR RUNQUE AND CONTEXT SWITCH

PROBLEM DESCRIPTION:
vmstat reports incorrect values for runque and
context switches.

PROBLEM SUMMARY:
vmstat reports incorrect values for runque
and context switches.

PROBLEM CONCLUSION:
vmstat code modified to resolve the problem.

------

APAR: IY23129 COMPID: 5765C3403 REL: 430
ABSTRACT: IPTRACE SHOWS SAME PACKET AGAIN

PROBLEM DESCRIPTION:
stokdd sends data to iptrace multiple times
even though data leaves adapter only once.

PROBLEM SUMMARY:
stokdd processes multiple interrupts on some
tx frames causing multiple logging of frames
by iptrace.

PROBLEM CONCLUSION:
We check for the frame before logging it again
so the frame is logged only once.

------

APAR: IY23157 COMPID: 5765C3403 REL: 430
ABSTRACT: READFILE/WRITEFILE NEEDS CHECK FOR DISCONNECTED SOCKET

PROBLEM DESCRIPTION:
If a socket becomes disconnected from the other side and a
ReadFile or a WriteFile is issued to that socket, the system
call should return with an error indicating that the socket is
disconnected. Currently, however, the system call returns with
"no data available" which causes the application to wait for an
i/o event which will never come.

PROBLEM SUMMARY:
If a socket becomes disconnected from the other side and a
ReadFile or a WriteFile is issued to that socket, the system
call should return with an error indicating that the socket is
disconnected. Currently, however, the system call returns with
"no data available" which causes the application to wait for an
i/o event which will never come.

PROBLEM CONCLUSION:
The solution is to check to see if the socket is connected at
the time the request is made and error off appropirately if it
is not.

------

APAR: IY23191 COMPID: 5765C3403 REL: 430
ABSTRACT: GDLC CRASH - . TRL.EXT:LANWRTSG +314 STB R30,0X380(R3)

PROBLEM DESCRIPTION:
IAR: . TRL.ext:lanwrtsg +314 (01649eb0): stb r30,0x380(r3)
     . TRL.ext:lanwrtsg +224 (01649dc0)
     . TRL.ext:g_find_self_gen +1e4 (01652c8c)
     . TRL.ext:find_self_gen +5c (01655c28)
     . TRL.ext:add_name_timeout +124 (0164aedc)c28)
     . TRL.ext:link_manager +12c (0164cd18)
     . TRL.ext:link_manager +12c (0164c

PROBLEM CONCLUSION:
Insure that the port stano variable is initialized to NO_MATCH
prior to handling SAP lbusy_ctr and addn_ctr conditions.

------

APAR: IY23196 COMPID: 5765C3403 REL: 430
ABSTRACT: RPCBIND VERSION 3 RETURNS WRONG ADDRESS

PROBLEM DESCRIPTION:
Curing certain network configuration circumstances, portmap
might return the wrong address for version 3/4 requests.
This is usually if the request comes from a network that is
routed between the client and server (not directly connected
to an interface), and this network has no corresponding
interface on the server with the same network prefix.
Also, even if a program registers a specific address with
the portmap, the best matching address will be returned.

LOCAL FIX:
None. Customers who have this problem should downgrade to
use version 2 of the protocol, which does not passback any
address information to the client.

PROBLEM SUMMARY:
The setup had the client on a network 172.28.1.x, and
only network 172.16.1.x configured on the server. Since
172.16.x.x are considered B-class networks, any similarities
in the most significant 16 bits are not considered when
returning the address to the client. This makes portmap pick
any configured interface address for the return value.
Making the application register a specific address instead of
any address didn't help, since the address registered is not
taken into consideration by portmap.

PROBLEM CONCLUSION:
Match on any number of bits when comparing interface addresses
with the configured addresses. This will give a return address
that is as close as possible to the client number-wise.
Also, make sure that requests for services registered with a
specific address will always return that address.

------

APAR: IY23197 COMPID: 5765C3403 REL: 430
ABSTRACT: MAN COMMAND CORE DUMPS

PROBLEM DESCRIPTION:
man command core dumps in save_charset().

PROBLEM SUMMARY:
man command dumps core.

------

APAR: IY23349 COMPID: 576565400 REL: 220
ABSTRACT: TPROF DOES NOT REPORT 64BIT APPLICATIONS CORRECTLY

PROBLEM DESCRIPTION:
tprof does not give the correct results in themixed env. of
32bit and 64bit applications.

PROBLEM CONCLUSION:
fixed the code to handle the Kernel segment and User segments
correctly when both 32bit and 64bit apps. are running

------

APAR: IY23369 COMPID: 5765C3403 REL: 430
ABSTRACT: LARGE QUEUE_DEPTH AND HEAVY FC DISK ACTIVITY CAN CAUSE I/O HANG

PROBLEM DESCRIPTION:
In configurations where a large number of FC disks are
associated with an FC adapter, and the queue_depth setting
for each hdisk is large, the number of commands attempting to
be issued may exceed the capacity of the FC protocol device
to handle them. This may expose a bug in the FC protocol
driver that causes starvation of a few hdisks.

PROBLEM CONCLUSION:
Modify the fairness algorithm for issuing I/O at the FC
protocol driver so that all devices are accounted for when
attempting to issue commands to the devices through the
FC adapter.

------

APAR: IY23370 COMPID: 5765C3403 REL: 430
ABSTRACT: FIXES REQUIRED FOR FUTURE ADAPTER

PROBLEM DESCRIPTION:
Various forms of I/O being suspended/failed. Failure of
failover in some cases. Configuration problems for some
hdisks.

PROBLEM CONCLUSION:
Fix problems with 9401 handler, change handling of ACA Active
and Queue Full, improve failover logic, find correct adapter
for configuration. Also incorporate HACMP-required changes.

------

APAR: IY23371 COMPID: 5765C3403 REL: 430
ABSTRACT: MODIFY RETURN CODE OF FAILED PRLI FROM ENXIO TO ADAPTER STATUS

PROBLEM DESCRIPTION:
When I/O is sent from a target driver to the FC protocol
driver, and a process relogin is required before this I/O is
issued, a failed process login will return a status of ENXIO
for this particular I/O. This will prevent proper failover
to an alternate path when working with particular subsytems,
namely, the disk drivers for use with the LSI Logic
subsystems.

PROBLEM CONCLUSION:
Modify the return status of a failed I/O due to relogin
failure from ENXIO to the status return by the FC adapter
(such as ENETDOWN, ENETUNREACH) to allow specific target
drivers above to properly fail these I/Os and redrive down
an alternate path.

------

APAR: IY23376 COMPID: 5765C3403 REL: 430
ABSTRACT: ATM MPOA - MPCSTAT CANNOT SPECIFY CLIENTS OTHER THAN MPC0

PROBLEM DESCRIPTION:
Unable to get mpcstat output when the MPOA device is not
mpc0

PROBLEM CONCLUSION:
Make mpcstat search the ODM for available mpoa devices.

------

APAR: IY23379 COMPID: 5765C3403 REL: 430
ABSTRACT: ADAPTER DIAGNOSTICS CORE DUMPS WHEN USING MALLOC DEBUG

PROBLEM DESCRIPTION:
  Customers running diagnostics with malloc debug on will
  receive a core dump while it is closing an adapter. The
  message says that an address has already been freed.

PROBLEM CONCLUSION:
  Removed free statements that were freeing memory allocated
  by catalog functions. The call to catclose was attempting
  to free duplicate memory addresses. Memory allocated
  by catalog functions should only be freed by other catalog
  functions.

------

APAR: IY23381 COMPID: 5765C3403 REL: 430
ABSTRACT: ADAPTERS IN 3RD AND 4TH DRAWERS CAN PING BUT CAN'T STREAM DATA

PROBLEM DESCRIPTION:
Some Bigbend adapters in the 3rd and 4th I/O drawer are not
not functional (unable to stream data)

PROBLEM CONCLUSION:
We had limit the maximum size in memory allocation for the
adapter.

------

APAR: IY23383 COMPID: 5765C3403 REL: 430
ABSTRACT: IOCP: LOCKS DURRING TERMINATION

PROBLEM DESCRIPTION:
An application will hang when it tries to exit if the iocp
file descriptor is closed before the individual sockets in
the iocp and at least one of the sockets has an outstand
ReadFile or WriteFile request.

PROBLEM CONCLUSION:
The solution is to clean up properly when a ReadFile or
WriteFile request is terminated because the iocp is being
closed.

------

APAR: IY23385 COMPID: 5765C3403 REL: 430
ABSTRACT: CANNOT GET PROPER IC VALUE FROM 64BIT XLIB WITH WNN6(XWNMO)

PROBLEM DESCRIPTION:
XGetICValues doesn't return correct values from a 64bit X
client with Wnn6.

PROBLEM CONCLUSION:
Fix Xlib to set correct values for long(64bit) IC variables.

------

APAR: IY23386 COMPID: 5765C3403 REL: 430
ABSTRACT: BACKPORT FEATURE 335295 TO SERVICE STREAM

PROBLEM DESCRIPTION:
Ypbind is unable to function when there is no NIS server
on the local subnet.

PROBLEM CONCLUSION:
Allow ypbind to search the file "ypservers" in the directory
/var/ypbind/binding/${DOMAINNAME}/. If the file exists, it
will contain a list of IP addresses which correspond to NIS
servers. These servers will be contacted instead of sending
a broadcast. If the file does not exist, or it none of the
servers listed in the file respond, a broadcast is made as
before.

------

APAR: IY23388 COMPID: 5765C3403 REL: 430
ABSTRACT: ATM 14106001 ADAPTER: SYSDUMP

PROBLEM DESCRIPTION:
An ATM device (atm0) has various clients, Classic IP and LANE.
After all clients have been brought into the 'detach' or
'defined' state, a command sequence like the one below often
causes a dump for the device driver of atm0:
chdev -l at1 -a state=detach
chdev -l at4 -a state=detach
chdev -l en3 -a state=detach
chdev -l tr3 -a state=detach
rmdev -l ent3
rmdev -l tok3
rmdev -l atm0
The dump only occurs when the LANE clients cannot register
with their server( wrong server address), as reported in the
error report by the error below:
LABEL: ATMLE_LOST_CONNECT
IDENTIFIER: 0E365476
The dump does not occur if all exisiting clients have
registered correctly with the switch. However, the above
commands should succeed if connection to the LANE server is
lost, for instance due to a hardware problem.

PROBLEM CONCLUSION:
correct the rmp pointer to the atm reserved_mapper.

------

APAR: IY23389 COMPID: 5765C3403 REL: 430
ABSTRACT: GXT6500P CONTEXT SWITCH HANGS

PROBLEM DESCRIPTION:
GXT6500P may hang under certain conditions when context switch
occurs between X and 3D DWA.

PROBLEM CONCLUSION:
There may be an intermittent problem restoring the geometry
context. To work around this, the device driver must ensure
that restores are always done, even if processes have been
killed or other errors causing an adapter reset have occurred.

------

APAR: IY23390 COMPID: 5765C3403 REL: 430
ABSTRACT: ADD 2048X153630HZ DIGITAL TO GXT6500/4500 ADAPTERS

PROBLEM DESCRIPTION:
GXT6500P and GXT4500P cannot drive hi res digital monitors at
timings faster than 1600x120060Hz.

PROBLEM CONCLUSION:
Add device driver code to support timing of 2048x153630Hz for
digital mode.

------

APAR: IY23391 COMPID: 5765C3403 REL: 430
ABSTRACT: GXT120P/130P GRAPHICS ADPAPTER RENDERS SLOWLY.

PROBLEM DESCRIPTION:
The GXT120P/130P graphic adapters may experience performance
problems when drawing rectangles and or spans.

PROBLEM CONCLUSION:
Wait for the raster engine to be finished after all rendering
commands have been sent to the graphics adapter.

------

APAR: IY23396 COMPID: 5765C3403 REL: 430
ABSTRACT: CDBOOT CORRUPTING LARGE DVD MKCD IMAGES >2GB

PROBLEM DESCRIPTION:
booting/mounting mkcd on DVD with more than 2 GB data fails

PROBLEM CONCLUSION:
replace fseek with fseeko

------

APAR: IY23399 COMPID: 5765C3403 REL: 430
ABSTRACT: GXT135P TU-35 ERRORS

PROBLEM DESCRIPTION:
some resolution might not work on some Rev C card

PROBLEM CONCLUSION:
use two set of MNPs, one for Rev C cards and one for Rev F
cards

------

APAR: IY23401 COMPID: 5765C3403 REL: 430
ABSTRACT: SECURITY: BUFFER OVERFLOW VULNERABILITY IN /USR/BIN/UUQ

PROBLEM DESCRIPTION:
Segmentation faults if invalid parameter is passed with the
'-r' option.

PROBLEM CONCLUSION:
Validate user passed parameters, when the '-r' option is
passed.

------

APAR: IY23402 COMPID: 5765C3403 REL: 430
ABSTRACT: SECURITY: BUFFER OVERFLOW SECURITY VULNERABILITY IN MUXATMD

PROBLEM DESCRIPTION:
Buffer overflow security vulnerabilities which will cause
the muxatmd core dumps.

PROBLEM CONCLUSION:
Replace "sprintf, strcpy" with "snprintf, strncpy" so that
we make sure it won't cause any buffer overflow.

------

APAR: IY23403 COMPID: 5765C3403 REL: 430
ABSTRACT: TT COMMANDS SHOW ERROR WITH SPECIAL CHARACTERS

PROBLEM DESCRIPTION:
ttcp, ttmv, ttrm and tttar commands will not work on
filenames with special characters.

PROBLEM CONCLUSION:
double quote(") was inserted before and after the filename.

------

APAR: IY23408 COMPID: 5765C3403 REL: 430
ABSTRACT: DIAGRPT -O OUTPUT HANGS CONSOLE

PROBLEM DESCRIPTION:
The diagrpt -o output is not properly terminated and the
console must be reset with ctrl-j stty sane ctrl-j.

PROBLEM CONCLUSION:
The ASL interface should not be initialized for the command
line options.

------

APAR: IY23409 COMPID: 5765C3403 REL: 430
ABSTRACT: TO AVOID NETIF_BACKUP RECEIVING DATA ON BOTH THE INTERFACES.

PROBLEM DESCRIPTION:
Customer is running telenet session from diffrent client to
the server on the interfaces configured to netif_backup.
When primary is up after the fail over both the interfaces
are responding to the packets which are coming to the
netif_backup interface. This results in receiving duplicate
packets. At this time user enables tn connection from client
to server, server is receiving duplicate packet after sending
SYN_ACK on first interface and it's sending RST flag
immediatly resulting in closing the tn connection.

PROBLEM CONCLUSION:
Fix the code so that only one adapter is supposed to be
active at a time in the case of netif_backup.

------

APAR: IY23415 COMPID: 5765C3403 REL: 430
ABSTRACT: GLXMAKECURRENT FAILS RUNNING JAVA3D APP

PROBLEM DESCRIPTION:
Window for Java3D test case did not appear when test case
was rerun.

PROBLEM CONCLUSION:
When the initial Java3D test case ended, GLX resource cleanup
occured because of notification by the X server. The client
application itself was still around, though, so when the test
case was restarted, the same X connection was being used.
Unfortunately, when the test case was stopped, some OpenGL
rendering commands were unrendered and the test case's GLX
context was still considered to be active. When the test
case was restarted and it tried calling glXMakeCurrent to
make a new context current, the GLX server code attempted to
flush the rendering commands in the previous context. This
process failed (since the context was still connected to the
window which no longer existed) so the GLX code returned an
X protocol error back to the client. This situation is bad
because it prevents a client from ever making another
context current. Therefore, the glXMakeCurrent code in the
GLX X server extension will be changed to ignore this
particular failure and it will continue to attempt to make
the requested context become current.

------

APAR: IY23416 COMPID: 5765C3403 REL: 430
ABSTRACT: ARP TABLE WIPED OUT FOR ATMS

PROBLEM DESCRIPTION:
some pvc arp output do not show up

PROBLEM CONCLUSION:
continue to make next deivce after detect error.

------

APAR: IY23418 COMPID: 5765C3403 REL: 430
ABSTRACT: NDPD-HOST COREDUMPS

PROBLEM DESCRIPTION:
ndpd-host core dumps with certain routing setups

PROBLEM CONCLUSION:
Use a temporary routing socket in function get_rtphy instead
of using main routing socket

------

APAR: IY23419 COMPID: 5765C3403 REL: 430
ABSTRACT: ARTIFACTS LEFT ON SCREEN USING THE GXT135P

PROBLEM DESCRIPTION:
Artifacts left on the screen after drag and drop when using
Desktop file manager on the GXT135P.

PROBLEM CONCLUSION:
Added a flush adapter at the end of the polysegment code.

------

APAR: IY23420 COMPID: 5765C3403 REL: 430
ABSTRACT: LECSTAT DISPLAYS INCORRECT VCC INFORMATION.

PROBLEM DESCRIPTION:
lecstat displays incorrect VCC information.

PROBLEM CONCLUSION:
The vci field is cast as uchar and should be a ushort.

------

APAR: IY23423 COMPID: 5765C3403 REL: 430
ABSTRACT: CIFSUSERPROC IS STARTED WHEN CONNECTING WITH INVALID USER.

PROBLEM DESCRIPTION:
Enable passthrough authentication server and try to connect
with a valid user on AIX and passthrough server, and with
invalid password. A new process cifsUserProc will be created
with every attempt.

PROBLEM CONCLUSION:
exit from cifsUserProc, if the request from the server is,
just to verify the user existency.

------

APAR: IY23424 COMPID: 5765C3403 REL: 430
ABSTRACT: OPLOCKS ALLOW DATA CORRUPTION BETWEEN 95 AND NT EXCEL USERS.

PROBLEM DESCRIPTION:
if oplockfiles = yes, Fast Connect server allow uses to
modify a shared MS office file (excell, power point) causing
data corruption

PROBLEM CONCLUSION:
allow only one user having full access to the file,
others can have read-access only.

------

APAR: IY23425 COMPID: 5765C3403 REL: 430
ABSTRACT: UNABLE TO SAVE POWER POINT FILES

PROBLEM DESCRIPTION:
Problems with saving a power pt files on Japanese WIN95.

PROBLEM CONCLUSION:
Fixed dos name mapping to handle mangled requests.

------

APAR: IY23426 COMPID: 5765C3403 REL: 430
ABSTRACT: SELECT FILE PROPERTIES CHANGES DATE TO 2497

PROBLEM DESCRIPTION:
Timestamp is wrong

PROBLEM CONCLUSION:
Checking if all the timestamp bits are set or not

------

APAR: IY23427 COMPID: 5765C3403 REL: 430
ABSTRACT: RAM-USAGE TOO HIGH IN DOSNAME TABLE

PROBLEM DESCRIPTION:
cifsServer daemon grows to consume major RAM usage. Can grow
to point of slowing-down the AIX system.

PROBLEM CONCLUSION:
Re-write DOS FileNameTable code to use MBS-strings malloc'd
to exact size needed, instead of using AsciiZString fixed-size
buffers.

------

APAR: IY23428 COMPID: 5765C3403 REL: 430
ABSTRACT: COREDUMP WITH 1024 CHARACTERS

PROBLEM DESCRIPTION:
Server core dumps

PROBLEM CONCLUSION:
Changing the assert statement to support more characters

------

APAR: IY23429 COMPID: 5765C3403 REL: 430
ABSTRACT: RAS: OUTPUT CIFS-BUILD INFO TO CIFSLOG

PROBLEM DESCRIPTION:
CIFS customers needing service may need to supply extra
data-collection, etc. to determine exact CIFS-version.

PROBLEM CONCLUSION:
Change cifsServer Makefile to timestamp every CIFS build,
and output that build-time to cifsLog when starting CIFS.

------

APAR: IY23438 COMPID: 5765C3403 REL: 430
ABSTRACT: ENTSTAT -D STATISTICS ARE NOT CORRECT ON ETHERCHANNEL

PROBLEM DESCRIPTION:
entstat does not show correct number of packets transmitted
in case of etherchannel

PROBLEM CONCLUSION:
code was fixed to correct this problem

------

APAR: IY23443 COMPID: 5765C3403 REL: 430
ABSTRACT: DIAGRPT CORE DUMP WHEN RUNNING DIAGNOSTICS

PROBLEM DESCRIPTION:
diagrpt -a causes core dump if a catalog file is missing
for a device.

PROBLEM CONCLUSION:
Give correct failure code when catalog is not found

------

APAR: IY23448 COMPID: 5765C3403 REL: 430
ABSTRACT: ALT_DISK BOOT IMAGES LACK SAN SUPPORT

PROBLEM DESCRIPTION:
The customer may get an LED 554 during fiber channel boot
after alt_disk mksysb install.

PROBLEM CONCLUSION:
Added FC boot support to alt_disk_install boot images.

------

APAR: IY23451 COMPID: 5765C3403 REL: 430
ABSTRACT: CC0D3B59 ERROR HAS WRONG DESCRIPTION LABEL IN ERRPT

PROBLEM DESCRIPTION:
Incorrect error message displayed in the AIX error log.
Displayed ADAPTER ERROR (wrong), should be PROBLEM RESOLVED

PROBLEM CONCLUSION:
Error message file used by the driver was changed to correct
the problem

------

APAR: IY23452 COMPID: 5765C3403 REL: 430
ABSTRACT: GXT135P: UPRINTABLE CHARACTERS MESSAGE IN BOOTLOG

PROBLEM DESCRIPTION:
GXT135P stderr messages in bootlog contain unprintable
characters.

PROBLEM CONCLUSION:
CFLAGS += -g -DDEBUG -DDEBUG_BUS -DCFGDEBUG needs
to be turned off

------

APAR: IY23453 COMPID: 5765C3403 REL: 430
ABSTRACT: SERIALIZATION PROBLEM IN DLOPEN,DLCLOSE,DLERROR AND DLSYM

PROBLEM DESCRIPTION:
serialization problem in dlopen,dlclose,dlerror and dlsym which
causes the application to crash.

PROBLEM CONCLUSION:
Order of mutex lock initialization changed to fix this problem.

------

APAR: IY23561 COMPID: 5765D5101 REL: 121
ABSTRACT: PHOENIX.SNAP ENHANCEMENTS

PROBLEM DESCRIPTION:
phoenix.snap enhancements

PROBLEM SUMMARY:
phoenix.snap should be able to run in a non-SP environment
(HACMP/ES)

PROBLEM CONCLUSION:
phoenix.snap will now collect rsct data specifc to HAMP/ES
regardless of whether it is on an SP.

------

APAR: IY23563 COMPID: 5765D5101 REL: 121
ABSTRACT: FALSE FAIL_STANDBY EVENT IN HAES441

PROBLEM DESCRIPTION:
false fail_standby event in haes441

PROBLEM SUMMARY:
A problem in RSCT Topology Services results in
"false" fail_standby events in HACMP/ES, and
possibly other forms of "false" failed adapter
indications. The problem occurs due to an
incorrect interaction (a race condition)
between the Topology Services NIM process and
the netmon (adapter status determination) library.
The following sequence may be observed in the
Topology Services NIM process log -- located at
/var/ha/log/nim.topsvcs.<interface>.<cluster_name> .00X --
up to a few minutes after the false adapter
indication:
10:04:17: Received an OPEN command <= (3)
10:04:17: Minimum packet size: 76
 ...
10:04:17: Local address: 192.114.129.195
 ...
10:04:17: Adapter has broadcast capability. Address
192.114.129.255.
10:04:17: Broadcast address is 192.114.129.255.
10:04:17: Bind to broadcast address succeeded.
10:04:17: Response successfully sent.
10:04:17: Sleep for 5.0 sec.
10:04:17: netmon_thread_main: got a netmon request. <= (2)
10:04:17: Netmon thread: file descriptors closed. <= (1)
The occurrence of the "file descriptors closed" (1)
entry right after the "got a netmon request" (2)
entry, a few lines AFTER the
"Received an OPEN command" (3) entry gives an indication
that the problem is being seen.

PROBLEM CONCLUSION:
With the fix, the false fail_standby event
in HACMP/ES should no longer occur.
When examining the Topology Services NIM
process log -- located at
/var/ha/log/nim.topsvcs.<interface>.<cluster_name> .00X --
the "file descriptors closed" entry should NOT occur
right after the "got a netmon request" entry (when the
latter occurs a few lines after the "Received an OPEN
command"
entry)
See example below:
10:51:41: Received an OPEN command
10:51:41: Minimum packet size: 76
 ...
10:51:41: Adapter has broadcast capability. Address
9.114.61.255.
10:51:41: Broadcast address is 9.114.61.255.
10:51:41: Bind to broadcast address succeeded.
10:51:41: Response successfully sent.
10:51:41: Sleep for 5.0 sec.
10:51:41: netmon_thread_main: got a netmon request.
10:51:42: Receive thread (re-)enabled.
10:51:42: netmon response: 2
10:51:42: netmon response: Adapter is up

------

APAR: IY23564 COMPID: 5765D5101 REL: 121
ABSTRACT: KEEP 2-3 LOG FILES WITH HIGH PRIORITY

PROBLEM DESCRIPTION:
keep 2-3 log files with high priority

PROBLEM SUMMARY:
The problem is due to that the original method to assign
priorities to the log and core files was designed
for keeping only one (highest priority) log file
in case there is no enough space.

PROBLEM CONCLUSION:
After we change the method to assign the priorities
it should keep at least 3 log files with highest
priorities and one recent core file together with the
corresponding log files.

------

APAR: IY23565 COMPID: 5765D5101 REL: 121
ABSTRACT: MISSING SUBSCRIPTION GROUP - ML0MEMBERSHIP

PROBLEM DESCRIPTION:
Missing subscription group - ml0Membership

PROBLEM SUMMARY:
The group ml0Membership group is missing
from the sample_test program. Thus the
users are having the difficult to test
the ml0Membership using the sample_test.

PROBLEM CONCLUSION:
This adds the ml0Membership group to the sample_test
so that the users can easily test the subscription
to the ml0Membership.

------

APAR: IY23571 COMPID: 5765D5101 REL: 121
ABSTRACT: TOO SHORT OF LOCALNAME FOR HA_GS_JOIN

PROBLEM DESCRIPTION:
too short of localname for ha_gs_join

PROBLEM SUMMARY:
An application uses ha_gs_join() to form a Group
Services group. Among the parameters, a field
gs_provider_local_name is limited to 16 bytes
unlike the other group names (limited to 32 bytes).
This max length is checked at ha_gs_join() and
returns an error code if it exceeds. This
different max values (and short length of local_name)
cause the confusion to the applications.
Furthermore, the gs_provider_local_name is used
only for the internal tracking.

PROBLEM CONCLUSION:
This fix will truncate the gs_provider_local_name
with 32bytes and prevent an error msg.

------

APAR: IY23697 COMPID: 5765C3403 REL: 430
ABSTRACT: DEADLOCK IN LOCKD CODE

PROBLEM DESCRIPTION:
During exceptional circumstances, a deadlock can occur in
the rpc.lockd code. The two stack traces deadlocking each
other are:
IAR: .slock_ppc+25c (0014f270): cmpi cr0,0x0,r19,0x0
LR: .slock_ppc+25c (0014f270)
f04495e8: . nfs.ext:lm_retry +38 (05572974)
f0449638: .lock_wakeup+78 (001e916c)
f0449688: .flckadj+28c (001e9748)
f0449718: .common_reclock+334 (001ea0f8)
f04497c8: .jfs_lockctl+ac (00216cb0)
f0449838: .vnop_lockctl+1c (001c149c)
f0449878: . nfs.ext:lm_unlock +d0 (0557abb4)
f04498f8: . nfs.ext:lm_nlm4_dispatch +348 (0557cb08)
f04499d8: . nfs.ext:svc_getreq +5ac (0557586c)
f0449f78: .threadentry+18 (00097e9c)
f0449fb8: .thread_terminate+2fc (00012248)
IAR: .slock_ppc+25c (0014f270): cmpi cr0,0x0,r19,0x0
LR: .slock_ppc+25c (0014f270)
f0269668: .simple_lock+64 (00009564)
f02696a8: . nfs.ext:lockfh_walk_vnodes +34 (0554f2d4)
f0269738: .vn_search+48 (0022ee98)
f0269788: . nfs.ext:lockfh_walk_vfs +28 (0554f4c4)
f02697c8: .vfs_search+54 (0022f524)
f0269818: . nfs.ext:unlock_lockmgr_granted +24 (0554ee64)
f0269868: . nfs.ext:lm_unlock_client +d0 (0557366c)
f02698b8: . nfs.ext:lm_crash +a0 (05583260)
f0269908: . nfs.ext:lm_nlm_dispatch +368 (05585314)
f02699d8: . nfs.ext:svc_getreq +5ac (0557586c)
f0269f78: .threadentry+18 (00097e9c)
f0269fb8: .thread_terminate+2fc (00012248)

PROBLEM SUMMARY:
lockd may deadlock

PROBLEM CONCLUSION:
Take the function causing the problem out of the lock.

------

APAR: IY23987 COMPID: 5765C3403 REL: 430
ABSTRACT: CHANGE FAILOVER TIMING AND SEQUENCE

PROBLEM DESCRIPTION:
Systems may fail to properly switch paths, or may experience
delays doing so.

PROBLEM CONCLUSION:
Wait longer (25 sec vs. 10 sec) when a 9136 error is seen.
Stop trying to guess the best path to switch to and simply
alternate paths once there's been a switch failure.

------

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