|
BCKTTSFX NLM
============
SYMPTOM:
Typically after re-booting the server and going though the process of backing
out the TTS files the server would hang with no messages.
SOLUTION:
Apply BCKTTSFX NLM 1157 10-03-94 9:19a
Solution Specifics:
Different versions of this patch are available for the following NetWare
Operating System's:
V3.11, V3.12, V4.01, V4.02
Example: v3.11 is in 311ITx.EXE or 311PTx.EXE,
v3.12 is in 312ITx.EXE or 312PTx.EXE,
v4.01 is in 401ITx.EXE or 401PTx.EXE,
v4.02 is in 402ITx.EXE or 402PTx.EXE.
v3.11 SFT is in SFTITx.EXE.
This Patch corrects a problem with backing out more than 263 Transaction
Tracking System (TTS) files.
It is important to Load BCKTTSFX.NLM after loading PATCH MANAGER in the
STARTUP.NCF file to allow the patch to work properly.
Installation Instructions:
NCF FILE: STARTUP.NCF
-----------------------------------------------------------------
COMCLSFX NLM
============
SYMPTOM:
Copying compressed files from Server A to Server B using NCOPY /R or NCOPY
/R/U leads to server reboot.
SOLUTION:
Apply COMCLSFX NLM 911 10-03-94 9:24a
Solution Specifics:
Different versions of this patch are available for the following NetWare
Operating System's:
V4.01, V4.02
If a user uses NCOPY /R or NCOPY /R /U to move a compressed file from a
NetWare 4.x server to another NetWare 4.x server, and then at some future date
that file is opened and accessed from either a client or the OS at the target
server; that file server will reboot.
This was caused by compression using a NoSleep WorkToDo that, when executed,
slept on disk IO's. This caused the WorkerThreadAvailList to be NULL and the
machine to ABEND with a page fault if the machine was under heavy load and all
server processes were busy. This condition was caused by ArcServe backing up
a file server with several compressed files.
Installation Instructions:
NCF FILE: STARTUP.NCF
-----------------------------------------------------------------
COMRAFIX NLM
============
SYMPTOM:
Accessing corrupt compressed files causes server ABEND, reset, or congestion
SOLUTION:
Apply COMRAFIX NLM 1066 10-03-94 9:12a
Solution Specifics:
Different versions of this patch are available for the following NetWare
Operating System's:
V4.01, V4.02
Installation Instructions:
NCF FILE: STARTUP.NCF
-----------------------------------------------------------------
COMSEQFX NLM
============
SYMPTOM:
The customer restored a backup. One particular file, when opened would cause
the server to ABEND when it was opened. The restored file was sparse, and had
been backed up from a volume with a smaller allocation size.
SOLUTION:
Apply COMSEQFX NLM 1038 10-03-94 8:48a
Solution Specifics:
Different versions of this patch are available for the following NetWare
Operating System's:
V4.01, V4.02
FIXED decompression problem. If a sparse file was created on a volume with
block size x and then copied compressed to a volume of block size y and x < y
then decompress build file would build an invalid fat chain causing a later
non-sequential IO request during decompression to ABEND.
Installation Instructions:
NCF FILE: STARTUP.NCF
-----------------------------------------------------------------
CORDFSFX NLM
============
SYMPTOM:
After creating and writing to the file as a regular os file, the file is
opened and written to as a directfs file and then open and read as a directfs
file and the data at the end of the file is corrupted
SOLUTION:
Apply CORDFSFX NLM 1033 12-21-94 12:56p
Solution Specifics:
Different versions of this patch are available for the following NetWare
Operating System's:
V4.02
Suballocation was reading blocks using InternalDiskGet. If any blocks were
cached in memory for the DFS file and these blocks were at the end of a file
the cached blocks would be written to the DFS file overlayingand data
previously written.
Oracle was opening a file in DFS mode and then opening the same file in CLIB
mode reading some data and closing it. A write would then be done on the DFS
handle When closed in DFS mode suballocation would read the data out of cache
for the conversion to suballocation and overwrite the data written in DFS
mode. Fixed by putting a check in trysuballocation for a handle in DFS mode
and then to invalid the cache related to this handle.
Installation Intructions:
NCF FILE: STARTUP.NCF
-----------------------------------------------------------------
CSEMFIX NLM
============
SYMPTOM:
Server Abends when mirroring duplexed drives. "Invalid semaphore passed to
kernel."
SOLUTION:
Apply CSEMFIX NLM 2464 12-23-94 12:01p
Solution Specifics:
Different versions of this patch are available for the following NetWare
Operating System's:
V4.02
Installation Instructions:
NCF FILE: STARTUP.NCF
-----------------------------------------------------------------
CTRSTKFX NLM
============
SYMPTOM:
Bug in LSLControlStackFilter functions NetWare v3.1x & v4.0x.
SOLUTION:
Apply CTRSTKFX NLM 923 10-03-94 9:26a
Solution Specifics:
Different versions of this patch are available for the following NetWare
Operating System's:
V3.12, V4.01, V4.02
Problems found with NLA Requests to turn on Promiscuous Mode while HSL.NLM
(Hub Support Layer) is loaded (NLA part of the NMS & RMON stuff) causes the
server to hand (NW v3.x & NW v4.0x).
Problem due to code in LSL function LSLControlStackFilter, at label
<ControlChainDefaults>, have after testing for entries in DefaultChainTable
have <jnz CallPreScanLoop> which should be the line <jnz CallDefaultLoop>.
Installation Instructions:
NCF FILE: STARTUP.NCF
-----------------------------------------------------------------
DHANDFX NLM
============
SYMPTOM:
The file server will exhibit a GPPE or other ABEND having to do with corrupt
directory handles. This is most notably seen when running Appletalk,NFS
gateway, or GroupWise NLM's, but can be seen when loading any set of NLM's
which exceeds the 20 handle limit.
CAUSE:
It was discovered that directory handles were getting overwritten or reused by
some other process before the current process was finished with it. This would
result in an ABEND or other corruption. It was found that connection zero
(commonly used by NLM's at the server) was not properly handling the case
where 20 or more handles were required.
SOLUTION:
Apply DHANDFX NLM 3048 12-23-94 11:48a
Solution Specifics:
Different versions of this patch are available for the following NetWare
Operating System's:
V3.11, V3.12, V4.02
The 3.12 version of this patch incorporates two formerly released patches for
3.12 NetWare which were called, DIRSRCFX.NLM and CRDIRFIX.NLM. DHANDFX.NLM
will NOT load concurrently with these patches. If you are running these two
patches please remove them prior to loading the DHANDFX.NLM
Installation Instructions:
NCF FILE: STARTUP.NCF
-----------------------------------------------------------------
EASUBFIX NLM
============
SYMPTOM:
When copying a long directory with EA's and only have R F rights the server
returns 140 error.
SOLUTION:
Apply EASUBFIX NLM 925 10-03-94 9:05a
Solution Specifics:
Different versions of this patch are available for the following NetWare
Operating System's:
V3.11, V3.12, V4.01, V4.02
Login to a 3.12 server that has a name space loaded. Go to a directory that
has a long directory in it. Make sure you only have R F rights in the source
directory. Then use FILER to copy the subdirectory structure, on the long
directories you will get an error. Problem was traced to NWFindFirstEA()
returns error code 140 ( No Set Privileges).
FIXED a bug OpenEAHandle, were is you only had read rights on a file, you
could not read the EA's on that file.
Installation Instructions:
NCF FILE: STARTUP.NCF
-----------------------------------------------------------------
IOHANDFX NLM
============
SYMPTOM:
On releasing IO handles, the OS does not clean up the application link field.
Do the following steps to reproduce the problem:
1. Load CDROM.NLM on a 3.12 server with a CD-ROM.
2. CD MOUNT # (to mount the CD).
3. CD CHANGE # ( to mount a different CD).
4. Take the CD out at the prompt and press any key to continue.
5. Unload CDROM.NLM.
6. Load CDROM.NLM.
You can repeat steps 5 and 6 after putting a CD back into the device and the
CD can no longer be seen until the disk driver is unloaded and loaded again.
SOLUTION:
Apply IOHANDFX NLM 1021 12-21-94 1:02p
Solution Specifics:
Different versions of this patch are available for the following NetWare
Operating System's:
V3.12, V4.01, V4.02
Installation Instructions:
NCF FILE: STARTUP.NCF
-----------------------------------------------------------------
LSLUNBFX NLM
============
SYMPTOM:
When unloading Madge driver, user receives error message: TCP - unrecognized
parameter 'xxx' (where xxx is always three extended ASCII characters).
ESI is expected to point to command line parameters for the Unbind command.
This unbind results from the unload of the board and there are no command line
parameters. However, ESI is not zero and TCPIP interprets the value as a
pointer to the parameters. LSL is not clearing ESI, or is only clearing it
when the first stack is unbound, when the unbind is caused by the unload of a
board. ESI needs to be cleared before calling the protocol with unbind in this
case.
The problem is in the LSL call <LSLUnBindThenDeRegisterMLID>, the register ESI
contains bogus values before the call to the Protocol Stack Control Entry
point to UNBIND, ESI should point to a string or zero. ESI should be set to
zero before making the call to the Protocol Stack, ie. <xor esi, esi>. This
fix has already been made in the NetWare v4.10 LSL code base. The same fix
needs to be made to the NetWare v3.12 and v4.0x LSL code bases.
SOLUTION:
Apply LSLUNBFX NLM 974 12-21-94 12:28p
Solution Specifics:
Different versions of this patch are available for the following NetWare
Operating System's:
V3.12, V4.02
Installation Instructions:
NCF FILE: STARTUP.NCF
-----------------------------------------------------------------
MEMALLFX NLM
============
SYMPTOM:
"Short term memory allocator out of memory"
Customer sees these messages frequently on his server. Usually happens during
heavy traffic times. Typically happens once a day but may skip a day
occasionally. Problem seems to stop when users log out and/or files are
closed. Performance at the WS's during the times this message is occurring is
extremely slow, may lose connection. The server is a Netframe with 128M of
ram, diagnostic NLM's have shown that there is plenty of contiguous, clean
memory available at the time the request for memory is being made but fails.
Drivers for NIC and HBA are Netframe type. NFS has been removed from the
server in attempt to isolate the problem but no help.
SOLUTION:
Apply MEMALLFX NLM 1330 10-03-94 9:21a
Solution Specifics:
Different versions of this patch are available for the following NetWare
Operating System's:
V4.01, V4.02
Implemented patch to AllocateMappedMemory() to not cause the guard pages to be
allocated and to also jump to the physical cache pool to allocate RAM if the
other areas were depleted.
Installation Instructions:
NCF FILE: STARTUP.NCF
-----------------------------------------------------------------
PBDSTFIX NLM
============
SYMPTOM:
PacketBurst Packet Abending Server
SOLUTION:
Apply PBDSTFIX NLM 1046 12-21-94 12:41p
Solution Specifics:
Different versions of this patch are available for the following NetWare
Operating System's:
V4.02
Installation Instructions:
NCF FILE: STARTUP.NCF
-----------------------------------------------------------------
PM402 NLM
============
SYMPTOM:
SOLUTION:
Apply PM402 NLM 7652 09-16-94 9:57a
Solution Specifics:
Different versions of this patch are available for the following NetWare
Operating System's:
V4.02
This README file accompanies PM401.NLM (patch manager for NetWare v4.01). The
following files are included for this revision:
This version of the patch manager replaces any and all previous versions of
the patch manager for NetWare v4.01. PM401.NLM can only be used with NetWare
v4.01.
PM401.NLM is the Patch Manager which tracks and manages all official patches
for NetWare v4.01.
Patches for NetWare v4.01 consist of three types:
DYNAMIC -- Dynamic patches are implemented as NLM files that can be
loaded/unloaded while the server is running. Unloading a dynamic patch will
restore the Operating System to its original "un-patched" state.
SEMI-STATIC -- Semi-static patches can also be loaded while the server is
running, but they cannot be unloaded. It is not possible to undo the effects
of a semi-static patch without first downing the server and bringing it back
up without loading the semi-static patch.
STATIC -- A static patch is a DOS executable program that modifies the
SERVER.EXE file. This type of patch is applied once and its effects are
permanent. It is suggested that a backup of the original SERVER.EXE file be
made before applying a static patch. (In other words, do not patch the
SERVER.EXE on your original distribution diskette!)
Dynamic and semi-static patches modify the Operating System in memory, not on
the disk. This means that dynamic and semi-static patches must be loaded each
time the Operating System is brought up in order for any 'fixes' to take
effect. It is recommended that the command "LOAD <patch>" be placed in the
file server's AUTOEXEC.NCF file to ensure that the patch is always in effect
as soon as the server is brought up.
In order to see which patches are currently loaded on the system, type
"PATCHES" at the file server command line. The patches will then be grouped
and displayed according to their type (i.e. - STATIC, SEMI-STATIC, or
DYNAMIC). Typing "MODULES" at the server console will show which dynamic
patches have been loaded, but will not show semi-static or static patches.
The Patch Manager (PM401.NLM) must be loaded before any dynamic or semi-static
patches can be loaded. If PM401.NLM is not already loaded, the <patch>.NLM
will attempt to locate PM401.NLM and, if found, will load it automatically.
PM401.NLM only needs to be loaded once; all NLM patches can then be loaded.
PM401.NLM cannot be unloaded until all patches that rely on it are first
unloaded.
*****************************************************************
DISCLAIMER
Novell, Inc. makes no representations or warranties with respect to any
NetWare software, and specifically disclaims any express or implied warranties
of merchantability, title, or fitness for a particular purpose.
Distribution of any NetWare software is forbidden without the express written
consent of Novell, Inc. Further, Novell reserves the right to discontinue
distribution of any NetWare software.
Novell is not responsible for lost profits or revenue, loss of use of the
software, loss of data, costs of re-creating lost data, the cost of any
substitute equipment or program, or claims by any party other than you.
Novell strongly recommends a backup be made before any software is installed.
Technical support for this software may be provided at the discretion of
Novell.
*****************************************************************
-----------------------------------------------------------------
READQFIX NLM
============
SYMPTOM:
When a Queue is full, the File Server will get into an infinite loop trying to
service Queue Management System (QMS) requests. Note that PCONSOLE makes QMS
requests for building a list of jobs, inserting and deleting print jobs, as
well as inserting and deleting print queues.
Symptoms observed when this problem occurs include, but are not limited, to
utilization up to 96 %, users loosing connections and Packet Receive Buffers
count going high.
SOLUTION:
Apply READQFIX NLM 1577 08-10-94 4:23p
By applying ReadQFix.NLM, the file server can correctly process a QMS request
for a queue which is full. Note that if a queue is full prior to loading
READQFIX, the following error message may appear on the console at any time
after loading READQFIX.
"WARNING: Error detected in Queue. ID=xxxxxxxx"
If this happens the queue with the given ID number should be deleted and
recreated.
Solution Specifics:
Different versions of this patch are available for the following NetWare
Operating System's:
V3.11, V3.12, V4.02
Installation Instructions:
NCF FILE: STARTUP.NCF
-----------------------------------------------------------------
REISRFIX NLM
============
SYMPTOM:
After installing the product update for 4.01 Proliant 2000 Pentium ABEND.
"Stack overflow detected by kernal"
SOLUTION:
Apply REISRFIX NLM 1227 10-03-94 9:09a
Solution Specifics:
Different versions of this patch are available for the following NetWare
Operating System's:
V4.01, V4.02
The latest change to their 4.01 servers was product update for 4.01 and after
applying the patches the servers have randomly been going down, when one
server went down other sent in image on all three machines.
In 4.01 OS file IPXPROTO.386 procedure SendPacket() interrupts are enabled at
line 2304. This is causing a stack overflow problem whenever the servers
receives a large number of type 20 (NETBIOS BROADCAST) packets. The patch
ReISRFix.NLM patches the code by pushing the flags on entrance before the CLI
is executed and then popping the flags on exit instead of re-enabling
interrupts with an STI.
Installation Instructions:
NCF FILE: STARTUP.NCF
-----------------------------------------------------------------
RELLOGFX NLM
============
SYMPTOM:
TTSEndTrans() broke with NETX; works with VLMS
SOLUTION:
Apply RELLOGFX NLM 1018 10-03-94 9:03a
Solution Specifics:
Different versions of this patch are available for the following NetWare
Operating System's:
V4.01, V4.02
Installation Instructions:
NCF FILE: STARTUP.NCF
-----------------------------------------------------------------
RSLEEPFX NLM
============
SYMPTOM:
On some conditions, unloading NWIP.NLM can abend a server. The running thread
is always some "server" thread. From the debbugger, we do not see anymore
NWIP threads at all. However, NWIP is still loaded. From the command
console thread, we see that clib is calling UnloadModule and was calling some
CYieldWithDelay type of routine, hence, it is blocked.
The server abends when accessing the adaptorId structure in IsServerAllowed.
This ID is suppose to be the LAN address that the service packet information
is to be sent to. It appears that Up to SendServerInfo but before calling
IsServerAllowed, the LocalNet table for some unknown reason thought NWIP
virtual driver is still alive. Hence, it was trying to inform us about a new
SAP. The supposely net entry address for NWIP actual contains RipWorkToDo
structure instead.
SOLUTION:
Apply RSLEEPFX NLM 1188 12-21-94 1:41p
Solution Specifics:
Different versions of this patch are available for the following NetWare
Operating System's:
V4.02
Installation Instructions:
NCF FILE: AUTOEXEC.NCF
-----------------------------------------------------------------
SALFILFX NLM
============
SYMPTOM:
Directory entries remained locked when error occurs in salvaging
SOLUTION:
Apply SALFILFX NLM 1098 10-03-94 9:27a
Solution Specifics:
Different versions of this patch are available for the following NetWare
Operating System's:
V4.01, V4.02
This causes connections to lock up that are doing salvaging of deleted files.
Installation Instructions:
NCF FILE: STARTUP.NCF
-----------------------------------------------------------------
SERCHKFX NLM
============
SYMPTOM:
A Third party company has a NLM that can take over when a server is down, and
continue to provide services for the downed server. The check for the socket
number causes their licensing to detect a duplicate server. It is not really
necessary to check, when the socket number will be forced.
SOLUTION:
Apply SERCHKFX NLM 1013 10-03-94 8:59a
Solution Specifics:
Different versions of this patch are available for the following NetWare
Operating System's:
V4.01, V4.02
Installation Instructions:
NCF FILE: STARTUP.NCF
-----------------------------------------------------------------
SERNUMFX NLM
============
SYMPTOM:
Serial number appears reverse order, byte swapped from the S/N on diskette or
server console.
SOLUTION:
Apply SERNUMFX NLM 944 10-03-94 9:01a
Solution Specifics:
Different versions of this patch are available for the following NetWare
Operating System's:
V4.01, V4.02
Installation Instructions:
NCF FILE: STARTUP.NCF
-----------------------------------------------------------------
SIGLOGFX NLM
============
SYMPTOM:
NWP_LOGIN when password has expired in bindery corrupts connection
SOLUTION:
Apply SIGLOGFX NLM 1011 10-03-94 9:28a
Solutions Specifics:
Servers 4.02 and 3.12 Signature level 1
Client VLMS dated 5-26-94 Signature level 2
Login 4.084 and 4.093 Login SPD 70145
Create user (TEST), assign a password and expire that password but have grace
logins available.
Login to user DS mode "LOGIN server/TEST" When asked to change password, reply
N Login to user bindery mode "Login server/TEST /B". You will get Abort,
Retry, Fail error. All you can do is reply F. Then Login gets several 8801
errors, invalid connection. Server gets "Server 4.0-2506 AnNCP request with an
invalid security signature was received from user TEST.... Possible intruder
or network is corrupt."
When the API calls VLM with NWP_LOGIN, everything after this fails. From
CodeView, the VLM request is made, then CodeView is exited. The server gets
several beeps and error messages then the dos prompt "invalid drive.." is
displayed. Login works in this situation if NETX.EXE is used
The problem we are seeing is that certain NCP packets do not function properly
on the server side with signing turned on. In this case it is a PING FOR NDS
packet. During a BINDERY Login to a 4.x server after the KEYED LOGIN NCP is
sent and returned, a PING is sent with signing turned on. The server does one
of two things in response: 1) It will either respond to the ping with no
signing, in which case the client will fail the Login because signing is
required; or 2) it will output intruder alert messages at the console and
never respond to the ping. This happens using the 4.x LOGIN.EXE when doing a
BINDERY emulation Login. Using a 3.x Login for BINDERY does not do the ping
and therefore succeeds.
Installation Instructions:
NCF FILE: STARTUP.NCF
-----------------------------------------------------------------
SPXDIAFX NLM
============
SYMPTOM:
RConsole from NetWare 4.01 retrieves RConsole servers through an SPX
diagnostic request (return all known servers). In the routine
SpxSeqPackRXESR there is a check for SpxSendEcb->status to determine if status
is ok. If the status is not ok the routine will return and not process the
diagnostic request. SPX was designed to work in this manner, since
processing the diagnostic request under certain conditions could cause a
server ABEND.
Once a server determines that the SpxSendEcb ->status is not OK, the request
is not processed. However, SpxSendEcb should be reset back to allow
processing of the next diagnostic SPX packet.
SOLUTION:
Apply SPXDIAFX NLM 989 10-03-94 8:53a
Solution Specifics:
Different versions of this patch are available for the following NetWare
Operating System's:
V4.01, V4.02
Installation Instructions:
NCF FILE: STARTUP.NCF
-----------------------------------------------------------------
SUBLCKFX NLM
============
SYMPTOM:
Deadlock occurs in 4.10
Server moves to 100% utilization and nlms block resources while loading.
SOLUTION:
Apply SUBLCKFX NLM 1436 12-21-94 1:18p
Solution Specifics:
Different versions of this patch are available for the following NetWare
Operating System's:
V4.02, V4.10
Installation Instructions:
NCF FILE: STARTUP.NCF
-----------------------------------------------------------------
SUBMOVFX NLM
============
SYMPTOM:
Server stuck at 40% Utilization, Clients receive wait for over 24 hours.
SOLUTION:
Apply SUBMOVFX NLM 1811 12-21-94 1:15p
Solution Specifics:
Different versions of this patch are available for the following NetWare
Operating System's:
V4.02
The suballocation move procedure would allocate some cache blocks and then
block on the VolumeSemaphore. The thread that had the volume semaphore would
call internal alloc and spin on the cache buffers in use by the suballoc move
procedure causing deadlock.
Installation Instructions:
NCF FILE: STARTUP.NCF
-----------------------------------------------------------------
WRLENFIX NLM
============
SYMPTOM:
WRLENFIX.NLM (Write Length Fix) fixes an illegal NETX packet which requests
to write over the 4 gig boundary (i.e. start offset = -1, bytes to write = 2).
To stop the ABEND we truncate the length of the write just like the VLM's do
so there is no wrap condition.
SOLUTION:
Apply WRLENFIX NLM 1037 12-21-94 12:36p
Solution Specifics:
Different versions of this patch are available for the following NetWare
Operating System's:
V3.12, V4.02
Installation Instructions:
NCF FILE: STARTUP.NCF
-----------------------------------------------------------------
WRPACKFX NLM
============
SYMPTOM:
Abend: Page Fault related to bad NCP 73 packet write to file, packet gives the
data lenght as e15a hex 57690 bytes to copy to buffer.
SOLUTION:
Apply WRPACKFX NLM 1070 12-21-94 12:45p
Solution Specifics:
Different versions of this patch are available for the following NetWare
Operating System's:
V4.02
WrPackFx.NLM stops a corrupt BYTES TO WRITE field in the Case 73 NCP from
generating a page fault when the InternalDiskWrite code is copying the packet
data to the appropriate cache buffers. If the BYTES TO WRITE field indicates
a value which will cause the source register during the copy to move into
memory which is not mapped in a Page Fault results.
Installation Instructions:
NCF FILE: STARTUP.NCF
|
