APAR: IY24556 COMPID: 5765E5400 REL: 440
ABSTRACT: CSPOC.LOG SHOWS ENCRYPTED PASSWORD

PROBLEM DESCRIPTION:
cspoc.log shows the encrypted password
when running clchkspuser/clpasswdremote.

PROBLEM SUMMARY:
cspoc.log shows encrypted password

PROBLEM CONCLUSION:
Do not print arguments sent to clpasswdremote in
script log_output.

------

APAR: IY25844 COMPID: 5765E5400 REL: 441
ABSTRACT: RESOURCE GROUPS ARE PROCESSED IN ALPHABETICAL ORDER, RATHER THAN

PROBLEM DESCRIPTION:
A change of resource group processing was introduced in 4.4.1.
Resource groups are processed in alphabetical order, rather than
by priority first and then alphabetically within the priority.
This will be changed so resource groups will be processed by
priority first and then alphabetically within the priority.

PROBLEM SUMMARY:
umount fails with busy because a resource group with an
application was not stopped first.

PROBLEM CONCLUSION:
resource groups will be processed by priority first and then
alphabetically within the priority.

------

APAR: IY26619 COMPID: 5765C3403 REL: 430
ABSTRACT: KSH NOT USING HASH TABLE FOR PATHNAME RESOLUTION

PROBLEM DESCRIPTION:
Hash function in 'ksh' does not work.
Lookup is performed for each element of PATH even though a
command is already in the hash table.

LOCAL FIX:
use full pathname for commands.

PROBLEM SUMMARY:
ksh searches the PATH variable for tracked aliases. This
is unnecessary since the path of the tracked command is
available to the shell.

PROBLEM CONCLUSION:
Shell searches the PATH variable for tracked aliases.
Code changes have been done so as to avoid searching into
PATH when the command invoked is a tracked alias.

------

APAR: IY27393 COMPID: 5765C3403 REL: 430
ABSTRACT: FILTER_ALL_LATESTS CAUSES CORE DUMP WHEN FILESET HAS SPACES IN

PROBLEM DESCRIPTION:
When trying to install software from LATEST available software,
the function filter_all_latest will core dump when it
encounters a fileset with a space in its name.

LOCAL FIX:
Install the fileset using the command line.

PROBLEM SUMMARY:
From smitty install_latest menu, after entering device,
attempting to list with F4 will core dump in filter_all_latest
with the following error:
1800-106 An error occurred:
/usr/lib/instl/sm_inst 12 : 29876 Segmentation fault(coredump)
#-------------------------------------------------------------
# No installable software products were found on the media.
#-------------------------------------------------------------

PROBLEM CONCLUSION:
The code in /usr/lib/instl/filter_all_latest
will be changed to handle
names (ISMP) containing spaces.

------

APAR: IY28229 COMPID: 5765E5400 REL: 440
ABSTRACT: CORRECT LIMIT ON CYCLE VALUE

PROBLEM DESCRIPTION:
In the SMIT panel for custom net module configuration, the
failure cycle limit, 1-30, is incorrect. Changing the limit
to 1-21474.

PROBLEM SUMMARY:
In the SMIT panel for custom net module configuration, the
failure cycle limit, 1-30, is incorrect. Changing the limit
to 1-21474.

PROBLEM CONCLUSION:
Changed the failure cycle limit in claddnim.c from 1-30
to 1-21474.

------

APAR: IY28503 COMPID: 5765E8200 REL: 230
ABSTRACT: CFGGMD SMIT PROBLEM, CFGGMD/STARTGMD NOT FOUND

PROBLEM DESCRIPTION:
Problem with smit panels:
When trying to Start ALL GeoMirror Devices Stopped -> Available
out of the GeoMirror Utilities ksh complains about startgmd
not being found.
Configure ALL GeoMirror Devices (Defined -> Stopped)
results in ksh: cfggmd: not found.

PROBLEM SUMMARY:
Problem with smit panels:
When trying to Start ALL GeoMirror Devices Stopped -> Available
out of the GeoMirror Utilities ksh complains about startgmd
not being found.
Configure ALL GeoMirror Devices (Defined -> Stopped)
results in ksh: cfggmd: not found.

PROBLEM CONCLUSION:
Include the full path name to startgmd.

------

APAR: IY28778 COMPID: 5765C3403 REL: 430
ABSTRACT: ADDING 'STOPSEARCH' CAPABILITY TO GROUP ROUTING FEATURE

PROBLEM DESCRIPTION:
As per group-routing feature, if a route has a group
associated with it that's denied access to this route,
the routing table is still looked up further eventhough
the group routing test fails for this route.
We need to add an option to route command such that a
flag can be set on a route. If this flag is set, and if
the routing table lookup matches this route for a group,
but is not allowed to use this route because of group
routing restrictions, then the routing lookup should
stop without further search.

PROBLEM CONCLUSION:
A new option '-stopsearch' is added to the route command.
Setting this option for a route using the 'route' command will
cause the 's' bit to be set (netstat -rn output shows this
flag). If this bit is set for a route, and if a group is
denied access to this route, the routing table will not be
looked up further for default route, thereby restricting access
to the members of that group.

------

APAR: IY28786 COMPID: 5765E5400 REL: 440
ABSTRACT: CLLSFS COMMAND MIGHT FAIL IF ODM IS CORRUPTED HACMP,HAES

PROBLEM DESCRIPTION:
If certain entries are deleted from ODM database
cllsfs command might fail with error.
So all other commands dependent on cllsfs will fail too.
So will the smit screens that need to list the file systems.
ex: Change/Show characteristics of a shared file system.

LOCAL FIX:
cllsfs command is changed to work thru the ODM
inconsistencies and prepare a list of file systems
Affected areas: hacmp, hacmpES

PROBLEM SUMMARY:
If certain entries are deleted from ODM database
cllsfs command might fail with the following error:
simp_get_fsinfo: malloc failed.
So all other commands dependent on cllsfs will fail too.
So will the smit screens that need to list the file systems.
ex: Change/Show characteristics of a shared file system.

PROBLEM CONCLUSION:
cllsfs command is changed to work thru the ODM
inconsistencies and prepare a list of file systems
    Affected areas: hacmp, hacmpES

------

APAR: IY28917 COMPID: 5765E5400 REL: 440
ABSTRACT: IP DISCOVERY FAILS HAS/HAES

PROBLEM DESCRIPTION:
When configuring your topology with SMIT, there is
a panel for "Discover Current Network Configuration"
which allows you to choose "Local Configuration".
This issues th cl_harvestIP_scripts command, which
sometimes will fail with "cl_harvestIP_scripts:
Problem accessing the configuration file"

PROBLEM CONCLUSION:
Some of the routines in the cl_harvestIP_scripts script are
run in the background. In some cases, the results of these
routines are needed before the routine actually finishes,
which causes the error. The backgrounding of these routines
will be fixed so that the routine is guaranteed to be
finished when the data is needed.

------

APAR: IY28929 COMPID: 5765B8100 REL: 230
ABSTRACT: VARIOUS FIXES FOR SS7_ISUP

PROBLEM DESCRIPTION:
Various fixes to SS7_ISUP for directtalk

------

APAR: IY28942 COMPID: 5765E5400 REL: 440
ABSTRACT: NEW FILESYSTEM NOT MOUNTED AFTER DARE.

PROBLEM DESCRIPTION:
A resource group uses mount all filesystems and a new
filesystem is added to a volume group defined in that resource
group. When resources are synchronized the new filesystem is
not mounted.

PROBLEM SUMMARY:
After adding a new file system to a VG synchronize resources
will not do a DARE because no change detected.

PROBLEM CONCLUSION:
save cksum if mount all so changes can be detected in a
resource group.

------

APAR: IY28977 COMPID: 5765C3403 REL: 433
ABSTRACT: ROUTING TABLE NEEDS TO BE UPDATED, WHEN AN IP ADDRESS IS DELETED

PROBLEM DESCRIPTION:
Routing table needs to be update with the correct outgoing
interface address, when an ip address is deleted/changed.

PROBLEM SUMMARY:
When we change/delete an ip address the routing table will
not be updated to reflect the changed address.

PROBLEM CONCLUSION:
Added a function to update the routing table, when an ip
address is changed/deleted.

------

APAR: IY28992 COMPID: 5765C3403 REL: 433
ABSTRACT: VMOUNT: A SYSTEM CALL RECEIVED A PARAMETER ... NOT VALID

PROBLEM DESCRIPTION:
User may experience trouble mounting nfs filesystems if using
lpp and spot resources that do not match at the fix level vrmf.
The error received is a vmount error.

PROBLEM SUMMARY:
User may experience trouble mounting nfs filesystems if using
lpp and spot resources that do not match at the fix level vrmf.

PROBLEM CONCLUSION:
NIM should make a copy of nfsmnthelp prior to copying over the
nfsmnthelp from the SPOT. During firstboot, the original file
should replace the SPOT version.
The code was changed to do the above.

------

APAR: IY29027 COMPID: 5765E5400 REL: 440
ABSTRACT: NFS CLIENT APPLICATIONS AFFECTED BY SOME UNNECESSARY STOPPING

PROBLEM DESCRIPTION:
The customer has been experiencing problems using NFS and
locking within HACMP clusters. These problems are interruptions
to their NFS clients/applications caused by what has been
determined to be unnecessary stopping of the NFS daemons,
specifically nfsd and rpc.mountd in the HACMP scripts. Also,
sometimes stopping of the rpc.lockd when not required which
can also have some affect on whether locks are successfully
reclaimed or not.

PROBLEM CONCLUSION:
All scripts which were stopping the nfsd and rpc.mountd
were changed to no longer stop these daemons. Likewise,
all scripts that were restarting these daemons had this
removed as well. Also, a change was made to not stop the
rpc.lockd when umounting filesystems until after a few
attempts have failed and only for filesystems that are
exported.

------

APAR: IY29064 COMPID: 5765E5400 REL: 440
ABSTRACT: SCRIPT FAILURE TRYING TO MOUNT ALREADY MOUNTED FILESYSTEM DURING

PROBLEM DESCRIPTION:
The customer forced down a node owning a resource group which
had filesystems=All specified. When attempting to bring the
node back up into the cluster, it failed due to attempting to
mount the filesystems which were already still mounted.

PROBLEM CONCLUSION:
The get_disk_vg_fs script was modified to check to see if
filesystems obtained from volume groups due to ALL option
are not already mounted before attempting to send them on to
be mounted by cl_activate_fs. The cl_activate_fs script also
was modified to check if filesystems passed in were already
mounted since HAGEO calls that script.

------

APAR: IY29065 COMPID: 5765E8200 REL: 230
ABSTRACT: GREP BROKEN IN GEO_MOUNT_FS

PROBLEM DESCRIPTION:
A call to grep in Geo_mount_fs.sh breaks because
of a text string longer than 2048 characters.

PROBLEM CONCLUSION:
Replaced grep with awk in script. Awk doesn't have a
problem with strings longer than 2048.

------

APAR: IY29070 COMPID: 5765C3403 REL: 430
ABSTRACT: GXT135P SLOW GRAPHICS WITH XFIG AND KDE KONSOLE SCROLLING

PROBLEM DESCRIPTION:
GXT135P graphics adapter shows slow rendering response,
compared to GXT130P. This can be seen with xfig application,
when moving xfig window around. Also aixterm scrolling also
shows problem.

PROBLEM CONCLUSION:
Change logic for pixmap caching to avoid unnecessary tiling,
and make better use of cache memory.

------

APAR: IY29071 COMPID: 5765E5400 REL: 440
ABSTRACT: SMIT ERROR, FAILURE TO TRANSLATE JA_JP (HAS,HA/ES)

PROBLEM DESCRIPTION:
If you are running HACMP with the Ja_JP locale, you will
run into a translation problem when using the cspoc SMIT
panel "Change / Show Characteristics of a Shared File
System in the Cluster". This panel is loaded when you
choose a resource group from a pop-up panel. However,
the data that is loaded includes two "yes,no" fields
which do not get properly translated into Ja_JP. It also
causes a 1800-083 error message, which will show up on
the screen and in the smit.log. This error message
shows up for the "AutoMount" disc_field_name and the
"Acct" disc_field_name.

PROBLEM CONCLUSION:
Change the details of the definition of those fields within
that SMIT panel to allow the proper translation to occur
before displaying the yes/no values.

------

APAR: IY29076 COMPID: 5765C3403 REL: 433
ABSTRACT: PROCESS IDS ARE MISSING FROM THE PID LIST OF SVMON WITH SL

PROBLEM DESCRIPTION:
svmon with -Sl option doesn't list all the
Process Ids associated with those segments.
Some of the segments which are associated
with process are also shown as Unused segments.

PROBLEM SUMMARY:
svmon -S -l reports some of the segments as unsued segments
even though these segments are associated with some
process.
This is due to a problem in svmon handling the vsid's which
doesn't have an associated esid.

PROBLEM CONCLUSION:
svmon is fixed to handle properly the segments which has
a vsid but not an esid.
This fixes the problem with the svmon -S -l reports.

------

APAR: IY29152 COMPID: 5765E5400 REL: 440
ABSTRACT: GET LOCAL AND GLOBAL NETWORK_DOWN WHEN ADAPTERS UNPLUGGED OR

PROBLEM DESCRIPTION:
The customer had a three node configuration with 4 IP networks
all attached through the same ethernet switch. When they
powered off the switch, expecting to just get global
network_down events for each network, instead they got a
swap_adapter event, several local network_down, global
network_downs, fail_standby, and rg_move events.

PROBLEM SUMMARY:
Disrupt a network connection, for example by powering off an
ethernet switch.
Instead of generating a single, global network down event,
hacmp generates several local network events, swap adapter
events and possibly rg move events.

PROBLEM CONCLUSION:
Coordinate the network events better such that only a single
global network event is generated (if possible).

------

APAR: IY29191 COMPID: 5765C3403 REL: 430
ABSTRACT: EXPR DOES NOT WORK PRPPERLY WHEN THE BRE CONTAINS $\

PROBLEM DESCRIPTION:
expr '$.' : "$." yields 0, and should yield 2

PROBLEM SUMMARY:
The expr command treats '$' wrongly in BREs

PROBLEM CONCLUSION:
The bug was traced back to a libc routine. libc code was
modified to fix this.

------

APAR: IY29269 COMPID: 5765E5400 REL: 440
ABSTRACT: HAES: NETMASK NOT FILLED IN SO RC.CLUSTER USES WRONG DATA

PROBLEM DESCRIPTION:
netmask information is not stored in the HACMPadapter ODM
class. But rc.cluster gets the netmask for here. The logic in
rc.cluster looks for boot addresses on the css0 interface and
if not found will ifconfig the base address up.

LOCAL FIX:
Fill the netmask field in HACMPadapter class
which corresponds to the value in subnet class.

PROBLEM CONCLUSION:
Fill the netmask information in the HACMPadapter ODm class
so it can be picked up by rc.cluster.

------

APAR: IY29322 COMPID: 5765C3403 REL: 430
ABSTRACT: MKTIME() PERFORMANCE ISSUE

PROBLEM DESCRIPTION:
mktime() routine uses iterative looping for
converting months out of range (<0 and >11)
into appropriate years. This would raise
a considerable performance degradation incase
the input tm structure has huge invalid values.

LOCAL FIX:
use modulus and division calculation for
converting months into appropriate years
instead for looping in a piece of code for
several iterations.

PROBLEM SUMMARY:
if the tm_mon field of tm structure is very huge, then mktime
will spend lot of time in a loop to bring the tm_mon value
to the range 0-11, adjusting the tm_year field.

PROBLEM CONCLUSION:
Looping has been replaced with appropriate mathematical
calculation, to avoid delay

------

APAR: IY29404 COMPID: 5765C3403 REL: 430
ABSTRACT: SYNC /USR/INCLUDE/RPCSVC/MOUNT.H

PROBLEM DESCRIPTION:
programs calling vmount fail with errno 22 (invalid parm)

PROBLEM CONCLUSION:
Syncronize /usr/include/rpcsvc/mount.h to reflect internal
NFS changes for nfs_args structure and NFSMNT_ACL. Also
change the kernel extension code that handles backwards
compatibility for nfs_arg structures from the user space to
avoid recompilation of user space code.

------

APAR: IY29451 COMPID: 5765C3403 REL: 430
ABSTRACT: PIODIGEST MAY FAIL TO CREATE PS QUEUE

PROBLEM DESCRIPTION:
piodigest may fail when creating a postscript
queue if a very large number of fonts reside
in /usr/lib/ps

PROBLEM SUMMARY:
piodigest may fail when creating a postscript queue if a large
number of fonts reside in /usr/lib/ps

PROBLEM CONCLUSION:
piodigest has been fixed so it can handle the increased number
of fonts

------

APAR: IY29454 COMPID: 569692600 REL: 110
ABSTRACT: CRASH IN THE FRAME LAYER FR_RNR_NET FUNCTION WHEN RNR FRAMES

PROBLEM DESCRIPTION:
Customer is receiving RNR constantly,
the fr_rnr_net function calls the fr_rej_net
that frees the message and return.
fr_rnr_net use the same conditions than
frn_rej_net to free the message again,
causing a crash

PROBLEM SUMMARY:
When RNRs are being handle with the poll final bit set
the machine will crash in the frame layer fr_net_rnr that
calls the freemsg

PROBLEM CONCLUSION:
We implement a flag that check if the path taken in the
fr_net_rej function called by the fr_net_rnr is taken.
If it is, we are not going to let the fr_net_rnr frees the
message because it was already freed in the fr_net_rej.

------

APAR: IY29525 COMPID: 5765C3403 REL: 430
ABSTRACT: GXT4500P MODELS INCORRECTLY RENDERED IN ENOVIA PORTAL

PROBLEM DESCRIPTION:
Environment 3DCOM (Enovia Portal) V5R7, AIX433 gold, GXT4500P.
When rendering with a V5 server and Local OpenGL viewer in
shaded and shaded plus edges modes, models are not complete.
This problem is exhibited when running Enovia Portal in
netscape, but not when running the Enovia as a java application
outside of netscape.

PROBLEM SUMMARY:
Environment 3DCOM (Enovia Portal) V5R7, AIX433 gold, GXT4500P
(and GXT4000P). When rendering with a V5 server and Local
OpenGL viewer in shaded and shaded plus edges modes, models
are not complete. This problem is exhibited when running
Enovia Portal in Netscape, but not when running the Enovia
as a java application outside of netscape.

PROBLEM CONCLUSION:
Zero length draw arrays were being handled incorrectly and
would not allow following draw array calls to be executed.
A check for zero length arrays was inserted into the draw
array functions to ensure a clean exit from the functions.

------

APAR: IY29542 COMPID: 5765E5400 REL: 440
ABSTRACT: BRING RESOURCE GROUP OFFLINE STICKY AND SOME OTHER DARE

PROBLEM DESCRIPTION:
The customer ran the resource management function to bring a
resource group offline with sticky indicated. This timed out
in the smit panel and though the resource group resources had
all been released, the status in clfindres still indicated
it as UP on the node but with STOP indicated in sticky location.
Other problems reported were failure to return a CWOF
resource group to its home node using bring resource group
online or cldare with default, and failure in moving
a rotating resource group with sticky indicated. These
problems were caused by a incomplete or "bad" fix for
a previous APAR IY27187.

PROBLEM SUMMARY:
***************************************************************
*USERS AFFECTED: *
* All users with the following filesets at these levels *
* cluster.base.server.events 4.4.1.6dres still indicated it
* cluster.base.server.events 4.4.1.7ted in sticky location.
* cluster.es.server.events 4.4.1.6 to return a CWOF resource
* cluster.es.server.events 4.4.1.7esource group online or
***************************************************************
group with sticky indicated. These problems were caused by a
incomplete or "bad" fix for a previous APAR IY27187.
*PROBLEM DESCRIPTION: *
* The customer ran the resource management function to
* bring a resource group offline with sticky indicated.
* This timed out in the smit panel and though the
* resource group resources had all been released, the
* status in clfindres still indicated it as UP on the
* node but with STOP indicated in sticky location.
* Other problems reported were failure to return a CWOF
* resource group to its home node using bring resource
* group online or cldare with default, and failure in
* moving a rotating resource group with sticky
* indicated. These problems were caused by a incomplete
* or "bad" fix for a previous APAR IY27187.
***************************************************************
*RECOMMENDATION: *
* Apply apar IY29542
***************************************************************

PROBLEM CONCLUSION:
The reconfig_resource_release script was modified to include
only TRANSIENT_LOCATION_REQUEST and STICKY_LOCATION_REQUEST
resource groups in the CURRENT_GROUPS and NEW_GROUPS and then
only if they had appeared in these variables previously from
the output of clsetenvgrp.
In addition, in HAES only, a change was made to clrgpa.C so
that the dare migration target node would be set to the
highest priority node that is currently active in the cluster.

------

APAR: IY29625 COMPID: 5765E5400 REL: 440
ABSTRACT: HAES: APPLY SNAPSHOT WILL LOSE CLUSTER.LOG REDIRECT FOR

PROBLEM DESCRIPTION:
daemon.notice output not redirected after applying a snapshot

PROBLEM CONCLUSION:
Restore daemon.notice file in clsnapshot.

------

APAR: IY29630 COMPID: 5765E5400 REL: 440
ABSTRACT: HACMP SHOULD NOT SHUT OFF GSCLVMD IF RVSD IS USING IT

PROBLEM DESCRIPTION:
If HACMP ESCRM is stopped, it will automatically stop the
gsclvmd daemon. However, in situations where both HACMP
and RVSD are run on the same node, gsclvmd may e in use
by RVSD. This shuts off all RVSD I/O, which is not the
intended consequence of shutting down HACMP.

PROBLEM CONCLUSION:
clstop should check, in the path executed on HACMP ESCRM, that
gsclvmd is using the group services stack, rather than the
hats stack. If HACMP ESCRM started gsclvmd, it will be using
the group services stack. If gslcvmd is using the hats stack,
its in use by RVSD.

------

APAR: IY29635 COMPID: 5765E5400 REL: 440
ABSTRACT: CLCONVERT SNAPSHOT ODM BROKEN

PROBLEM DESCRIPTION:
Converting an HAES431 snapshot to HAES441 should
remove the clresmgrd.log stanza from the HACMPlogs
class of ODM.

PROBLEM CONCLUSION:
Code must be added to HAES431toHAES44 to remove the
clresmgrd.log stanza from the HACMPlogs class of ODM.

------

APAR: IY29638 COMPID: 5765E5400 REL: 440
ABSTRACT: HACMP: ATM SWAP_ADAPTER DOESN'T ACTUALLY SWAP THE ADAPTERS

PROBLEM DESCRIPTION:
swap_adapter for a Classical IP network does not actually
swap the adapter.

PROBLEM CONCLUSION:
Fix cl_swap_ATM_IP_address to call ifconfig with the correct
interfaces.

------

APAR: IY29640 COMPID: 5765E5400 REL: 440
ABSTRACT: HAES: CLSTRMGR CORE DUMP WHEN MISC DATA EXCEEDS 1024

PROBLEM DESCRIPTION:
If the misc data for any resource group exceeds 1024 bytes,
the clstrmgr will dump core.

PROBLEM CONCLUSION:
Check the length of the misc data before processing

------

APAR: IY29643 COMPID: 5765E5400 REL: 440
ABSTRACT: HAES: REMOVE PERIODS FROM IFS OWNERSHIP ENVIRONMENT VARIABLES

PROBLEM DESCRIPTION:
No symptoms, the IFS ownership environment variables are
not passed in the environment to child processes (i.e.
sub-event scripts) because of the periods in the
environment.

PROBLEM CONCLUSION:
The periods will translated to 'x' like they are in the IFS
state environment variables.

------

APAR: IY29644 COMPID: 5765E5400 REL: 440
ABSTRACT: HAES: NODE_ID AND HANDLE MISMATCH CAUSE INVALID RG_MOVE

PROBLEM DESCRIPTION:
If the node_id and node_handle in HACMPnode do not match,
then erroneous rg_move events or no rg_move events at all
will happen after local_network_down events.

PROBLEM CONCLUSION:
Replace the incorrect use of node_id with node_handle when
examining affected resource groups after a
local_network_down.

------

APAR: IY29645 COMPID: 5765E5400 REL: 440
ABSTRACT: HAES CLUSTER MANAGER CORE DUMPS WHEN HAGEO SITES ARE DEFINED

PROBLEM DESCRIPTION:
HAES cluster manager core dumps and halts node when
HAGEO sites are defined. There is no consistent time
or action which provokes the problem, but there is a
consistent symptom in that the output of the clRGinfo
command will be corrupted, that is, it will show an
innacurate number of nodes for a resource group.

PROBLEM CONCLUSION:
clstrmgr was freeing space that was never allocated.
Fix the code to not free the space unless allocated.

------

APAR: IY29646 COMPID: 5765C3403 REL: 430
ABSTRACT: GETGRSET WITH NISPLUS RETURNS BAD DATA

PROBLEM DESCRIPTION:
getgrset() returns bad data for users who have logged in using
NISPLUS authentication. This will cause the "groups" command to
give bad output data.

PROBLEM SUMMARY:
***************************************************************
*USERS AFFECTED: *
* All users with the following filesets at these levels *
* bos.adt.prof 4.3.3.75
* bos.adt.prof 4.3.3.76
* bos.adt.prof 4.3.3.77
* bos.adt.prof 4.3.3.78
* bos.adt.prof 4.3.3.79
* bos.adt.prof 4.3.3.80
* bos.adt.prof 4.3.3.81
* bos.adt.prof 4.3.3.82
* bos.rte.libc 4.3.3.75
* bos.rte.libc 4.3.3.76
* bos.rte.libc 4.3.3.77
* bos.rte.libc 4.3.3.78
* bos.rte.libc 4.3.3.79
* bos.rte.libc 4.3.3.80
* bos.rte.libc 4.3.3.81
* bos.rte.libc 4.3.3.82
* bos.rte.libc 4.3.3.83
***************************************************************
*PROBLEM DESCRIPTION: *
* getgrset() returns bad data for users who have logged
* in using NISPLUS authentication. This will cause the
* groups command to give bad output data.
*
* This problem is found to exis with AFS and DCE user
* authentication as well where in the groups command is
* not returning all the groups. Only the first group
* member is displayed.
***************************************************************
*RECOMMENDATION: *
* Apply apar IY29646
***************************************************************

PROBLEM CONCLUSION:
Remove unitialized variable reference in nisplus_getgrset()

------

APAR: IY29656 COMPID: 5765E5400 REL: 440
ABSTRACT: C-SPOC ADD FILESYSTEM RETURNS 1800-106 AN ERROR OCCURED.

PROBLEM DESCRIPTION:
Using C-SPOC to Add A Journaled File System to a Previously
Defined Logical Volume returns "1800-106 An Error Occured".
This happens when a resource group does not use the last
node in the cluseter.

PROBLEM SUMMARY:
C-SPOC add filesystem returns 1800-106 An Error Occured.

PROBLEM CONCLUSION:
Use first node returned by "cllslv -s -n -v" for the LV to
find the LV type.

------

APAR: IY29685 COMPID: 5765C3403 REL: 430
ABSTRACT: THREAD COMPONENT DUMP FAILURE

PROBLEM DESCRIPTION:
Because of a possibly timing situation involving having
a zombie thread in a still existing process at the time of
a memory dump/system crash, it is possible to have the thread
component fail to dump. This will be apparent by having a
dump where the component dump table (cdt) "thread" is not
present in the dump as viewed with dmpfmt of the startup
of KDB.

PROBLEM CONCLUSION:
The problem occurs because the dump routines try to access
memory in a segment that no longer exists. Attempting to read
this memory forces a page fault, which fails the entire
dump routine. The fix checks the existance of this segment
before trying to read the data.

------

APAR: IY29688 COMPID: 5765E6800 REL: 300
ABSTRACT: 3DMON CORE DUMPS

PROBLEM DESCRIPTION:
3dmon core dumps while trying to monitor Multiple host
activity

PROBLEM CONCLUSION:
Fixed the problem in 3dmon code.

------

APAR: IY29703 COMPID: 5765E5400 REL: 440
ABSTRACT: BOTH TYPES OF DISK FENCING ,SSA AND 9333, ARE AVAILABLE FOR ONE

PROBLEM DESCRIPTION:
Concurrent resource group containing both SSA and 9333
concurrent volume groups does not come on line.

PROBLEM CONCLUSION:
Change the logic in the fencing code to check the disk type,
and apply the appropriate kind of fencing.

------

APAR: IY29704 COMPID: 5765E5400 REL: 440
ABSTRACT: CLUSTER WON'T STABILIZE AFTER DISK FENCING OCCURS

PROBLEM DESCRIPTION:
Partitioned cluster node is fenced out from concurrent
resource group in 4node cluster with SSA disk fencing
activated while a node on three-node side of partition
is also fenced out from other concurrent rg's. This
leads to each side of the partition having node_down
EVENT FAILED, leading to event_error EVENT FAILED and
subsequent config_too_long. This occurs on a cluster
with AIX 5.1, HAES 4.4.1, and RSCT 2.2.0.

PROBLEM CONCLUSION:
The logic in the fencing operation is changed to not
report an event_error should fencing fail - though
the failed fencing operation is reported. This will
allow the half of the partition which retains access
to a quorum of disks to continue. Note that it is
still possible that neither half of the partition
can retain a quorum - the outcome is timing dependent.
This is a fundamental limitation of the product.

------

APAR: IY29772 COMPID: 5765C3403 REL: 430
ABSTRACT: LQUERYVG RETURNS ERROR DURING IMPORTVG

PROBLEM DESCRIPTION:
importvg succeeds but customer sees the following error msg:
0516-010: lqueryvg: Volume group must be varied on; use
        varyonvg command

PROBLEM SUMMARY:
lqueryvg call inside redefinevg returns
error if currently importing the VG.

PROBLEM CONCLUSION:
lqueryvg will now get piece of data from
disk rather than volume group data area.

------

APAR: IY29790 COMPID: 5765C3403 REL: 430
ABSTRACT: PROBLEM PRINTING KANJI CHARACTER IN JA_JP

PROBLEM DESCRIPTION:
There is a problem printing the Kanji character
represented by EE63-EE03. It gets erroneously
printed as ED63-EDE0.

PROBLEM SUMMARY:
In Ja_JP, there is a problem printing the Kanji character
represented by EE63-EE03. It gets erroneously
printed as ED63-EDE0.

PROBLEM CONCLUSION:
The Kanji character will now print correctly.

------

APAR: IY29840 COMPID: 5765C3403 REL: 433
ABSTRACT: TRACEROUTE HANGS SOMETIMES

PROBLEM DESCRIPTION:
traceroute command can hang sometime when it doesn't get
response for a particular hop. It should timeout after
5 minutes (which is default, altered by '-w' option).
But when there are other ICMP replies received, traceroute
doesn't keep track of the time and never times out.

LOCAL FIX:
Avoid any other ICMP (replies) traffic when traceroute
is executed.

PROBLEM SUMMARY:
traceroute command can hang sometime when it doesn't get
response for a particular hop. It should timeout after
5 minutes (which is default, altered by '-w' option).
But when there are other ICMP replies received due to
other processes in the system, traceroute code will
process those ICMP packets and will not keep track
of the timeout value. Due to that, traceroute remains
in a state of hang (as long as it receives other ICMP
packets and when it doesn't get a reply for the probe
which it sent, from any router)

PROBLEM CONCLUSION:
Modify the code to have a different timeout variable
which keeps track of the time. This will take care
of the timeout in case router doesn't respond to any
particular probe.

TEMPORARY FIX:
Avoid any ICMP reply traffic on the system when
traceroute is executed

------

APAR: IY29841 COMPID: 5765C3403 REL: 430
ABSTRACT: INIT RESETS PERMISSIONS OF TTYS

PROBLEM DESCRIPTION:
The PID of a getty that was associated with a tty
gets reused. If this PID gets assigned to init for
cleanup, it causes init to reset the permissions of
the old tty, even though it is not associated with any
getty, currently.

PROBLEM SUMMARY:
init resets the tty permissions, making it world readable.

PROBLEM CONCLUSION:
When init cleans any orphans, it
1) Checks if there is an entry in the utmp for that PID.
2) If its there, then it checks if it is associated with
a tty and resets the permissions of that tty. Now an
additional check is made whether that tty entry has
been marked as DEAD. This ensures that when PIDs are reused,
init does not reset the permissions of TTYs associated
with that PID, earlier.

------

APAR: IY29880 COMPID: 5765E5400 REL: 440
ABSTRACT: HACMP: QUORUM SETTINGS NOT IN EFFECT FOR FIRST VARY-ON AFTER

PROBLEM DESCRIPTION:
After HACMP performs a lazy update on a volume group the
quorum is not set correctly until the volume group is varied
off and back on again.

LOCAL FIX:
Vary the volume group off and on either manually (if
appropriate) or through HACMP.

PROBLEM CONCLUSION:
Vary the volume group off and back on if quorum has been
disabled so the setting takes affect.

------

APAR: IY29887 COMPID: 5765C3403 REL: 430
ABSTRACT: PROBLEM WITH RESTORE COMMAND ON DATABASE FILES

PROBLEM DESCRIPTION:
The restore command does not allow to preserve the
non sparseness of database files when restored from
from an archive in back byname format.

PROBLEM SUMMARY:
The restore command fails to maintain the non sparseness
of database files thereby resulting in unallocation
of blocks of memory

PROBLEM CONCLUSION:
The restore code introduces a new flag ('-e')
which restores the data files maintaining their non
sparseness. The new option goes through a non sparsing
algorithm by validating blocks of nulls greater than
4k bytes.

------

APAR: IY29915 COMPID: 5765C3403 REL: 430
ABSTRACT: SPMI CLIENTS SHOWS PEAKS IN GLOBAL CPU METRICS

PROBLEM DESCRIPTION:
Spmi clients, like xmperf, ptrlog, shows peaks & wrong values
for CPU % metrics.

PROBLEM CONCLUSION:
Fixed Spmi to handle the % calculations properly and buffer
the old values properly. Added code to handle wraps before
calculating percentages.

------

APAR: IY29935 COMPID: 5765E5400 REL: 440
ABSTRACT: CSPOC: ADDING A MIRROR TO A LOGICAL VOLUME FAILS

PROBLEM DESCRIPTION:
Adding a copy to a logical volume fails in CSPOC.
mklvcopy of AIX asks for customer input if certain
conditions match like the version of the volume group
being less than 6. So the smit screen appears to be
hung.

LOCAL FIX:
Send the -f flag to mklvcopy command which will forcibly
add a copy without checking the version of the volume
group.

PROBLEM CONCLUSION:
Send the force flag to mklvcopy command which will force the
operation without asking for any confirmations.

------

APAR: IY29964 COMPID: 5765E5400 REL: 440
ABSTRACT: HACMP: CLLSNODE SEGFAULTS WHEN A NODE CONTAINS MANY ADAPTERS

PROBLEM DESCRIPTION:
Cllsnode segfaults when a node contains a large number of
adapters.

PROBLEM CONCLUSION:
Dynamically allocate enough memory to store a string that is
built up in iterations. Previously an arbitrary amount of
fixed space was used.

------

APAR: IY30001 COMPID: 5765C3403 REL: 430
ABSTRACT: VMOUNT: A SYSTEM CALL RECEIVED A PARAMETER ... NOT VALID

PROBLEM DESCRIPTION:
When installing with lpp_sources and spots at different vrmf
levels, the customer may not be able to NFS mount upon reboot.
The error below may appear:
#mount synergy:/usr/sys/inst.images /mnt
mount: 1831-008 giving up on:
synergy:/usr/sys/inst.images
vmount: A system call received a parameter that is not valid.

PROBLEM CONCLUSION:
Modify NIM method c_bosinst_env to only copy
/sbin/helpers/nfsmnthelp once during any type of install.

------

APAR: IY30002 COMPID: 5765C3403 REL: 433
ABSTRACT: RESTVG MAY HANG

PROBLEM DESCRIPTION:
Restore of a user volume group may fail because a data file
cannot be found. It may hang as it tries to find it.

PROBLEM SUMMARY:
Executing restvg may hang. This has been seen only
on fast systems with LTO 3580/3581 ultrium tape
drives. However, this could possibly fail using
other drives. It is a timing issue where the tape
drive is accessed once, and then accessed again,
but it is not ready at to be accessed the next time.

PROBLEM CONCLUSION:
Read only one block on the first read and put in a
sleep to give drive time to get ready.

------

APAR: IY30036 COMPID: 5765E5400 REL: 440
ABSTRACT: HACMP&ES: CLVER GIVES ERRORS ABOUT CUAT FOR RGS WITH NFS

PROBLEM DESCRIPTION:
If there exists a Resource Group in the cluster which has some
filesystems exported, but does not belong to all the nodes in
the cluster, Then when clver is run on the node that does not
have this resource group will give error messages about the
filesystems exported in that resource group.

LOCAL FIX:
clver should only check for filesystem mounts
if the resource group involves that particular
node.

PROBLEM CONCLUSION:
clver should only check for volume group major numbers if the
resouce group having the volume group involved the local node.

------

APAR: IY30088 COMPID: 5765C3403 REL: 430
ABSTRACT: CLOSEDIR() RETURNS UNEXPECTED EBADF ERROR FOR AN APPLICATION

PROBLEM DESCRIPTION:
The closedir() library routine returns an unexpected
EBADF error for an application program.

PROBLEM SUMMARY:
The closedir() library routine reports EBADF error
when the file in a directory is owned by an invalid
user.

PROBLEM CONCLUSION:
The EBADF error is reported by closedir() library
routine. This happens when the application program
is compiled with cc. The same is not reported with
xlc_r.

------

APAR: IY30120 COMPID: 5765E5400 REL: 440
ABSTRACT: CLSTAT DOES NOT CHANGE SUBSTATE FROM 'RECONFIG' AFTER CLDARE

PROBLEM DESCRIPTION:
With a running cluster, after doing a DARE of cluster
topology or resource changes, sometimes clstat will stay
in the RECONFIG state after the DARE finishes.
This has been constantly reproduced when changing the name
of a resource group and DAREing in the change. Sometimes
both nodes in a 2 node cluster have clstat show RECONFIG
substate after the DARE completes, sometimes it happens only
on the node on which the DARE did not originate from,
sometimes it doesn't happen on either nodes.

PROBLEM CONCLUSION:
Have clinfo set it's internal cluster map's substate to STABLE
after receiving a TE_NEW_PRIMARY event.

------

APAR: IY30121 COMPID: 5765E5400 REL: 440
ABSTRACT: CL_SETUP_KERBEROS SHOULD BE EXECUTABLE FROM ANY DIRECTORY

PROBLEM DESCRIPTION:
If you run cl_setup_kerberos you get following errors.
rcp: cl_ext_krb: A file or directory in the path name does not
exist.
chmod: /tmp/kerberos/a2_ensvc/cl_ext_krb: A file or directory
in the path name does not exist.
ksh: ./cl_ext_krb: not found.

PROBLEM CONCLUSION:
Add the path for the cl_ext_krb file.

------

APAR: IY30122 COMPID: 5765E5400 REL: 440
ABSTRACT: CLADDRES CORE DUMPS

PROBLEM DESCRIPTION:
If one of the resource fields on the claddres command has a
list of resources with more members in the list than the
number of different types of resources being added, the stack
will be corrupted and a core can result.

PROBLEM CONCLUSION:
The malloc of HACMPresource structures only allocates space
for "argc" structures. If one of the argments is a list
of resources in quotes, that exceeds argc, the stack will
be corrupted. The code will be changed to allocate enough
space for all of the resources specified.

------

APAR: IY30123 COMPID: 5765C3403 REL: 430
ABSTRACT: ADD IFREQS TO DEVICES.PCI.00100300.DIAG

PROBLEM DESCRIPTION:
Running Diagnostics on a SCSI adapter or attached SCSI devices
may display the following error followed by a Service Request
Number 803-xxx.
Could not load program /usr/lpp/diagnostics/da/dpciscsi:
Symbol resolution failed for /lib/libpdiag.a(shr.o) because:
Symbol pdiag_dd_open (number 12) is not exported from dependent
    module /unix.
Symbol pdiag_dd_close (number 13) is not exported from
dependent
  module /unix.
Examine .loader section symbols with the 'dump -Tv' command.

PROBLEM CONCLUSION:
An update to the Diagnostics Test Unit in filesets
devices.pci.00100c00.diag devices.pci.00100f00.diag and
devices.pci.00100b00.diag.

------

APAR: IY30124 COMPID: 5765E5400 REL: 440
ABSTRACT: HAS: CL_ACTIVATE_FS FAILS IF FS CHECK IS LOGREDO

PROBLEM DESCRIPTION:
cl_activate_fs fails if a RG has filesystem consistency check
parameter set to
logredo and the file system is to be NFS exported.
During node up, the owning node attempts to varyon and
activate the FS in
preparation for NFS exporting. This operation fails with the
following error:
/usr/sbin/cluster/events/utils/cl_activate_fs 261 : FS_LIST:
parameter not set
The node goes into event_error.

PROBLEM CONCLUSION:
fix the syntax error in the script.

------

APAR: IY30125 COMPID: 5765E5400 REL: 440
ABSTRACT: SMIT: CHANGE NETWORK ALIAS SELECTION INVALID

PROBLEM DESCRIPTION:
Change an IP-based Network SMIT screen shows network name in
the field reserved for whether or not aliasing is configured:
Change an IP-based Network
Type or select values in entry fields.
Press Enter AFTER making all desired changes.
Network Name Ether_alias
New Network Name
Subnet(s) 192.168.124.0/24
Add Subnets
Remove Subnets
Use IP aliasing for IPAT Ether_alias
This occured after removing one non-IP network.

PROBLEM CONCLUSION:
Replace the broken logic in cl_harvestIP_scripts to correctly
parse for the aliasing flag from the HACMPnetwork ODM.

------

APAR: IY30126 COMPID: 5765E5400 REL: 440
ABSTRACT: CLADDNIM: INVALID BAUD RATE (LOG) FOR RS232 NIM, IF LOG PARAM.

PROBLEM DESCRIPTION:
claddnim: Invalid Baud Rate (log) for rs232 nim, if log
parameter.

PROBLEM CONCLUSION:
Allow log parameter.

------

APAR: IY30128 COMPID: 5765E5400 REL: 440
ABSTRACT: CORRUPT OR MISSING HACMP.LOGS ODM MESSAGE

PROBLEM DESCRIPTION:
HACMPlogs ODM is missing or corrupted. Message in hacmp.out
after every event and no history/cluster.mmddyyyy log.

PROBLEM SUMMARY:
If APAR IY28996 is applied the Cluster daily history files
will not be created. The default directory for the Cluster
daily history files is /usr/sbin/cluster/history. This
problem will also cause the following error message to appear
in hacmp.out after every EVENT.
HACMPlogs ODM is missing or corrupted.

PROBLEM CONCLUSION:
Change the odmget in the clevlog script from cluster.mmdd to
cluster.mmddyyyy.

------

APAR: IY30140 COMPID: 5765C3403 REL: 430
ABSTRACT: OD -C HANGS WHEN MULTI BYTE LOCALES ARE USED

PROBLEM DESCRIPTION:
od command with -c option hangs when LANG
is exported to multi byte locales.

PROBLEM SUMMARY:
od -c command hangs when multi byte
locale is used.

PROBLEM CONCLUSION:
Code changes will be made to od command to
properly handle the multi byte characters with
-c and -C options.

------

APAR: IY30151 COMPID: 5765C3403 REL: 430
ABSTRACT: WAITPID DUMPS CORE ON 64BIT, WHEN CALLED FROM EXCEPTION HANDLER

PROBLEM DESCRIPTION:
Segmentation fault in waitpid at 0x90000000002297c
0x90000000002297c (waitpid+0xa4) 83230000
        lwz r25,0x0(r3)
(dbx) t
waitpid(??, ??, ??) at 0x90000000002297c
sigchld_handler(0x0) at 0x10000071c
_vp_sleep(??, ??) at 0x9000000003d6110
_usched_dispatch(??, ??, ??) at 0x9000000003d4358

PROBLEM CONCLUSION:
Checked for null value and handled.

------

APAR: IY30198 COMPID: 5765C3403 REL: 430
ABSTRACT: SYSTEM CRASH IN HD_KEXTEND WITH > 32000 PPS

PROBLEM DESCRIPTION:
System crashes due to DSI in hd_kextred or other function
which manipulates pps when customer has > 16000 pps and 2 lv
copies.

PROBLEM SUMMARY:
system crash in hd_kextend on system with a logical
volume having > 16000 logical partitions and mirroring.

PROBLEM CONCLUSION:
variable of type short was expected to hold pp count
greater than 32000.

TEMPORARY FIX:
reduce size of logical volumes in PP number.

------

APAR: IY30270 COMPID: 5765C3403 REL: 430
ABSTRACT: COMPILE TIME ERRORS WITH _XOPEN_SOURCE MACRO

PROBLEM DESCRIPTION:
When compiling code that includes m_types.h and math.h
with the macro, _XOPEN_SOURCE=500 (for strict UNIX98
compilation), user will get following errors.
1) Conflicting declaration of the symbol "label"t"
2) Wrong number of arguments for certain routines in
math.h

PROBLEM SUMMARY:
User will get compilatio errors as:
"/usr/include/sys/m_types.h", line 70.3: 1540-0400 (S)
"label_t" has a conflicting declaration.
"/usr/include/sys/m_types.h", line 62.16: 1540-0424 (I)
"label_t" is declared on line 62 of
"/usr/include/sys/m_types.h". "/usr/include/math.h", line
341.30: 1540-0215 (S) The wrong number of arguments have been
specified for "_power_ilogb()". "/usr/include/math.h", line
345.29: 1540-0215 (S) The wrong number of arguments have been
specified for "_power_logb()". "/usr/include/math.h", line
349.34: 1540-0215 (S) The wrong number of arguments have been
specified for "_power_nextafter()".

PROBLEM CONCLUSION:
The headers math.h and m_types.h have been changed to fix the
problem.

------

APAR: IY30291 COMPID: 5765C3403 REL: 433
ABSTRACT: 2D EDGE VERTEX ARRAYS CAUSES A SEGMENTATION FAULT.

PROBLEM DESCRIPTION:
Use of 2D Edge vertex arrays causes a segmentation fault.

PROBLEM CONCLUSION:
listcomp.c must be changed so that the slowindex is
equal to _VA_EDGEFLAG_1B (0x1000) and fast index is
equal to _VA_SLOW_EDGEFLAG (0x2000). If this is not
changed, the slow path is entered and _VA_SLOW_EDGEFLAG
is incorrectly interpretted as a fog mask.

------

APAR: IY30292 COMPID: 5765C3403 REL: 433
ABSTRACT: WEBSM NIM MKSYSB RESOURCE CREATION FAILS TO MAKE LARGE FILE

PROBLEM DESCRIPTION:
When using WebSM to create a mksysb NIM resource
the customer is given the option to create the filesystem
in which it will be stored. If the customer chooses this
option and the mksysb image is larger than 2 gigabytes
the mksysb creation fails because the new
filesystem is not large file enabled.

LOCAL FIX:
Create the filesystem Large File Enabled manually
and then use it in the WebSM definition of the
mksysb resource. Do not allow WebSM to create
the default filesystem suggested.

PROBLEM SUMMARY:
A failure will occur when WebSM is used to create a mksysb
NIM resource larger than two gigabytes and the option to
create the filesystem that the mksysb image will be stored
in is seleted. The reason for this is that the filesystem
created by WebSM is not large file enabled.

PROBLEM CONCLUSION:
WebSM is using one of NIM's methods (m_fs) to create the
filesystem on the server, but it had no option to create
filesystems large file enabled.
Functionality was added to both m_fs and
to websm to indicate that a large file enabled
filesystem is desired.

TEMPORARY FIX:
Create the filesystem manually.

------

APAR: IY30320 COMPID: 5765C3403 REL: 433
ABSTRACT: KRSHD CORE DUMP

PROBLEM DESCRIPTION:
krshd core dump

PROBLEM SUMMARY:
rcp will fail, with krshd core dumping.

PROBLEM CONCLUSION:
read_service_key() calls krb5_init_context() to create a new
kerberos context and initialize it with the default values
obtained from the configuration file. read_service_key() was
not checking for the return value for krb5_init_context().
When krb5_init_context() fails, "context" variable will be
pointing to zero. krshd dumps core, when we try to free
"context" which is pointing to NULL. Added a check for the
return value of krb5_init_context() in read_service_key().

------

APAR: IY30346 COMPID: 5765C3403 REL: 430
ABSTRACT: MEMORY LEAK IN RTREQUEST_NOLOCK()

PROBLEM DESCRIPTION:
Customer will experience memory leak.

PROBLEM CONCLUSION:
Free the memory before calling senderr().

------

APAR: IY30347 COMPID: 5765C3403 REL: 430
ABSTRACT: MOUNTD SHOULD HAVE OPTION TO NOT RESOLV HOST ADDRESS

PROBLEM DESCRIPTION:
Unable to mount when hostname can not be resolved
unable to turn off updating of /etc/rmtab file

PROBLEM CONCLUSION:
add support for new -x option to allow for no hostname
resolution for mounting. Also add -r option to turn off
the updating of the /etc/rmtab file

------

APAR: IY30349 COMPID: 5765C3403 REL: 430
ABSTRACT: SYSTEM CAN PANIC UNDER HEAVY CONFIG TESTING

PROBLEM DESCRIPTION:
The system can panic under heavy config testing. The stack
trace will show cxma_poll on the last configured cpu and a
process in the middle of the unconfig which may be on another
cpu or the same cpu.

PROBLEM CONCLUSION:
The solution is to remove the panic added during early
developement hoping to catch any coding errors. The panic is
being hit during a timing hole which does not pose a real
risk. The poll would complete successfully so the assert is
too restrictive. There is no assert that would possible that
would actually catch a problem.

------

APAR: IY30357 COMPID: 5765C3403 REL: 430
ABSTRACT: TRACEROUTE CRASHES WHEN GIVEN EXTREMELY LONG ARGUMENTS

PROBLEM DESCRIPTION:
when a large string is given as a parameter to traceroute
command it coredumps.

PROBLEM CONCLUSION:
Buffer boundary check was added.

------

APAR: IY30395 COMPID: 5765C3403 REL: 433
ABSTRACT: IPV6 DEFAULT ROUTE WITH A GATEWAY DOES NOT WORK

PROBLEM DESCRIPTION:
IPv6 traffic will not be sent correctly if a default route
with a gateway is configured

LOCAL FIX:
Add a more specific route to your routing table in addition
to the default route

PROBLEM CONCLUSION:
Code has been dropped to fix this problem

TEMPORARY FIX:
Add more specific routes to the routing table
instead of relying on the default route

------

APAR: IY30401 COMPID: 5765C3403 REL: 430
ABSTRACT: MEMORY LEAK AND LOGIN PROBLEM IS LDAP LOADMODULE

PROBLEM DESCRIPTION:
There are memory leaks in the LDAP loadmodule.
This is noticed when the ldap authentication
was used to connect to the informix engine.
The msc process which does the authentication
was growing in memory and eventually doesnot
take any more connections. This happens only
with LDAP connections and not local user connections.
There is problem with ldap user login that is
fixed with this apar. When a ldap user enters
the username and types a wrong password and in
the same login/telnet session enters the correct
username and password, the user will not be
accepted with error message 'Invalid username
or password entered'. In a new telnet session
there is no problem with correct username and
password.

LOCAL FIX:
For Memory leak, the ldap authentication
should not be used and only local authnetication
should be used.

PROBLEM CONCLUSION:
The buffer allocation is freed when the required
attribute is not found before returning.
For the login problem, the correct attribute is used.

------

APAR: IY30414 COMPID: 5765C3403 REL: 430
ABSTRACT: UNABLE TO RESTORE BOOTABLE DVD IMAGE TO FC DISK

PROBLEM DESCRIPTION:
Customers cannot select a FC scsi disk, as the target for
the install, if installing from a generic system backup
on CD or DVD.
The field:
1 Disk(s) where you want to install
will not list the FC disks.

PROBLEM CONCLUSION:
Added drivers to the cdfs list that mkcd uses to match the
list of drivers on the product CDs.

------

APAR: IY30416 COMPID: 5765C3403 REL: 430
ABSTRACT: GETHGETHOSTNAME CAN'T RE-READ RESOLV.CONF IN MULTI-THREAD

PROBLEM DESCRIPTION:
gethostbyname() in multi-threaded program can't re-read
/etc/resolv.conf file even if RES_INIT flag is unset.
_res.options &= ~RES_INIT;

PROBLEM CONCLUSION:
Make sure the /etc/resolv.conf file is re-read even if
RES_INIT is unset.

------

APAR: IY30419 COMPID: 5765C3403 REL: 430
ABSTRACT: 'ARP -A' DOES NOT DISPLAY ALL ENTRIES

PROBLEM DESCRIPTION:
When 'arp -a' command is issued, the output will not show
all the entries from the ARP table. Output will show only the
first entry from each bucket of ARP table. Though we can see
the entries by issuing 'arp <hostname>' which will display if
the particular entry is in the ARP table. This will be seen
after applying fileset bos.net.tcp.client 4.3.3.79.

PROBLEM CONCLUSION:
arptab structure was modified in 4.3.3.79. But 'arp' command
code was not rebuilt to pick up the modified structure. Since
'arp' binary could not identify the additional entries in the
structure, when the table is getting dumped, it could print
only the first entry from each bucket. Solution is to rebuild
the 'arp' binary and ship the new one.

------

APAR: IY30449 COMPID: 5765C3403 REL: 430
ABSTRACT: SYSTEM CRASH IN MULTICAST APPLICATION

PROBLEM DESCRIPTION:
multicast applications might crash the system

PROBLEM CONCLUSION:
incrementing imo->imo_num_memberships after in_addmulti and
saving the return value of in_addmulti in
imo->imo_membership imo_num_memberships

------

APAR: IY30453 COMPID: 5765C3403 REL: 430
ABSTRACT: LOG LINK ERROR/RECOVERY WHEN POLL LINK ENABLE

PROBLEM DESCRIPTION:
10/100 Ethernet driver does not log errors every time the
cable is unplugged.

PROBLEM CONCLUSION:
Fix the driver to log link error/recovery properly when the
cable is plugged or unplugged.

------

APAR: IY30589 COMPID: 5765C3403 REL: 430
ABSTRACT: CP COMMAND AND BAD TARGET DIRECTORY

PROBLEM DESCRIPTION:
copy of more than one file, using meta characters in filename,
to a non-existing directory path, will not give proper
error message.

PROBLEM CONCLUSION:
code fixed to see appropriate error message is displayed

------

APAR: IY30590 COMPID: 5765C3403 REL: 430
ABSTRACT: STRIPNM FAILS ON LARGE LIBRARIES

PROBLEM DESCRIPTION:
stripnm core dumps or gives out of memory error if it is
asked to display symbols of large library ( size of library
may be greater than 2MB )

PROBLEM CONCLUSION:
Fixed stripnm by increasing the hash table size such that
it can accomodate more symbols. This will prevent out
of memory errors and hence core dumps.

------

APAR: IY30592 COMPID: 5765C3403 REL: 430
ABSTRACT: INTRPPC_ERR OCCURS WHEN ADAPTER RESTARTS DUE TO SOME IOCTLS.

PROBLEM DESCRIPTION:
INTRPPC_ERR occurs when the adapter needs to restart due to
some IOCTL commands.

PROBLEM CONCLUSION:
The fact that the adapter needs to restart on certain IOCTL
commands is a h/w limitation. The device driver will be
modified to handle pending interrupts properly in this case.

------

APAR: IY30621 COMPID: 5765C3403 REL: 430
ABSTRACT: TO MODIFY MESSAGE FILE ENTRY FOR CHINA TIMEZONE

PROBLEM DESCRIPTION:
setting timezone to BEIST-8BEIDT fails with error

PROBLEM CONCLUSION:
modify the necessary odm and message files

------

APAR: IY30625 COMPID: 5765C3403 REL: 430
ABSTRACT: STICKY LIBRARY IMPLEMENTATION FOR AIX 4.3

PROBLEM DESCRIPTION:
Some libraries once loaded must not be unloaded. If such
library is unloaded will cause exception in the program.
One of such libraries is libpthread. When it is loaded as a
dependency of the executable it will not be unloaded,
however, when it is loaded by dlopen() function it will be
unloaded by dlclose() function.

PROBLEM CONCLUSION:
The library which cannot be unloaded will notify the system
loader about it with a system call. This notification should
be done during the library initialization phase.

------

APAR: IY30626 COMPID: 5765C3403 REL: 430
ABSTRACT: SYSTEM CRASH WHILE DELETING ARP ENTRY

PROBLEM DESCRIPTION:
The arp -s command followed by a arp -d command will
fail for permanent arp entry causing the system to crash
This is because a arp timer table was getting
corrupted

PROBLEM CONCLUSION:
Fixed the problem to run arp -s and arp -d command
to execute correctly.

------

APAR: IY30677 COMPID: 5765C3403 REL: 430
ABSTRACT: GUARDPAGES FOR USER DEFINED STACK

PROBLEM DESCRIPTION:
This is not a defect. It is a service aid to debug user
defined pthread stack corruption (User can pass their own
stack using the pthread_attr_setstackaddr subroutine). Till
now the facility was available to only pthread stack created
internally by the system.
It has some limitations:
1. User has to provide stack size also.
2. The guard pages and alignment pages has to be curved out
of the user's stack (which reduced the stack size). If the
stack size is in the border line then it may not be even
possible to construct a guard page.
But in majority of cases it would still be useful.

PROBLEM CONCLUSION:
By default, it would be turned off and can be invoked by
exporting the enviromental variable:
"AIXTHREAD_GUARDPAGES_FOR_USER_STACK".
Further, even if it is exported guard pages will be
constructed only if both the attributes stackaddr and
stacksize has be set by user for the thread. Also if the
combination of stacksize and guardpages turns out to be such
that that guard page cannot be constructed it will be ignored.

------

APAR: IY30939 COMPID: 5765C3403 REL: 430
ABSTRACT: MAY NOT BE ABLE TO INSTALL A FILESET ON CLIENT THROUGH NIM

PROBLEM DESCRIPTION:
If customer puts on the bos.sysmgt.smit from IY18432 (43) or
IY21183 (51), then also need bos.sysmgt.nim.master from
IY28646 (43) or IY29816 (51).

PROBLEM CONCLUSION:
IY28646 (43) and IY29816 (51) fixed this problem. This apar
will add the requisite so later changes to bos.sysmgt.smit
will pull in IY28646 or IY29816.

------

APAR: IY30969 COMPID: 5765B8100 REL: 230
ABSTRACT: ISDN_CALL_TRANSFER CS FAILS TO TRANSFER CALLS. ERRORID 20501

PROBLEM DESCRIPTION:
The ISDN_Call_Transfer custom server occasionally gets into a
bad state and will no longer transfer calls. Each attempt to
transfer a call will result in an errorid 20501 being generated,
with the following error text:
ISDN_Call_Transfer: MakeCallStatus is not valid at this time.

PROBLEM SUMMARY:
ISDN_CALL_TRANSFER CS FAILS TO TRANSFER CALLS.
ERRORID 20501

PROBLEM CONCLUSION:
Layer 4 was fixed so that the TERMINATE_CNF
is always sent freeing up the CHP properly when the transfer is
complete; or when it fails due to one of the parties involved
in the transfer hanging up.

------

APAR: IY30974 COMPID: 5765B8100 REL: 230
ABSTRACT: ISDN_CALL_TRANSFER CS FAILS TO TRANSFER CALLS. 20501 TRACE

PROBLEM DESCRIPTION:
The ISDN_Call_Transfer custom server occasionally gets into a
bad state and will no longer transfer calls. Each attempt to
transfer a call will result in an errorid 20501 being generated,
with the following error text:
ISDN_Call_Transfer: MakeCallStatus is not valid at this time.

PROBLEM SUMMARY:
ISDN_CALL_TRANSFER CS FAILS TO TRANSFER CALLS.
ERRORID 20501

PROBLEM CONCLUSION:
Layer 4 was fixed so that the TERMINATE_CNF
is always sent freeing up the CHP properly when the transfer is
complete; or when it fails due to one of the parties involved
in the transfer hanging up.

------

APAR: IY31018 COMPID: 5765B8100 REL: 230
ABSTRACT: DIRECTTALK BEANS GRAMMER UPDATES

PROBLEM DESCRIPTION:
Updates to DirectTalk Beans grammers

PROBLEM SUMMARY:
directtalk Beans grammer updates

PROBLEM CONCLUSION:
Grammer updates applied

------

APAR: IY31131 COMPID: 5765C3403 REL: 430
ABSTRACT: DIRECT CONNECT OF 6228 ADAPTERS CAUSES MICROCODE PROGRAM ERRORS

PROBLEM DESCRIPTION:
Running cfgmgr to a FASTt500 direct connect (arb loop)
causes Microcode Program Errors on 6228 host adapters.

PROBLEM CONCLUSION:
If the class 2 FLOGI that is used for topology discovery is
rejected with a reason code of class not supported. The
adapter driver will now re-initialize the link in Arbitrated
Loop Mode with a new set of retries. This fix coupled with
a Microcode upgrade of the FASTt will correct the issue with
the Microcode Program Errors.

------

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