-
pqau corrupted
images into ciApe memory has been correctedK

Problems Corrected


Ascend Communications Inc
.


P
age
24



Issue

Resolution

MIB II

The following responses to SNMP MIB
-
2 queries for ifSpeed and ifType
have been corrected.



ifSpeed query
response from the UIO card was incorrect. The logical
port bandwidth, not the physical port rate, is the correct response.



ifSpeed query response from the HSSI card showed an incorrect
number.




ifSpeed query response from the ATM
-
DS3 card showed an incorre
ct
number.



ifType query response from the ATM
-
DS3 card has been changed
from
lapb

to DS3.

MIB variables



The Channelized DS3 MIB values now have correct default values.



The MIB values cardDiagNonFatalTime and cardDiagFatalTime no
longer report the wrong va
lues.

Module Identification
Memory (MIM)

MIM is supported via a console command (show hardware).

Console commands



When executing show Lport attributes console command, a logical
port that is added when its physical port was down was reported as
physical
port 0 instead of the actual physical port id. This problem has
been corrected.

The show card console command has been corrected to indicate that the
memory statistic it reports is Memory available and not Memory utilized,
as it had reported previously.

S
ystem



Corrected a problem with the IOPA parity detector.



Modified node number field to correctly report node numbers greater
than 127 for Bulk Statistics file names.



Enhanced error checking to prevent unknown or corrupted images
from downloading into the s
witch.

Statistics

Added internal counters to transmit statistics for frame sizes over 10000
bytes and for TX interrupts with no frames on queue.

36.0 IOP Type A and B card
crash

The problem that caused IOP Type A and IOP Type B cards to crash with
Error
36.0 has been corrected.

SNMP



Corrected a problem where if multiple IP addresses are assigned to a
community name, only the first address would be recognized for SET
actions.



When PRAM synching two cards in one switch from two NMS
workstations simultaneou
sly, the SNMP
pram sync complete

trap
returned for each event indicated the same card number. This problem
has been corrected.



A problem has been fixed where Community Names greater than 31
bytes in length were being accepted by the B
-
STDX.



On a UIO card,
SNMP traps were not being generated when physical
port type did not match with the type defined by the user. This
problem has been corrected.

SNMP

(C96
-
9535)

SNMP get requests for lportDTEFailReason and lportDCEFailReason
were returning improper values. T
his problem has been fixed.

Ethernet

The Ethernet link remains alive if any traffic is received during the time
-
out interval (60 seconds). Previously it would remain alive only if it saw
directed IP or ARP traffic.

OSPF

Relaxed a restriction which preven
ted release 4.01 from functioning in a
network of class ‘B’ IP addresses.

Problems Corrected


Ascend Communications Inc
.


P
age
25



Issue

Resolution

LMI

Corrected a problem where NNI and UNI
-
DCE logical ports would
initialize and report PVC’s active before receiving an active bit from a
remote logical port of

a msC segmentK

qcqm

qhe problem that caused the file transfer to become suspendedI if the
directory in which either the card memory dump file or the moAj upload
file was placed was write protectedI has been fixedK

fllegal base fm Address

qhe switch no l
onger allows the illegal base fm address ORRKORRKORRKORR to
be setK


4.5.2

CP 30/40/50


Issue

Resolution

CP 30/40/50 cold
boot/power up

In a switch with a CP 30/40/50 card, IOP cards no longer fail with 'Bus
Initialization' errors when you coldboot or powered
up the switch. Once
the cards get shot (reloaded), they become operational.

4.5.3

IOP


Issue

Resolution

Redundant IOP Warmboot

When you warmboot or PRAM Synch redundant IOPs, the active IOP no
longer fails or is shot (reloaded) by the standby, resulting in a s
witchover.

4.5.4

Channelized DS3


Issue

Resolution

Admin Down Physical Port
on Channelized DS3 Card

When you set the physical port Admin Status to Down, the card no longer
performs a cold boot.

ISDN, SMDS and ATM
DXI/FUNI support

The Channelized DS3 card now

supports Frame Relay , Frame Relay to
ATM Interworking, ISDN, and ATM DXI/FUNI.

Trunks on channelized DS3

This release supports trunks on a channelized DS3 card.

4.5.5

HSSI


Issue

Resolution

Intermittent failures

The problem of a HSSI card intermittently fai
ling when you continuously
bounce a trunk that is configured on it has been fixed.

HSSI driver oversubscribed

(Solution: S96
-
940)

The HSSI driver has been fixed to prevent a deadlock condition between
the driver and the DMA controller when the HSSI is bei
ng oversubscribed
at low bandwidth.

HSSI driver and circuits

(Solution: S96
-
892)

The HSSI driver has been fixed to prevent circuits from bouncing when
small packets (less than 160 bytes) are transmitted at a high throughput.

ATM UNI and HSSI

The problem
that caused a HSSI card to report a bad function in the trace
area when an SNMP
get

of a FEAC status was submitted to a switch
containing both an ATM
-
UNI and a HSSI card has been corrected. The
problem occurred because the HSSI card was inadvertently picki
ng up the
SNMP get of a FEAC status that was actually being sent to the ATM
-
UNI
card.

Trunk Protocol

On a HSSI trunk card, if a trunk line was connected to one of the ports that
had a noisy line with numerous frame bit errors, the card would crash with
a
fatal diagnostic error of 41.0. This problem has been corrected.

Proble
ms Corrected


Ascend Communications Inc
.


P
age
26


4.5.6

ATM


Issue

Resolution

ATM CS/IWU card in a 4.1
switch

An ATM CS/IWU card no longer loads the wrong boot flash when
inserted in a 4.1 switch. When the switch is later upgraded to 4.2, the
CS
/IWU card will load the correct code and boot without failure.

9000 cell trunk not showing
statistics (C97
-
11515)

You can now display statistics for the B
-
STDX side of a CBX to B
-
STDX
cell trunk.

ATM UNI CS cards

CP boot operations no longer cause the AT
M CS cards to stop responding
with the NMS message “response too big”.

TSU byte frames

qhe Aqj ftr card no longer has problems receiving TSU byte frames on
the Aqj interfaceK

Aqj service interworking

qhe problem of Aqj service interworking circuits not
passing Aqj
traffic that has the CmCp bit set has been correctedK

kkf ports for CpLftr
cards

Aqj CpLftr cards now support kkf logical ports typesK

iogical port bandwidth

qhe logical port bandwidth between VMMM Cp and RMM apP ports is now
equalK

Aqj rkf
Cp cards

qhe Aqj rkf Cp card now supports traffic shaping featuresK

aefined paths on Aqj
cards

qhe problem of using defined paths for interworking circuits EAqj cardsF
has been correctedK

Circuit statistics wrong

fnterworking circuit throughput statistic
s are no longer doubled on the Cp
or ftr endpoint of the circuitK

eot inserting flms

qhe Aqj ftr lCPLpqj
-
N card no longer goes into red alarm or
experiences any cell errorsLdiscards when
hot swapping

another IOP.

OPTimum cell trunks

ATM cards now support

more than one ATM OPTimum cell trunk.

ILMI support

The problem with logical ports configured on the DS3 CS card and IWU
OC3/STM
-
1 card with ILMI enabled not coming up has been fixed.

Physical port statistics
wrong

The physical port statistics screens n
ow provide an accurate number of
frames, bits per second, and octet counts for CS and IWU physical ports.

Physical port Summary
statistics

Summary transmit statistics for the physical port of a trunk are now
correctly reported.

LMI

In order to allow Fram
e Interworking circuits to be established quickly, the
LMI handler has been modified to get an LMI exchange 3 seconds after
“link up” rather than waiting until the T391 timer has expired.

The ATM NNI port now defaults to LMI “link up” rather than “link dow
n”
until the first successful ijf pollK

Aqj ketwork fnterworking
msCs

Aqj ketwork fnterworking circuits now become active without a long
delayK


4.5.7

SMDS

Issue

Resolution

SMDS PDU violations

SMDS over ISDN no longer shows SMDS PDU violations with SA not
fo
und. The SIP3 Header area for the SMDS PDU Violation Dialog screen
displays the ISDN E.164 address in ASCII format.

Modify IA area mask

Modification of the IA area mask no longer time out if traffic is running.
The set
completes prior to NMS the timeout p
eriod.

SMDS areas missing

The area table is no longer missing SMDS areas following an upgrade to
04.02.00.00.

Problems Corrected


Ascend Communications Inc
.


P
age
27



Issue

Resolution

SMDS Optimum Trunk/GA
traffic

GA traffic arriving on an SSI with an optimum trunk path to a node with
the GA area no longer

causes the card to become CPU bound. In a
previous release this issue occurred under the following conditions:



GA traffic entering node 1 on a DXI that is multiplexed to the SSI at
node 2 (and carries the optimum trunk)



Group Address exists at node 2 and
the GA area is known at node 1

SMDS GA traffic on

DSX and UIO cards

(C96
-
8795)

When DSX and UIO cards receive SMDS GA traffic at a very high data
rate, this causes the cards to crash with memory corruption. This problem
has been corrected.

SMDS header

T
raffic from an SMDS DXI
-
DTE port through an SSI
-
DTE port failed at
the SMDS DSU because the direction bit in the DXI2 header was
incorrect. This problem affected IA frames only. The problem has been
corrected.

SMDS vector table
corruption

The problem wher
eby a 4
-
port T1/E1 card would fail due to vector table
corruption upon receiving excessive amounts of group addressed traffic
has been corrected.

SMDS GA on IOP
-
B

(Solution: S96
-
808)

SMDS GA handling on the IOP
-
B platform has been fixed to prevent a
frame

buffer leak where a trunk (Direct or Optimum) exists on the same
card as the SSI receiving the GA traffic.

SMDS DXI on HSSI card

(Solution: S96
-
714)

An SMDS problem has been fixed where if a DXI port is configured in the
same HSSI card as a Direct Trunk
then AIA frames input into the DXI
cannot be sent over the Direct Trunk to a remote HSSI. This problem only
occurred if the area of the AIA was not known to the switching system.

SMDS Path setup

During SMDS virtual path setup, the SMDS component allowed a

virtual
path to be setup even on a frame
-
relay only card. This problem has been
fixed.

SMDS Trunk between
clouds

When a direct trunk carrying traffic between two SMDS clouds is broken,
traffic reverts to an SSI which has a “management only” trunk. Previ
ouslyI
data would be erroneously discardedK

moAj pynch problems
with pjap logical ports

After a card reboots from a moAj syncI a logical port which is
operational and passing traffic will incorrectly send an
Lport Down

trap if
it receives no responses to
its heartbeat polls.

SMDS logical and physical
port statistics

The switch no longer reports invalid values in the SMDS logical and
physical port statistics.

SMDS Protocol Error
Checking

Protocol error checking now correctly occurs on SMDS traffic.

SMDS
traffic
oversubscription

Oversubscribing SMDS traffic on a 10
-
port DSX or 8
-
port UIO card will
no longer cause the card to crash with a 36.0 diagnostic error code.

Corrupt IA Screen Member
map

The problem that caused the PRAM upload to inadvertently indic
ate
differences in the IA screen map on a DXI of an E1 card that had many
members and was PRAM synched, followed by PRAM upload, has been
fixed.

SMDS SS

The problem that caused DXIs muxed to a remote SSI with particular slot
number associations to fail to

switch traffic has been fixed.

CP with large configuration

(C96
-
9153)

Control Processors with a large SMDS configuration may get shot by a
standby CP when warm booted. This problem has been corrected.

SNMP

If a Group Address is modified (added to or rem
oved from), while the
card(s) with the Individual Address(es) is down, the SNMP set will fail.


Problems Corrected


Ascend Communications Inc
.


P
age
28


4.5.8

Logical Ports


Issue

Resolution

Logical port conflict traps

When the user deletes a logical port on a switch and then adds a logical
port on a different car
d that uses the same interface number, the switch
would sometimes report an 'Lport Conflict' trap. This was due to the
original logical port being marked as deleted, but left in the interface table.
The deleted logical port will now be correctly removed fr
om the table.

4.5.9

PVCs


Issue

Resolution

PVC re
-
routes after


CP reset

While OSPF is reinitializing, if an IOPs PVC manager executes the
periodic load balance re
-
route, then the evaluated PVC(s) no longer
experience the effects of a load balance reroute if
a better route does not
exist. OSPF is re
-
initialized upon a CP reset (either cold
-
start, warm boot,
switch
-
to
-
redundant, or PRAM sync).

Resilient NNI/UNI (fault
tolerant PVCs)


PVCs now function properly following an upgrade.

PVCs fail to come up

The pr
oblem of circuits failing to come up under extreme conditions has
been corrected.

Switches with multiple
trunks

Circuits no longer fail to come up because they are trying to use a trunk
which has reached the maximum number of allowed circuits on a single
trunk. The OSPF cost of using this trunk is correctly calculated and
circuits are correctly routed to a lower cost trunk.

Asymmetric circuits

The problem of reverse circuit parameters being set to the same values as
the forward circuit parameters has bee
n fixed.

LMI

(C96
-
9437)

When a PVC down event is received during PVC re
-
route process, any full
status inquiry message will get a PVC state active response until the LMI
asynchronous update delay timer is expired.

PVC

Corrected the problem that caused va
lid data to be discarded when the
Admin Status value of a circuit is set to up before the logical port Admin
Status value is set to up. This problem would occur following an external
diagnostic test on a logical port.

Manually Defined Paths

PVC routes con
figured with defined paths requiring greater than 8 hops
would occasionally fall off the defined path and not recover without an
admin down/up of the PVC. The fix supports the documented limit of 16
hops.


Management DLCI

The problem that limited Manageme
nt DLCI on FR
-
UNI
-
DTE and FR
-
NNI has been corrected.

PPP
-
1490 Translation
Logical ports

Corrected a problem where circuits on a PPP
-
1490 Translation logical
ports on 10 port DSX
-
1 cards on ports 5 through 10 would cause a system
reboot when activated.

PV
C routes

Circuit routes are no longer torn down when the circuit is on a negative
virtual bandwidth trunk. However, the route may momentarily fail and
then recover.

PVC Manager

The Circuit Failure Reason is no longer incorrect when a PVC is
established fo
r the first time and the port at the destination IOP has no
modem signals.

PVC re
-
routes after CP reset

PVC no longer re
-
routes after CP reset.

Generation of FECNs and
BECNs

The problem where a card configured with a PVC with no congestion
generates FECN
s and BECNs has been corrected. Performance has
improved to full T1/E1 speed on 21 channels with 128 byte frames.

Problems Corrected


Ascend Communications Inc
.


P
age
29



Issue

Resolution

PVC Manager

(C96
-
8116)


(C96
-
8689)

Problem with re
-
routes not being performed correctly when older revisions
of switch so
ftware exist in the network has been fixed.

If a “switch to redundant” or a warmboot was done to a CP, spurious re
-
routes of msCs would occurK qhe algorithm to determine if re
-
routes are
necessary has been corrected for this situationK kote that leaving th

switch tuning parameters at default levels may minimize the occurrence of
this problemK


4.5.10

UIO


Issue

Resolution

UIO physical port type
mismatch

If a UIO card's physical port type does not match the configured type, a
trap is generated to the NMS.


4.5.11

ATM

DS3


Issue

Resolution

DS3 Loopback

If you run a Far
-
End Loopback test followed by an External Test, the test
no longer fails with a “Bad Data Compare” or “No Data Received” error.

Aqj apP Cards

ECVS
-
RVNVF

Problem with executing ‘Internal Diagnostic’ whe
re test reported
“FAILURE: Bad Data Compare” or “FAILURE: Time out
-

ko data
Received” on good cards has been fixed.

Aqj apP ioopbacks

ECVS
-
TUVTF

then initiating a apP far
-
end loopbackI if the remote port was already in
loopbackI the line would become loo
ped at both ends preventing access to
the remote portK qhis problem has been corrected by ensuring that a looped
back signal may not set the local port in far
-
end loopbackK

Aqj apP ioopbacks

car
-
end ioopback fnitiated
status

ECVS
-
TUVTF

A new statusI car
-
e
nd loopback fnitiatedI may be reported at the kjpK
qhis status is reported when a far
-
end loopback is initiated from a physical
port on an Aqj apP cardK kjp OKMKOKNQ or greater is requiredK

Aqj apP ioopbacks

iogical port remains down

ECVS
-
TUVTF

fnitiati
ng a far
-
end or line loopback from an Aqj apPLbP logical port
would cause the port to remain down after attempts to clear the loopbackK
qhis has been corrected by requiring all apP loopbacks to be initiated at
the physical port level on Aqj apP cardsK

Aqj

bP lmqimum qrunk
oedundant cards

qhe problem that caused the Aqj bP lmqimum trunk redundant cards to
crash with internal error PSKM and switch over after a system reboot has
been fixedK

apP iine ioopback

qhe loopback status no longer displays incorrect m
ode or invalid values
for the oeceived cbAC ptatus on a apP card that has frame chip mjTPONK

apP iine ioopback to
mayload ioopback or vice
versa

qhe problem that caused the loopback status and the received cbAC status
to incorrectly reflect the change whe
n running a line loopback test
followed by a payload loopback testI without running the stop test in
between testsI or vice versaI has been correctedK

aata corruption

aata corruption no longer occurs after you change the apP physical port
Admin ptatus to
Down
.


Problems Corrected


Ascend Communications Inc
.


P
age
30


4.5.12

DS1 and DS0 Loopback



Issue

Resolution

DS0 Far End Loopback

The following DS0 far
-
end loopback issues have been corrected:



When you send the test bit pattern to a DS0 that is in far
-
end
loopback, it no longer goes out of sync.



The NMS prevents y
ou from selecting multiple DS0s for far
-
end
loopback testing. This prevents the port from getting stuck in far
-
end
loopback mode.



When deactivating a DS0 far
-
end loopback an appropriate trap is now
sent to the NMS

Loopback

(Solution: S96
-
924)

A DS0 or DS1
, which is already in loopback, will no longer accept another
loopback activation from being set until the first is stopped/disabled. Only
one loopback is allowed at a time.

DS1 Loopback

Nx56 Far
-
end out
-
of
-
band is now supported.

4.5.13

BERT Testing


Issue

Reso
lution

BERT Pattern

The problem where the pattern “3 of 24” generates an incorrect pattern has
been correctedK

Bboq Bit brror oate

vou can inject an error rate at a single bit rateK

Bert lut of crame

iower byte user
-
defined Bboq patterns no longer resul
t in Bert lut of
crameK


4.5.14

Bulk Statistics


Issue

Resolution

Negative Values in Reports

The problem whereby a Node ID greater than 128 bytes resulted in
negative values in the Bulk Statistics reports has been corrected.


Memory Corruption

The memory corr
uption problem in Bulk Statistics collection, which
resulted in Timer Chain Corrupted error, has been corrected.


4.5.15

ISDN


Issue

Resolution

Card crash

DS3 card no longer crashes when a cable is removed from a 4
-
port ISDN
card.

IPX over PPP

(C96
-
9445)

IPX
over PPP traffic traversed anywhere along the circuit path would crash
a high speed IOP card.

PPP Echo request Timeouts

cause physical port traps

(C96
-
7997)

If CPE equipment over a PPP link passing through a DSU have the Admin
value set to down, PPP echo
request time
-
outs recognize this condition and
report circuit down with UNI/NNI down as the reason code. However,
physical port down and up traps were also inadvertently reported. This has
been corrected.

Inverse ARP

(C96
-
9280)

Inverse ARP responses and g
ratuitous ARP requests from a PPP
-
1490
logical port through a switch to a frame relay CPE would not succeed if
passing through a trunk.

Problems Corrected


Ascend Communications Inc
.


P
age
31



Issue

Resolution

IPCP configuration

(C96
-
9412)

IPCP configuration request commands issued without the IP address
opt
ion are now ACK’d with no IP address provided rather than NAK’d
with fm address MKMKMKM returnedK

klqbW qhe above is utilized on an fm unnumbered serial portK fm
unnumbered serial ports on routers need another serial port or bthernet
port on the same route
r with an fm address configured and operational
before the unnumbered serial port will pass dataK

fpak mmm
-

dratuitous Aom
feature

fn this releaseI an individual dial
-
in user is mapped to a

nailed up

PVC.
Cascade maintains a local ARP cache, mapping the
DLCI to the client IP
address. As dial
-
in users connect and disconnect, the ARP cache is
updated and the PVC state changes are communicated to the egress routers
via LMI asynchronous status updates. The egress router then transmits
inverse ARPs and the ARP
s attempt to resolve changes in the DLC I state.
The B
-
STDX software intercepts the inverse ARP in its 1490
-
to
-
PPP data
stream and responds with the appropriate information. This configuration
works well except in instances where a particular dial
-
in clien
t (or specific
IP address) or DLCI is reused before the egress router has updated its ARP
cache.

Reusing the same DLCI before the egress router has updated the ARP
cache, causes a scenario where if a second call is placed on the same DLCI
with a different
IP address, it is not recognized by the router, and data will
not pass. Gratuitous ARP ensures that when a new call is accepted by the
B
-
STDX, the new IP address associated with the DLCI is updated in the
router’s cache.

偐m



A problem has been fixed in P
PP where an Echo Request that was not
8 bytes in length was ignored. PPP now sends an ARP when IPCP
goes into open state.



The BCHAN_MSG that is passed from the ISDN Call Control Task to
the LCP in the PPP stack was not being freed. This condition was
causi
ng a memory leak when calls were being placed and torn down.


ISDN

Corrected the following issues:



ISDN Timer Expires were causing the Telenetworks routines to be re
-
entered. This caused the globals that stored the PKT addresses to be
corrupted which in t
urn caused unpredictable results.



The ISDN main routines for each of the layers made a copy on the
stack of all of the ISDN messages that go through them. This was not
necessary because a common memory pool is used by all of the ISDN
tasks and the block ca
nnot be reused until it is freed. This became an
issue when the Timer Expires began to be posted to the ISDN Tasks
queues as the amount of traffic through these routines increased
significantly.



Timers that were being restarted were not being stopped then
started
so they would not expire when expected.



When the D
-
channel drops for any reason and Layer 2 has
information frames that are in the information queue waiting for
acknowledgment, the queue entry was not being properly removed.
As a result of the queu
e entry not being removed properly, the block
that was holding the information frame on the queue was being freed,
but the address of the block was not being removed from the queue.
This condition would cause the block to be freed twice which in our
system

causes a fatal error.

Restrictions and Special Considerations


Ascend Communications Inc
.


P
age
32


5.

Restrictions and Special Considerations

5.1

Networks with 800 or More OSPF LSAs

Networks with 800 or more OSPF LSAs require a CP+, CP40, or CP50 control processor card.

5.2

Login to Console

When trying to login to the console running Releas
e 4.2.3 and 4.2.4, be sure that your console cable has
pin 8 (DCD) on the switch side “high.” This is accomplished by connecting pin 8 to pins 20 or 4. The
cable that is currently shipped with the switch has the correct pin settings and will work properly.

Cables
missing pin 8 that worked in previous releases will no longer work with Release 4.2.3 and 4.2.4 or higher.
Crossing pin 8 to pins 20 or 4 to make DCD high enables modems to disconnect properly when exiting
from a switch.

The pinout is not required

in Release 4.2.5, however, you may have problems accessing uninitialized
switches.

5.3

Script Download

When you send an initialization script to a B
-
STDX switch via the console interface, a 2/10 second delay
is required between each set command.

5.4

Hardware

Do n
ot pull down the reset latch on the active CP at any time. If you want to remove the current active CP
card, switch the card to Standby mode, then remove it. For a single CP system, the switch should be
powered off before the CP is removed.

The DS0 loopbac
k feature is only supported on channelized T1s which are Rev C19 or greater. If you
have a card prior to Rev C19, the loopback feature on the NMS will be disabled (grayed out) for that card.

5.5

IOM

If you make pport parameter changes when the pport is admin’d

down, these changes may be ignored by
the switch software. Therefor, if you want to make changes to the physical layer parameters such as line
length or PLCP mode, you should make them when the pport admin state is up.

5.6

RIP

If you are using SunOS 5.5.1 (
System V Release 4) and want to use the RIP protocol to update the routing
table entries, you must run the in.routed process on the host system. Once the process is started, any
manual additions and deletions of the routing entries cause the table to freez
e. The table no longer accepts
updates from RIP. To resolve this problem, you must restart the in.routed process. In addition,

you may
also need to remove and/or flush the existing entries before restarting.

If you use RIP with multiple gateways on a flat
network (all gateway nodes and NMS server are on the
same LAN segment), you may end up managing a gateway node through another gateway node located
multiple hops from this node. This also means that nodes located close to gateway nodes may not be
managed t
hrough

that gateway node.

Before you enable RIP on a gateway node, verify that an appropriate firewall is in place to prevent the
router network and Cascade network from learning routes that are not meant for each other.

5.7

STDX Interoperability

This release
of switch software supports interoperability with STDX switches running switch code
versions 2.4.0.0 and above.



Restrictions and Special Considerations


Ascend Communications Inc
.


P
age
33


5.8

Software Download

The NMS is temporarily unable to reach the switch while the software is being committed to flash
memory during a B
-
STDX syst
em software download.

5.9

Real
-
time Clock

An NMS script sets the real
-
time clock for all switches that have a community name of
cascade.

If a
switch does not have a community name of cascade, the real
-
time clock is not set.

5.10

Connecting to B
-
STDX Through a Modem

Connection

If you are connecting to a B
-
STDX switch through a modem connection you must check and set the
modem as follows:



Enable Auto Answer



Override Data Terminal Ready

5.11

Switch Management


Issue

Description

Trunk limitation

Although a Release 04.02.02.
00 or greater network can support up to 50 trunks
per node, in a mixed software network (switches with 04.02.02.00 code and 4.1
code), any node that has over 40 trunks configured will fail to come up when
added and/or fail to recover when bounced. The trun
k appears to remain in a
loading state. To resolve this problem, delete the extra trunks (over the 40
limitation) on the node or nodes which have them. Until you upgrade the entire
network to Release 04.02.02.00 or greater, the trunk limitation remains at
40.

Multicast DLCI

The following multicast DLCI restrictions exist:



Service is temporarily disrupted if you delete a PVC from a Multicast
DLCI.



Do not use Multicast DLCIs with SMDS OPTimum trunks.

Install console
command

The install console command does
not support the initial configuration of
SMDS and ATM services.

Traps Generation

Dynamically disabling the Traps Generation field of a switch’s NMS entry has
no effect until you moAj pynch the switchK

janagement fnterface

vou cannot use multiple manageme
nt interfaces simultaneously on the same
switchK

Console interface
password

qhe console interface requires a user name and passwordK vou can use any user
nameI when the Cm has a moAj imageX the password is any of the defined
readLwrite community namesK ko
te that login and logout traps are sent to the
kjpK

pynchronizing moAj

fmmediately after you moAj pync a CmI there is a PM
-
second window of time
during which any kjp user activity may cause undesirable resultsK qhereforeI
it is recommended that you do not

perform any switch configurationI flm
moAj pynchI or any other kjp operations during this PM
-
second period of
timeK

pkjm set

then you add a multicast aiCf with OMM membersI the pkjm set may be
rejected and result in a “bad value” error for MIB object
ckt
jcastjemberiistKNOKNMNK qhe set succeeds when you reduce the number of
members to NMMK


Restrictions and Special Considerations


Ascend Communications Inc
.


P
age
34


5.12

IOP Type A


Issue

Description

Maximum IOP card
performance

Maximum IOP card performance has been tested at 12 to 14 Mbps,
depending on the network configuration.

IO
P traps

In situations where the IOPs are generating a large number of traps, the CP
may become overloaded and drop traps.

Redundant IOP Coldboot

When you coldboot redundant IOPs, there is no way to determine which
IOP will become the active IOP.

5.13

Logical
Port


Issue

Description

PPP
-
IPX encapsulation

(Point
-
to
-
Point) to

(Internet Packet Exchange
Protocol)

The Cascade switch supports only PPP to IPXCP RFC 1552. To get
IPXWAN support, you must attach a router between the Cascade switch
and the Novell network

that supports both IPXWAN and IPXCP.

Cascade
supports PPP encapsulation of IP with no option support (except IP
address assignment for ISDN), IPX with no option support (additionally,
no IPXWAN support) and Bridging with no option support.

5.14

T1/E1


Issue

D
escription

T1 physical port
configuration

The following T1 physical port configuration is not recommended due to
the possibility of a false alarm condition:

Link Framing Type:
D4,


DS0s selected:
1
-
24

Zero Encoding Method:
B8ZS
Logical Port Bit Stuffin
g:
OFF

Internal Foreground
Diagnostics

Executing T1/E1 Internal Foreground Diagnostics on a single logical port
affects all logical ports on the physical port.

Cable restriction

When you use a cable that is longer than 100 feet, the T1 card does not
work

unless it is connected directly to the network.

5.15

HSSI


Issue

Description

Frame errors

The switch reports statistics that include frame errors on a HSSI physical
port that is being overdriven on the receive. The count reflects BTUs
which are being dropped
. When this condition occurs the frame error
counts may exceed the rate of the frames coming in. In previous releases
the BTUs dropped were not reported in the physical port statistics.


Restrictions and Special Considerations


Ascend Communications Inc
.


P
age
35


5.16

ATM CS/IWU





Issue

Description

ATM OPTimum Frame
Trunks

The EFC
I and CLP bits are set to zero on all cells sent on a B
-
STDX 9000
ATM OPTimum Frame Trunk from an ATM CS/IWU card. Data
traversing an OPTimum Frame Trunk on an ATM CS/IWU card will not
map the DE and FECN bits to the CLP and EFCI bits. The DE and FECN
i
nformation is carried within the trunk header of the frame and will be
reconstituted at the end of the trunk. The DE to CLP and FECN to EFCI
mapping is done on all Direct and OPTimum cell trunk lports as well as on
NNI lports.

5.17

Power Consumption

The
-
48 V
DC and 120 VAC power supplies for the B
-
STDX switch have been tested to deliver 700 watts
of +5V power. When you calculate configurations, use 40 watts for IOP Type C cards (ATM UNI CS/ATM
IWU), use 45 watts for IOP Type B card (HSSI), use 40 watts for CP
cards and IOP Type A cards (T1, E1,
DSX, and UIO), and use 35 watts for the Channelized DS3 card.

5.18

Quality of Service (QoS)

QoS functions only if the entire PVC path supports QoS. This means that if any intermediate node does
not support QoS, then QoS will
not report statistics for that circuit.

The QoS Min/Max/Ave round trip delay is now calculated over a 15
-
minute interval instead of over the
life of the circuit.

5.19

SMDS


Issue

Description

Trunk card failure

After extensive trunk bouncing, it is possible to
crash the IOP Module.
This condition exists if there is a single trunk from a switch to the
network. The workaround for this problem is to create multiple trunks
between switches.

SMDS IA traffic over ATM
OPTimum trunk

If the SMDS frame BA size is > 8132,

SMDS IA traffic over an ATM
OPTimum trunk is discarded at the trunk port.

SMDS Group Members

You cannot delete a group address without first deleting all of the
members in the group.

SMDS Switching System

After you disable the SMDS Switching System (SS
) on a currently
functioning SS node, you must synchronize the CP’s PRAM and then
coldboot the Cm card before the disable will take affectK droup Address
traffic from a higher numbered node may still pass over this node’s trunks
until you coldboot the remo
te node’s CP.

Cm tarmboot

qhe B
-
pqau is temporarily unable to switch pjap user data when you
warmboot an active CmK

pjap qrunks

ln any system where pjap devices are configuredI make sure you set
the card service capability for all trunk cards to jultiser
viceK qhese cards
must physically be flm mlus cards Ethis applies for pjap device cards as
wellFK pjap data cannot pass over a direct trunk if the trunk is on a card
that has its service capability defined as crame
-
oelay lnlyK aue to the
nature of the pjap

switching system software Ewhich routes pjap
traffic locally without having to access the pjap cloudFI if you have any
trunk card configured as crame
-
oelay lnlyI it is possible that no pjap
data will be allowed to exit from the switchK qhis situation can
occur even
if other pjap trunk cards or ppf cards existK

Restrictions and Special Considerations


Ascend Communications Inc
.


P
age
36



Issue

Description

Multiservice

When you change an IOP card’s capability from Frame
-
oelay lnly to
jultiservice Eor vice
-
versaFI the moAj on the card is configured with a
different size layoutK qhi
s condition causes the card to fail itself and reboot
so that the new moAj layout can take affectK Background diagnostics for
the card displays a pervice Capability Changed diagnostic messageK kote
that when you replace an flm cardI if the replacement moAj

layout is
different from the configured layoutI the card will fail and reboot with the
new layoutK

pjap pp on NM
-
port apu

ln a NM
-
port apu cardI if the network passes bi
-
directional pjap fA
traffic at VKO jbpsI ports may begin to experience frame errorsK


pjap pp on O
-
port eppf

A O
-
port eppf card configured for pjap auf may experience discarded
fA frames on the opposite auf at OQKR jbpsK

fALdA area mask changes

then changing an fALdA area maskI it will take approximately NO
seconds for the switch to com
plete any updates of these area routing
tablesK ao not change the fALdA area masks until you upgrade the entire
network to MQKMOKMOKMM or greaterK ff you change these masks before you
upgrade to MQKMOKMOKMM or greaterI traffic delivery will not be reliable
K

Changing ppf muxing

qhe kjp now prevents you from changing aufLppf muxing on switches
running pre
-
MQKMOKMOKMM switch codeK oelease MQKMOKMOKMM changes the
implementation of this operation to make it easier to performI howeverI
these changes are not back
ward compatibleK

Class B addressing

ao not introduce Class B addressing into an pjap network until you
upgrade all nodes in the network to MQKMOKMOKMM or greaterK ff you use
Class B addresses before you upgradeI traffic delivery will not be reliableK

5.20

Ma
nagement Trunk

Management Trunks are supported only on networks in which

every node is running the following
minimum release versions of the switch software:



B
-
STDX, Release 04.01.00.00



STDX, Release 02.04.00.00

Do not enable management trunks on any node
in a network that is running a version of the software that
does not support management trunks.

5.21

ATM

The maximum ATM frame size is 8160.

5.22

ISDN

ISDN enables you to dynamically configure the IP address for dialing in, however, you must manually
configure the
subnet mask from the PC.

5.23

New Frame Relay Rate Enforcement

To use
jump

rate enforcement, circuits must be configured with a CIR equal to the Bc.

5.24

Frame Relay SVC Maximums

Frame Relay SVCs are available to users of the CascadeView/UX NMS Version 02.04.00.00 o
r later. If
you are using Version 02.04.00.00 or later of the NMS, you can now set and monitor switch
-
wide SVC
configuration parameters on each of your B
-
STDX 8000/9000 switches. See the
Cascade Documentation
Set Addendum

for more information about Frame R
elay SVCs.

Restrictions and Special Considerations


Ascend Communications Inc
.


P
age
37


5.24.1

Cards that Support Frame Relay SVCs

All low
-
speed cards (IOPA) cards support Frame Relay SVCs. In addition, the following high
-
speed cards
(IOPB) support Frame Relay SVCs:



HSSI



Channelized DS3

5.24.2

Card Limits

The following limits indicate the maxim
um numbers for Frame Relay SVC circuits for the IOPA and IOPB
images. These limits are based upon one address defined as the Calling Party, and one address defined as
the Called Party.

IOPA (Low
-
speed) Card Limits

Maximum number of Frame Relay SVCs per po
rt: 991

Maximum number of Frame Relay SVCs per card: 2214

IOPB (High
-
speed) HSSI Card Limits

Maximum number of Frame Relay SVCs per port: 981

Maximum number of Frame Relay SVCs per card: 1338

IOPB (High
-
speed) Channelized DS3 Card Limits

Maximum number of
Frame Relay SVCs per port: 991

Maximum number of Frame Relay SVCs per card: 4262

5.24.3

Node and Port Prefix and Address Limits

The following node and port prefix and address limits exist for Frame Relay SVCs:



64 node prefixes per switch



The total number of SVC p
ort prefixes + SVC port addresses must be less than or equal to a total of
512 for each card

There is also a limit of 8 SVC E.164 addresses for any IOPB or IOPA that is configured for Multi Service
mode. For example, you could potentially configure a chann
elized DS3 card with 4262 SVC circuits and
have all 4262 SVC circuits use only 8 different destination addresses.

You should also note that some cards (such as the T1) can be configured for Frame Relay or Multi Service
and the limits vary sharply based on

the configuration. For example, a T1 card that is configured for Frame
Relay service can support a total of 512 E.164 addresses. However, if the same card is configured for Multi
Service, the limit is 8.

Use the following steps to check on a card’s confi
guration:

1.

Select the switch.

2.

Choose Monitor

Cascade Objects

Show Detail. The Switch Back Panel displays.

3.

Select the card from the Switch Back Panel.

4.

Choose the Get Attr… option button. The View Card Attributes dialog box displays.

If you are not sure how
the card has been configured, select the card from the Switch Back Panel and
choose the Get Attr… option button.

Restrictions and Special Considerations


Ascend Communications Inc
.


P
age
38


5.25

Changing the Default Password

You can use the
set debug password

console command to change the debug password. The default
“mootman” password

is disabled when the debug password is changed.

NOTE:


If the default password is changed and the new password is lost, there is no way to access the
switch through the console.

5.26

IPX Traffic

If IPX data traffic is received on an Ascend PPP Lport, and forwarded
over an OPTimum trunk on an ATM
UNI CS or ATM IWU OC3/STM
-
1 card, the IPX traffic will be discarded.

5.27

Undocumented New Features

The following new features are not included in the
Network Configuration Guide for B
-
STDX/STDX

or the
Diagnostic and Troubleshoot
ing Guide for B
-
STDX/STDX
.

5.27.1

B
-
STDX 9000 ATM CS and IWU I/O Module Direct Cell Trunk

Shaper Priority Assignment

The section describes the mechanism used in this release to prioritize circuit traffic traversing a direct cell
trunk on the B
-
STDX 9000 ATM CS

and IWU cards.

In the initial release of the cell trunk, all traffic on the trunk was assigned to a single shaper for
transmission from the card. Frames were segmented into cells and the cells transmitted at the rate
configured. This mechanism gave all c
ircuit data the
same

transmission priority, so that VBR rt, VBR nrt
and UBR circuits all would send data equally.

This release uses the CS/IWU card’s NEC chip’s shaper priority to provide a means of
prioritizing

circuit
data for transmission. The NEC SAR c
hip performs the segmentation and reassembly function on the
CS/IWU cards. In addition it provides 16 shapers used to limit the speed at which data is transmitted from
the ATM port. The shapers are configured with a peak cell transmission rate, sustainable

cell transmission
rate, maximum burst size (MBS), and a shaper priority. The NEC chip services circuits associated with the
shapers depending on the priority of the shaper. The lower the priority value the higher the priority of the
shaper. Shaper 0 has t
he highest priority. Shaper 15 has the lowest priority. The CS/IWU cards reserve the
first shaper (shaper 0) of the 16 to be used as a line rate shaper (PCR=SCR=line rate). You can configure
the other 15 shapers on the Set ATM Physical Port Attributes di
alog box. An example of the default Set
ATM Physical Port Attributes dialog box for an IWU card is shown below.


Restrictions and Special Considerations


Ascend Communications Inc
.


P
age
39



5.27.2

Shaper Guidelines

To allow for the prioritization of circuits by the circuit’s QoS class, the switch code creates three groups of
shapers th
at are used for VBR rt, VBR nrt and UBR circuits. There are five shapers assigned to each group.
When you modify the shaper settings, the following guidelines apply:



Shaper priority set for shaper n+1 must be >= the priority set for shaper n



VBR rt circuit
s are assigned to shapers 1
-

5



VBR nrt circuits are assigned to shapers 6
-
10



UBR circuits are assigned to shapers 11
-

15

Shaper 0 is set to PCR=SCR and can not be modified. This shaper carries control circuits.

Assignment of a circuit to an individual sh
aper within a QoS shaper class is based on the SCR (CIR) of the
circuit. At connection establishment the QoS value is used to select a shaper group, then the circuit is
assigned to the shaper whose SCR most closely approximates the SCR (CIR) of the circuit
. The conversion
formula is the same as that used by the CAC when determining bandwidth allocation for circuits. For
example:

SCR
0

= CIR * IOH1 / 8,

where:

IOH1 is the Interworking Overhead
=


((N_Avg + AAL
HS) / 48)/N_Avg,


For an average frame size of 256, IOH1 is approximately equal to 0.02148

Restrictions and Special Considerations


Ascend Communications Inc
.


P
age
40


This enables you to set the shapers to whatever PCR, SCR and MBS desired within each of the three QoS
groups. The shaper entry priority field d
etermines the transmit priority of the traffic associated with the
shaper. For example, using the following shaper parameters, a circuit with QoS VBR nrt with an SCR =
50000 would be assigned to shaper 9 (second QoS group for nrt and 60000 is the next hig
hest SCR). A
circuit with QoS of UBR, SCR=80000 would be assigned to shaper 11. A circuit with QoS VBR rt , SCR=
49000 would be assigned to shaper 3.


Shaper Id

Sust. Cell Rate

Peak Cell Rate

Burst Tolerance


QoS







1

353207

353207

2



2

89285

3532
07

2



3

60000

353207

100



VBR rt

4

48000

48000

2



5

24000

24000

2









6

353207

353207

2



7

80000

353207

100



8

72000

72000

2


VBR nrt

9

60000

60000

2



10

44642

353207

2









11

353207

353207

2



12

72000

72000

2



13

50000

50000

2


UBR

14

35000

35000

2



15

22321

353207

2




The NEC chip schedules the shaper about every 70 ns. When this scheduling takes place a cell is
segmented from the frame associated with the circuit assigned to the highest priority shaper. The following
exa
mple shows how the priority works when multiple circuits assigned to different priority shapers all
present data to be transmitted simultaneously.

Using the shapers set in the example above, 7 circuits are active on a cell trunk. Circuits
VC a

and
VC b

h
ave QoS VBR rt. Circuits
VC c
,
VC d

and
VC e

are of QoS VBR nrt. Circuits
VC f

and
VC g

are UBR
circuits.


Circuit Name

QoS class

SCR




VC a

VBR rt

72000

VC b

VBR rt

68000




VC c

VBR nrt

44000

VC d

VBR nrt

44000

VC e

VBR nrt

44000




VC f

UBR

2
0000

VC g

UBR

20000


Restrictio
ns and Special Considerations


Ascend Communications Inc
.


P
age
41



The selected shapers are configured with:




Shaper #2

Shaper #10

Shaper #15

PCR

353207

353207

353207

SCR

89285

44642

22321

MBS

2

2

2

priority

1

2

3

For shaper #2 the I/M ratio is 1 cell in 4 cell times (S
CR/PCR = ¼)

For shaper #10 the I/M ratio is 1 cell in 8 cell times (SCR/PCR = 1/8)

For shaper #15 the I/M ratio is 1 cell in 16 cell times (SCR/PCR = 1/16)


Each of the circuits are assigned to the following shapers:

Shaper #2

VC a

VC b




Shaper #10

VC c

VC d

VC e




Shaper #15

VC f

VC g


The cell stream emitted for this configuration would appear as follows:

Slot

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

Shaper #2

a

b



a

b



a

b



a

b



Shaper #10



c

d



e




c

d



e


Shaper #15








f








g


Cell stream
vc a

and
b

have 1 cell sent every 4 cell times, then
vc c

and
d

send their cells, at this point the
scheduler recognizes that shaper #2 has higher priority and jumps back and segments a cell from
a

and
b

again. Then the scheduler retur
ns to shaper #10 and sends a cell for
vc e
. The scheduler finds no work to
do for shaper #2 or shaper #10 so it sends a cell from shaper #15
vc f
. The cycle then repeats.

5.27.3

Configuring the Shaper Settings

If you do not change the default Pport shaper setting
s, the cell trunk operation will remain the same as it
was in the previous software release. In this case all shapers are set to line rate and all priorities are
equal
.
If you use the default shaper settings, the switch will still assign the circuits based

on QoS to the first
shaper in each QoS shaper group, but all circuits will be serviced with
equal

priority since the shapers
priorities are equal.

NOTE:

You must use caution when setting the shapers for circuit prioritization when circuits traversing
the cell t
runk have the Graceful Discard option
ON
. With graceful discard
ON,

the frame relay
circuit can burst data up to the access rate of the ingress card. The potential problem is that SCR
is calculated using CIR which does not take into account the bursts allo
wed by Graceful Discard.
If the SCR set in the Pport shaper field closely matches the SCR, calculate from the CIR value then
when the circuit exceeds the Burst Tolerance value data will accumulate in the VC transmit
queues. If the source of the data contin
ues to exceed the CIR then discard is inevitable due to
shaper congestion.


Restrictions and Special Considerations


Ascend Communications Inc
.


P
age
42


5.27.4

G.826 Statistics

The G.826 statistics display performance monitoring statistics for digital transmission paths.

G.826 statistics are supported on the following B
-
STDX I/O modules:



4
-
port E1 card



Channelized E1 card

5.27.5

Displaying G.826 Statistics

To display G.826 statistics:

1.

On the network map, select the switch object that contains a 4
-
port E1 or channelized E1 I/O module.

2.

From the Administer menu, select Cascade Parameters
-
> Set Para
meters. The switch back panel
dialog box appears.

3.

Select (double
-
click) the 4
-
port E1 or channelized E1 physical port for which you want to display
G.826 statistics. The following dialog box appears. (For information about the Set Physical Port
Attrib
utes dialog box, refer to the
Network Configuration Guide for B
-
STDX/STDX
.)



4.

Select G.826 Statistics. The following dialog box appears.


Restrictions and Special Considerations


Ascend Communications Inc
.


P
age
43




The following table describes the E1 Statistics fields.


Field

Description

Switch Name

Displays the name of the
switch

IP Address

Displays the IP address of the selected switch.

Pport ID

Displays the physical port ID.

Refresh Time

Displays the time of the last refresh.

E1 Current Statistics


Errored Blocks

A count of detected block errors (frames with CRC or
fr
aming errors) within a 15
-
minute interval.

Errored Seconds

A count of one
-
second intervals with one or more
errored blocks or at least one defect within a 15
-
minute
interval.

Severely Errored Seconds

A count of one
-
second periods containing


30% errored

blocks or at least one defect within a 15
-
minute interval.

Background Block Errors

A count of block errors detected but not occurring as part
of a severely errored second within a 15
-
minute interval.

Errored Seconds Ratio

A ratio of accumulated errored
seconds to total seconds
in the available time.

Severely Errored Seconds Ratio

A ratio of accumulated severely errored seconds to total
seconds in the available time.

Background Block Error Ratio

A ratio of background block errors to total received
block
s (E1 frames) in the available time.


Usage Based Billing


Ascend Communications Inc
.


P
age
44


6.

Usage Based Billing

Usage
-
based billing for SMDS requires the use of a real
-
time clock that is present on all CP/CP+ cards. This
clock is used to schedule billing aggregation periods and mark the beginning and ending

of billing periods
within usage records.

Whenever you insert a new CP card into a switch, use CascadeView/UX to set the value of its real
-
time clock.
This ensures that the real
-
time clock on the switch remains synchronized between a redundant CP pair and
the
clock on the CascadeView/UX station.

6.1

Bringing Up a New Switch

You can configure Billing on a switch at either of the following times:



When the switch is off
-
line (the switch is unmanaged)



When the switch is on
-
line and actively being managed

The clock

is automatically set when billing parameters are changed while the switch is reachable and
being actively managed. Thus, CascadeView/UX sets the clock when you enable billing after you perform
the initial switch configuration (i.e., after the switch becom
es reachable).

If you enable billing while the switch is off
-
line and unreachable, the clock may not be set properly. Thus,
you must set the clock manually immediately after the switch becomes reachable and before the CP cards
are PRAM
-
synched. If you set
the clock after the CP cards are PRAM synched, you should reset the CPs
so that the NMS can correctly schedule the billing software's aggregation period.

6.2

Installing a new CP card

You must manually set the real
-
time clock whenever you install a new CP card
into an existing switch (for
example, when a failed CP card is swapped
-
out or when a redundant CP is added). To avoid affecting the
billing operation, set the clock immediately after the installed CP card becomes operational.

6.3

Setting the Clock

To set the c
lock from CascadeView/UX:

1.

Select the desired switch on the network map. A selected switch appears highlighted.

2.

From the Administer menu, choose Cascade Parameters


Set Parameters.

3.

Choose the Set Sw Attr command button.

4.

Select the Billing button.

5.

Choose A
pply. The real
-
time clock is set when the Apply button is activated on the CascadeView/UX
Billing screen.

NOTE:

Note that CascadeView/UX automatically sets the clock once per day. For more information about
this, refer to the Cascade
SMDS Billing System Administ
rator's Guide
.

Known Anomalies


Ascend Communications Inc
.


P
age
45


7.

Known Anomalies

7.1

Anomalies in 04.02.05.00


Issue

Description

Upgrade from 4.2.2.6

When upgrading from version 4.2.2.6, the primary CP can no longer
see the standby CP when issuing a
show card

or
show sys

command. To workaround this proble
m, issue the
show software
flash

command.

PRAM differences

When you perform a PRAM Compare, PRAM differences will be
noted for cktAtmQoS and cktAtmRQoS on Management DLCIs. This
discrepancy has no operational impact on the switch.

Defined Paths

When a
circuit on a defined path is deleted, the defined path object
remains in PRAM. This may result in circuit set failures due to the
maximum number of defined path circuits. To workaround this
problem, delete the defined path prior to deleting the circuit.

B
-
STDX makes the router
go up and down every
second (C97
-
16722)

The B
-
STDX IW circuit will continuously bounce while receiving
OAM cells.

ATM Management
VPI/CVI on ATM CS/IWU
ports

You must reset the port on ATM CS/IWU cards after the configuration in
or
der for the ATM Management VPI/CVI to function.

Trunk on CS/IWU cards
sometimes fail to come up

Trunks configured on cell cards (ATM CS/IWU) sometimes fail to come
up following a cold boot. To recover the trunk, you must reset the far
-
end
trunk endpoin
t on the card.

Bulk Statistics collection
occurs at the top of the hour
when the collection interval
is set to '0'.

When the Bulk Statistics collection interval is set to '0' on the switch, a
TFTP transfer occurs for that switch at the top of the hour. E
xpected
functionality is that Bulk Statistics TFTP transfer is disabled when the
NMS interval is set to 0.

Circuit lost counts are not
correct.

Circuit lost counts (cktOutLostFrames, cktOutLostOctets,
cktOutLostDEFrames, and cktOutLostODEFrames) are inacc
urate.
Instead, they are instantaneous snapshots that are reset frequently by QOS
processing.

7.2

Anomalies in 04.02.03.00

7.2.1

General

Issue

Description

Switch Authentication

On switches which have been configured for authentication, if it is
disabled, you will
not be able to enable debug mode on the console.

G.826 Performance
Statistics

The accuracy of the statistics have been observed to be off by
approximately 2%.

Invalid Interface Numbers

When performing a query on the MIB
-
II object, the if (interface) numb
ers
associated with OPTimum trunks are reported as “invalid”. The
configured feeder ports associated with these OPTimum trunks report as
usual.

PRAM Conflict When
Moving an IOP Card

Moving an IOP card from one switch to another will cause the IOP to
rep
ort a “PRAM Conflict” condition. After this condition, a PRAM sync
operation may cause the IOP card to crash. The CP will shoot the card and
the IOP will come up successfully without any intervention.

Known Anomalies


Ascend Communications Inc
.


P
age
46


7.3

Anomalies in 04.02.02.00

7.3.1

General


Issue

Description

Mismatched subnet masking
with RIP enabled

If there is a router on the same LAN as the gateway switch, and:




Both the router and gateway switch have RIP enabled



The router has a Class C subnet mask (255.255.255.0) and the
gateway switch has a Class B sub
net mask (255.255.0.0)


Then the switch may respond to some RIP updates with "frame too small"
packets. The background diagnostics for the CP may also show a non
-
fatal error of 2.2 "outbound heap low".

ATM Frame Optimum
trunks

OSPF and traffic delivery p
roblems exist on ATM frame optimum trunks
that are configured on frame
-
based cards other than ATM UNI. When this
condition occurs you will see large negative bandwidth values on the
trunk. Circuits on this trunk will not load balance off of the trunk and

new
circuits may be routed over this trunk.

Incorrect red frame
calculation

Red frames are calculated incorrectly when a card is busy. This condition
occurs on circuits that are configured with a rate enforcement value of
simple.

Transmit Discards

(C96
-
6332)

Transmit discards and congestion can occur on a looped V.35 port
operating at 6 M.

Direct trunk to a 6000

A direct trunk connection to an STDX 6000 may not come up until the
9000 card is rebooted.

Loopback status inaccurate

A failed far
-
end loopbac
k activation may be reported as successful.

T1 direct trunk

A T1 direct trunk may not recover after a cold boot of the switch.

QoS Statistics

(C95
-
3359)

QoS Statistics may be inaccurate by 5 percent.

Circuits backed up to a DTE
backup port

Circuits back
ed up to a DTE backup port do not always come up.

Standby CP getting shot

While downloading switch software to a switch configured with a
redundant CP, the standby CP will sometimes get shot.


7.3.2

Loopbacks


Issue

Description

DS1 closed loop

When running an

unframed CSU Line loopback test on a DS1 port of a
channelized DS3 card, it is possible for a closed loop condition to occur on
the DS1 port. When the far
-
end device loops back the loopback code, the
DS1 also goes into loopback.

Loopback state of OC3 an
d
Channelized DS3

Traps that indicate whether a port is in or out of a loopback state are not
generated for all loopbacks on an OC3 card or for DS1 loopbacks on a
channelized DS3 card.

7.3.3

Channelized DS3


Issue

Description

Adding DS3 card

The addition of a
DS3 card occasionally fails when doing multiple adds
and/or deletes on the same DS3 card..

Known Anomalies


Ascend Communications Inc
.


P
age
47


7.3.4

T1/E1


Issue

Description

E1 configured for 56K

If a logical port is configured for 56 K on an E1 ISDN card, the card
crashes when a call comes in.


7.4

Anomalies in 0
4.02.00.00

7.4.1

General

Issue

Description

MIB walk

You cannot perform a complete MIB walk with this release.

Multicast DLCI frame
discards

Cards discard data when the multicast circuit contains many members.

TFTP Server

If you PRAM synch a card when the TFTP

server is not running, the card
becomes unreachable from the NMS, until the PRAM synch operation
times out.

Erase PRAM

If you use the NMS Erase PRAM for an IOP card in maintenance mode,
the switch rejects the SNMP request, and PRAM is not erased. If a ca
rd is
in maintenance mode, use the switch console to erase PRAM.

Switch unreachable at the
time of statistics request

(C95
-
2080)

If an IOP is extremely busy (100% CPU utilization) card is busy and
physical or logical port statistics are requested for the
associated port, the
switch becomes unreachable until the statistics window is closed.

Provisioning PRAM
compare differences

Refer to the
Software Release Notice for CascadeView/UX Version
02.04.03.00

for more information about PRAM compare/upload.

Upgra
de PRAM compare
differences


A switch configured with a non
-
redundant CP displays a PRAM
difference following a software upgrade.

Foreground diagnostics

Foreground diagnostics fail on full channelized T1/E1 cards.

OSPF
-

Ethernet port down

A switch with
a configured gateway Ethernet address will not stop
advertising the route if the Ethernet port goes down. To avoid this
problem, any switch that is not intended as a gateway should not have an
address entered on the CP. If a switch was configured as a gate
way, and
the Ethernet is unplugged, the Ethernet address must then be deleted and
the CP warmbooted to delete the advertised OSPF entry.

Rev 1 LMI

Rev 1 LMI does not work on HSSI, CS DS3, or IWU OC3/STM
-
1 cards.


7.4.2

CP


Issue

Description

Redundant CP card
s

Continuously toggling the active/redundant CP cards may cause IOPs to
reset.

7.4.3


T1/E1

Issue

Description

Frame errors

Frames of 8190 to 8192 in length on T1/E1 cards are dropped as a “frame
error” when received by a physical port at the switch.

Frame dis
cards on E1
interface (C97
-
12000)

E1 statistics report frame discard errors. These errors are reported when
the port receives CBX 500 signaling.

Known Anomalies


Ascend Communications Inc
.


P
age
48


7.4.4

DSX
-
10

Issue

Description

Background diagnostic error
27.2 (bus transmits full)

DSX cards may report a backgr
ound diagnostic error of 27.2 (bus
transmits full) when the card is heavily stressed. Such conditions include:
a card that is fully configured with trunks and PVCs, and the maximum
utilization is defined for all trunks.

7.4.5

HSSI


Issue

Description

Hot swappi
ng with UIO
cards

Hot swapping a UIO card that is switching SMDS frames to a HSSI trunk
may cause the HSSI IOP to lose its connection to the CP.

Overclocking bandwidth

Overclocking the HSSI card beyond 45 Mbps may cause it to fail.

Foreground diagnostics

on
HSSI logical ports

Foreground diagnostics executed on a HSSI logical port causes the
operational status of the physical port to remain down. Bounce (admin
down/up) the physical port to correct this problem.


7.4.6

ATM Features


Issue

Description

VPC circui
ts not supported

VPC circuits will not pass traffic over a cell trunk configured on any of the
ATM cards supported on the 9000.

OPTimum cell to direct cell
trunk

When you reconfigure a trunk from a direct cell trunk to an optimum cell
trunk, you must perf
orm a PRAM sync for the trunk to come up.


7.4.7

ATM DS3/E3 Cards


Issue

Description

ATM PVC

ATM FUNI to FUNI PVC allows only 2M of data to pass through when
the CIR = 20M.

ATM PVC

You cannot create an ATM to ATM PVC where VPI x 256 + VCI > 1023.
The switch

rejects the set with “Invalid length or type”.

Aqj apPLbP trunk

ff you perform a coldboot of an Aqj apPLbP trunk with the endpoints in
an Admin aown stateI then Admin rp the endpointsI the trunk may not
become operationalK qo work around this problemI yo
u must warmboot the
cardK

iink status down

qhe link status displays down if ijf oev N is used for an Aqj kkf
logical portK

Aqj bP card

A fully configured E3 card may crash with a 30.3 error “Heap out of
memory”.

7.4.8

Channelized DS3 Cards


Issue

Description

PRAM sync required

If you add a channelized DS3 card to a switch configuration for the first
time, you must PRAM sync the card after you use the NMS to configure
the card.

Known Anomalies


Ascend Communications Inc
.


P
age
49


7.4.9

SMDS


Issue

Description

SMDS traffic on non
-
switching nodes

The switching system
does not prevent SMDS traffic from using paths that
contain non
-
switching nodes. This may result in SMDS traffic not being
delivered.

Physical port summary
statistics

Summary statistics for the physical port of a trunk will not account for
SMDS Group Addr
ess traffic when the direction of the traffic is from a
lower
-
numbered node to a higher
-
numbered node.

The following error condition frames are not detected:

-

invalid link ID (set to 1)

-

L2 header subfield (set to any invalid value)

-

BA size too small

-

missing or invalid version field

SMDS Billing

If the Adjunct Processor is serviced by NIS, the AP installation changes
that you make to
/etc/services

have no affect.

SMDS paths

SMDS virtual paths will never get established to a switch if the best path
t
o the switch goes over a trunk which already has the maximum number of
circuits supported on it.

7.4.10

Console


Issue

Description

show card command

The
show card

command times out for 15 seconds on any card that is in
the process of booting up.

7.4.11

PVCs


Issue

De
scription

Zero CIR circuits

You may see some statistics which report green frames on zero CIR
circuits even when all frames are red.

PVCs fail to establish

Under extreme conditions (over 9000 PVCs and continuous trunk
bouncing), PVCs may not be establish
ed due to low memory. This is a
non
-
fatal error.

Manually defined paths

When you attempt to define a path over an active circuit with Admin
Status Up and an NNI endpoint, the circuit may fail with either “PVC
segment down beyond node 2” or “Admin Status D
own.” The workaround
for this condition is to bounce the circuitK qhe circuit should then come up
using the defined pathK

oesilient kkf reroute traps

oeroute traps are not generated when circuits are rerouted using resilient
kkf servicesK

Circuit fails t
o come up

lccasionally a circuit will fail to come up with a fail reason of not enough
bandwidthK qhis usually occurs when the bandwidth of the circuit is equal
to Eor slightly underF the available bandwidth on a trunk or logical portK

Circuits not load b
alance
rerouting

lccasionally some circuits will not load balance reroute to a better pathK
qhis usually occurs when the best path already has almost the maximum
number of circuits on it and there is an alternative path available for the
circuitK

Known Anomalies


Ascend Communications Inc
.


P
age
50


7.4.12

DS0 Loo
pback


Issue

Description

DS0 Far End Loopback

DS0 far end loopback does not get activated when a DS0 is not selected.
The NMS indicates that the loopback will be activated.

Multiple DS0 Far
-
end
Loopback enable/disable

If you enable and disable DS0 far
-
en
d loopback tests more than one time
on a DS0 that is on a physical port in a red alarm state, the following error
occurs when you disable the test a second time: “value given has the
wrong type or length”.

7.4.13

ISDN


Issue

Description

ISDN T1 card

The ISDN T
1 card supports short haul applications from 0 to 110 feet.

SMDS frames lost

Frames may be lost in an SMDS over ISDN configuration.

Card crash

Under extreme conditions, while continually toggling the Admin status of
a physical port on the ISDN card up a
nd down, the 4
-
port ISDN card may
fail.

PRI failure,

Cable Disconnect/
Reconnect

The PRI card may fail with an internal error 30.x while dialing many calls
at once.

New circuit not activated

When an MP call is established, the circuit is created for 1 DS
0 of the MP
bundle. If the DS0 the circuit is activated on is disconnected, the circuit
goes inactive but the other DS0s in the bundle remain connected. A new
circuit is not established for the next DS0 in the bundle.

DS0 not disconnected

If a DS0 of an M
P bundle (other than the DS0 the circuit is connected on)
is admin’d down, the call is not disconnected and the DS0 is not released
from the jm bundleK

Base fm address

Base fm Address assignment requires a warmboot in order to take affectK

mmm Authentica
tion failures

mmm Authentication failures on kon
-
fpak logical ports generates a trap
referring to fpak authenticationK

aynamic fm address
Assignment


qhere are issues with dynamic fm address assignmentK cor exampleI after
getting the client to connect su
ccessfully OO timesI at the OPrd callI the
Client reports a BAa
-
fm error messageK qhe switch sends a trap that states
the fm Address is in useK pince the client is dialing inI the switch should
assign a unique address to the clientK ff the address assigned

is in useI the
switch should assign the next available addressK

vou must warmboot the flm to recover from this errorK ff the client dials
in with a specific fm addressI then mmm convergesK

7.4.14

OSPF


Issue

Description

Duplicate NMS paths

If there are switche
s with duplicate NMS paths configured, it is possible
for OSPF paths to be calculated incorrectly if the switch’s CPs are
rebooted simultaneouslyK qhis may make the switches unreachable from
the kjpK qo recoverI warmboot the Cm of each switch separatelyK


Switch Software Upgr
ade Instructions


Ascend Communications Inc
.


P
age
51




8.

Switch Software Upgrade Instructions

This section describes the procedure for upgrading a B
-
STDX switch to the current 04.02.05.00 version of
software.

For information on installing a new switch into your network, refer to the
Network Management
Statio
n Installation Guide
.

8.1

Prerequisites

The Network Management Station must be running CascadeView
02.04.03.00
(for UNIX environment)
before you can install the B
-
STDX software supplied with this release in any switch. If you have not yet
upgraded CascadeView/
UX, refer to one of the following Software Release Notices (depending on the
NMS version that you are using) for more information about this NMS release:



Software Release Notice for CascadeView/UX Version: 02.04.03.00

NOTE:

Use the following upgrade path:


1. I
nstall the NMS.


2. Upgrade Bulk Statistics Version to Version 2.5.4.0 (if applicable).



Bulk Statistics Version 2.2.0.0
cannot

collect statistics from the upgraded B
-
STDX switches.
However,

Bulk Statistics Version 2.5.2.0 can collect statistics from the

following switches:

-

B
-
STDX switches running Switch Software Release 4.0.18 and higher

-

STDX 3000/60000 switches running Switch Software Version 2.3.7 and higher


3. Upgrade the STDX switch software.

Note that if you have STDX 6000 switches in your
netw
ork, you
must

upgrade to release 2.4.0.0 STDX software (at a minimum) before upgrading
your B
-
STDX switch to 4.2.5.0.



A B
-
STDX must be running release 04.00.xx.xx or greater before you can upgrade to this version.
If you are running a release prior to re
lease 04.00.xx.xx, you must first upgrade to release 04.00.x.
Refer to the appropriate
Software Release Notice

for upgrade instructions.

8.2

Upgrade Procedures

There are slight differences in the upgrade process depending on the hardware present in the B
-
STDX
switch.

NOTE:

Follow the upgrade procedures exactly as specified. PRAM upload and PRAM compare may
show differences if you do not perform a PRAM sync.



If you are upgrading a switch with a single CP, refer to
Installing the Software into a Switch with

a
Single CP

on page
52
.



If you are upgrading a switch with redundant CPs, refer to
Installing the
Software into a Switch with
Redundant CPs

on page
53
.

Switch Software Upgrade Instructions


Ascend Communications Inc
.


P
age
52


Using Multifile Download

If you are using UNIX NMS Version 02.04.02.00 or greater, you can use multifile download to install the
switch software. Depending on the type of hardware installed in the switch, use one of the following
upgrade pat
hs:



If you are upgrading a switch with a single CP, refer to
Installing the software into a Switch with a
Single CP Using Multifile Download

on page
54
.



If you are upgrading a switch with redundant CP
s, refer to
Installing the software into a Switch with
Redundant CPs Usi
ng Multifile Download

on page
55
.

NOTE:

Follow the upgrade procedures exactly as specified.

8.2.1

Installing the Software into a Switch with

a Single CP

This section provides

installation procedures for a switch with a single CP. Make sure that copies of the
software images (
.rom

files) are located in the
/tftpboot/cv_switchSoftware

directory before you
begin this installation.

1.

Verify that the

TFTP daemon is running. If it is not, refer to the
Network Management Station
Installation Guide
for instructions on how to configure the TFTP daemon to start automatically.

2.

From the CascadeView map, highlight the switch. Then, from the Administer menu, s
elect Cascade
Switches


Download Switch Software.

3.

Select the file
cp1.rom