Release Notes
for LS2020 Software Release 2.1.01

LSCle02937 Cell Payload Scrambling Must be Enabled for OC-3 Interfaces

Cell payload scrambling was a configurable parameter for OC-3 interfaces in Release 2.0, and it should not have been. In Release 2.1 it may only be enabled.

During the upgrade of a network to Release 2.1, there may at some point be a mixture of Release 2.1 and Release 2.0 nodes running in the network. If there is an OC-3 trunk connecting a Release 2.1 node to a Release 2.0 node, and either the Admin value or the Oper value for cell payload scrambling is set to disabled on the OC-3 trunk port of the Release 2.0 node, the trunk does not come up operationally.

Workaround

For each OC-3 trunk connection, set the Admin value for cell payload scrambling to enabled on the OC-3 trunk port of the Release 2.0 node, and make sure that the Oper value also gets set to enabled.

To check these values, use SNMP commands in the CLI. For example, if your Release 2.0 trunk port is on card 3, port 0, use the getsnmp command as follows:

cli>getsnmp clc1InfoAdminScramble.3000

Name: clc1InfoAdminScramble.3000 Value: 1

cli>getsnmpclc1InfoOperScramble.3000

Name: clc1InfoOperScramble.3000 Value: 1

If either of these values is anything except 1, enter the following commands in protected mode:

*cli>set config lock  (open the database for writing)

*cli>set port 3.0 inactive  (deactivate the port)

*cli>setsnmp clc1InfoAdminScramble.3000 1  (set to enabled)

*cli>set port 3.0 active  (activate the port)

*cli>set config unlock  (release the database)

Activating the port triggers the application of the Admin value to the Oper setting. The trunk should come up operationally.

If any OC-3 edge ports are configured with cell payload scrambling disabled, they should also be reconfigured as above.

LSCle02186 The trunkmon Tool Has Been Withdrawn From Release 2.1

The trunkmon tool is no longer supported in Release 2.1.

LSCle02935 The set port testing Command Has No Effect

The CLI set port testing command has no effect on the specified port. Use of this command is discouraged. Support for this command will be removed in a future release.

Network Management Caveats: The StreamView CFG Tool

LSCle01393 - Aggregate Circuit Bandwidth Not Compared to Port Bandwidth

Due to a limitation in the StreamView Configuration tool, the aggregate allocated bandwidth of the circuits configured through a port is not compared to the raw bandwidth or allocated bandwidth available through the port. Configuring circuits with aggregate allocated bandwidth greater than the port bandwidth will result in a failure to establish some of the configured circuits when the configuration is downloaded to the switch.

Workaround

To help ensure that all circuits configured through a port are set up, you must ensure that the aggregate bandwidth of the circuits does not exceed the capacity of the port.

LSCle01401 Creating an Instance of a Second Network Processor Card

The StreamView Configuration tool does not permit execution of the Copy operation on a Network Processor (NP) card in order to create a second instance of the card.

Workaround

To create a second Network Processor card instance, select the Empty tag for the second slot and create the new card using the Add function.

LSCle02679 Use Leading Zeros when Specifying MAC Addresses

Under some circumstances, the StreamView CFG tool requires Media Access Control (MAC) addresses to be specified with leading zeros, so that, for example, a MAC address such as the following is not accepted:

1:2:3:4:5:6

Workaround

When using the CFG tool, specify a leading zero in any MAC address field whose value is in the range 0x00 - 0x0F, as in the following example:

01:02:03:04:05:06

LSCle02684 Modifying Line Card MIB Objects While the Card is Down

It is not possible to use the Send All operation of the CFG tool to change the administrative status of a card from down to up while simultaneously changing other MIB variables associated with the card.

While the administrative status of a line card is down, no Line Card Control (LCC) process is associated with that card. As a result, certain MIB objects defined for the card are not available to the SNMP agent. When the Send All operation is used to update the configuration of a line card, the SNMP agent rejects some of the SNMP set operations normally forwarded to the LCC process associated with the card. Such rejections cause the overall Send All operation to fail, and the administrative status of the card remains down.

Workaround

Step 1 Enable the card by setting the administrative status to active. (Use either the CFG tool or the CLI command set card card# active.) Ensure that this is the only card parameter that is changed at this time.

Step 2 Download the change in administrative status via the "Send Changes Only" operation.

Step 3 Wait for the card to come up. You may monitor the status of the card via the CLI or via the Configure "Verify/Discard" operation.

Step 4 Continue the configuration operation once the card has come up.

LSCle02686 Value in Chassis Name Field May be Inaccurate

Under certain circumstances, the value displayed in the Chassis Name field of the StreamView CFG tool may be inaccurate.

For example, this may happen if a switch icon has been renamed using OpenView facilities, and is then selected in the StreamView Topology Map, and the Configuration tool is started from the Topology Map tool (in order to configure the selected switch).

Workaround

Use StreamView rather than HP OpenView to rename icons.

Start the Configuration tool and select the chassis you wish to configure from the list of switches shown rather than from the topology map.

Network Management Caveats: The StreamView PVC and VLI Tools

LSCle00956 StreamView PVC Tool Has No Save As File Menu Item

The PVC and VLI configuration tools do not support a Save As function.

Workaround

To create copies of the database, use the Save As option of CFG. Do not copy the .dir and .pag files with UNIX shell commands.

LSCle02872 Use Update: Send All to Update VLI Information

In the StreamView VLI tool, the Update: Send Changes Only operation does not work. The button has been deactivated.

Workaround

To update the VLI configuration of a switch, use the Update: Send All operation. This is the only active option.

LSCle00962 Deleted DLCIs May Not Be Reused Prior to Configuration Download

In the Release 2.1 PVC tool, under certain circumstances it is not possible to reuse the data link connection identifier (DLCI) previously assigned to a deleted Frame Relay circuit, even though the modified configuration has been successfully downloaded to the switch.

Workaround

Flush the current in-memory database image maintained by the PVC tool by selecting the Read DB button, then add the new circuit using the deleted DLCI.

LSCle02919 Verify May Report Changes in Default Secondary Scale Value

Following a Verify operation, the StreamView PVC tool may report a difference in the value of the secondary scale factor associated with a PVC, when in fact it is unchanged from the default value. The chassis reports the actual (default) percentage of the maximum rate, but the database reports the internal value as "default".

Workaround

If the secondary scale has not been modified, ignore the reported difference. The chassis value is written to the database when you apply the change.

LSCle02924 Number of Virtual Circuits (VCs) Not Compared to Configured Card Maximum

The StreamView PVC tool does not verify that the number of circuits configured on a port falls within the configured number of circuits maxVCs that may be configured on the port. Configuring a greater number of circuits than the allowed value results in a failure to establish some of the configured circuits when the configuration is downloaded to the switch. The same is true for frame relay circuits and the per-port Max Supported VCs value.

Workaround

To help ensure that all circuits configured through a port are set up, the number of circuits configured must not exceed the number of VCs allowed through the card or port.

• For a port operating in Frame Relay service, the maximum number of circuits which may be configured on a port is set by the value of the frDlcmiMaxSupportedPVCs management information base (MIB) object, which corresponds to the CFG port field Max Supported VCs.
  

• For a port operating in UNI service, the maximum number of circuits which may be configured on a port is set by the value of the cardMaxVCs MIB object, or the expert mode field Max VCs.
  

• The aggregate number of circuits on all ports on a card may not exceed the value of the cardMaxVCs MIB object in all service cases.
  

LSCle02941 StreamView PVC Tool sendupdate Causes Sun Crash

If you run StreamView on a Sun workstation using SunOS 4.1.4 (as opposed to 4.1.3), the workstation issues a panic message when you perform a SendUpdate function with more than 15 PVCs on a single port, using the PVC StreamView Tool. The message refers to the file ufs_lockf.c.

Workaround

Obtain patch #102264-02 from Sun and install it, as follows:

Step 1 Connect by ftp to sunsolve1.Sun.com.

Step 2 Log in using the anonymous userid.

Step 3 When prompted, use your login ID as a password.

Step 4 Change directory to the patch directory, using the following command:

cd pub/patches


Step 5 Enter the command binary to make the transfer in binary mode.

Step 6 Enter the command get 102264-02.tar.Z to copy the needed patch file.

Step 7 Enter the command ascii to make a file transfer in ASCII mode.

Step 8 Enter the command get 102264.readme to copy the readme file for the patch.

Step 9 Follow the instructions in the readme file to rebuild the Sun kernel.

Step 10 Reboot the workstation.

LSCle02947 Cancelling a Verify Operation and Executing a Save Operation May Cause PVC to Exit

Under certain circumstances, if you perform the following sequence of actions with the StreamView PVC tool, it may report an internal error and then exit:

1. Start a Verify operation against two switches.
   

2. Cancel the running Verify operation with the Cancel button.
   

3. Schedule a download to a switch.
   

4. Initiate the download by executing the Save operation.
   

Workaround

Avoid cancelling a running Verify operation. If you have cancelled a Verify operation and want to download a configuration next, execute the Read DB operation before starting the download.

Network Management Caveats: The Command Line Interface (CLI)

LSCle01068 Rate Statistics for Ports are Inaccurate

The Rate information produced by the CLI when displaying port statistics is not accurate.

Workaround

Increasing the interval between displays of the statistics increases the rate accuracy. Very short intervals have a small baseline, so that the brief delay between request and retrieval of statistics can significantly skew results. By waiting 10-20 seconds between display requests, the baseline increases, the delays become less significant (a smaller percentage of the baseline), and the rates become much more accurate.

LSCle01625 Use the CLI to Set IP Addresses to 0.0.0.0

In certain operational circumstances it may be desirable to set one or more of the IP addresses associated with the chassis and the Network Processor to a value of 0.0.0.0. Logically, this changes the status of the specified IP address to "not set". In Release 2.1, the StreamView CFG tool does not allow you to assign a value of 0.0.0.0 to an IP address.

Workaround

Use the CLI to set a chassis or network processor IP address to a value of 0.0.0.0.

LSCle02390 Turning Off tty Output Paging

By default, the LS2020 CLI uses a functional equivalent of the UNIX more command to paginate multi-line terminal output. In some circumstances this may not be desired.

Workaround

You may disable more-style output pagination by specifying the -nomore switch on the command line when you start the CLI, as follows:

LSNode:2# cli -nomore 

...

LSCle02634 The CLI May Falsely Report No Filters Defined or No Multicast Groups Defined

Under certain circumstances, such as when a high SNMP processing load is placed on the LS2020 SNMP agent, the LS2020 CLI show bflt, show ipflt, and show ipxflt commands may report that no filters exist, or the show mcast command may report that no multicast groups exist. This is due to the CLI failing to receive a response to an SNMP request associated with the command.

Workaround

Usually, failure to receive a response to an SNMP request is indicated by a message saying "No response - try again", or "Request timed out". When such a message appears in conjunction with a message that no filters or multicast groups exist, disregard the messages and repeat the show command.

LSCle02658 Multicast Groups: Non-LAN Ports are Accepted as Destination Endpoints by CLI

The Release 2.1 CLI tool permits you to define a multicast group containing one or more destination endpoints which are not local area network (LAN) service (for example, Ethernet or FDDI) ports.

Workaround

While it is possible to define a multicast group containing non-LAN ports, the LS2020 signalling and ATM management facilities does not establish a multicast connection branch terminating at a destination port other than a LAN service port. Therefore there is no danger that multicast LAN traffic will be accidentally delivered to a device attached to a Frame Relay, Frame Forwarding, ATM UNI, or circuit emulation edge interface.

LSCle02661 CLI May Fail to Identify MIB Version If Started Too Soon

If, during the course of a node reboot, the CLI is started prematurely, such that certain MIB objects are not yet available because the processes that have registered for them have not yet completely booted up, then it is likely that the CLI will display the following error message at start-up time:

Unable to determine MIB version... 

If this is the case, exit from the CLI and wait until a self-directed ping responds with success before restarting the CLI.

LSCle02921 The show card Command Does not Differentiate Between T3 and E3 Access Cards

The CLI show card display does not distinguish between T3 and E3 rate access cards. The display identifies E3 access cards and ports as T3 cards and ports. The correct port type is displayed by the CLI show port command.

Workaround

To resolve the ambiguity, use the CLI show port command to display the status of a port thought to be an E3-rate port.

LSCle02922 Setting Port Peak Cell Rate

The syntax of the command set card cardno peak-cell-rate cell-rate may suggest that one is setting the aggregate rate for all ports on the specified card. In fact, this command sets the per-port peak cell rate for each port on the specified card.

The syntax of this command will be made unambiguous in a future release.

LSCle02952 Delete Filter Operation May Report "Invalid Filter"

Under unusual circumstances, the CLI may report that a filter being deleted is invalid, even though the filter is deleted successfully.

Workaround

Use the show bflt, show ipflt, or show ipxflt command to verify that a filter exists and is valid before you delete it. After you delete the filter, use the same command to verify that it has been deleted.

Network Management Caveats: The LightStream Topology Map

LSCle02044 Adding Switches to an Existing Topology Map Domain

The StreamView Topology Map tool may not allow you to add a LS2020 switch to an existing LS2020 domain.

Workaround

To include a switch in a domain, you must recreate the domain. Delete the original domain, then select all icons desired for the domain (including the switch that is to be added), and recreate the domain.

LSCle02657 LS2020 Domain Icon May be Hard to Find

A group of LS2020 switches and interconnecting trunks are grouped together to form a Topology Map domain, and the domain icon is placed in the HP OpenView New Object Holding Area. If the Holding Area contains many new objects, not all objects may be completely visible on the display. As a result, it may seem that the domain icon has disappeared.

Workaround

In this situation, check the entire New Objects Holding Area for the iconified domain. Reduce the height of the window. This scales all objects smaller. The domain icon should be visible at the end of the New Objects Holding Area.

LSCle02890 Termination Following a Close while Synchronizing

Under certain unusual circumstances, closing the StreamView Topology Map while it is in the synchronizing state may cause the program to terminate and generate a core file. This typically happens if the StreamView Topology Map (LTM) tool is terminated before initialization has completed.

Workaround

Wait for the StreamView Topology Map (LTM) tool to become fully operational before exiting.

Other Network Management Caveats

LSCle01141/LSCle01540 Stopping an Unwanted Download

No function is currently provided to stop a long download.

Workaround

It is possible to abort downloads with the LynxOS kill command. Log in as root to the network management system (NMS) station and use the command ps ax to find the process ID (PID) of each cfg_a process. Do not kill the cfg_a process with the lowest PID, it is the parent cfg_a process. Any cfg_a process with a higher PID is a child process, which you may kill. If more than one download is in process, they can be distinguished only by remembering the order in which the downloads were invoked (one with a lower PID was invoked earlier, and one with a higher PID was invoked later). However, even when the correct process is killed the results can be unpredictable, because some part of the configuration might have been downloaded and some not. It may be better to wait for a download to terminate normally.

LSCle01083 Use HP OpenView Release 3.3 or Later

Use HP OpenView Release 3.3 or later to manage your LS2020 network.

If you are using HP OpenView to manage your LS2020 network, it is strongly recommended that you use Version 3.3 or greater in order to avoid certain problems with the tool which may have an adverse effect on LS2020 simple network management protocol (SNMP) agent and system performance. Before Release 3.3, HP OpenView formats SNMP requests into a single buffer which can exceed the SNMP maximum of 484 bytes if a data collection has more than 20 variables.

Workaround

If you have HPOV Release 3.2 or earlier, when you set up data collection using HP OpenView, limit the size of collections to less than 20 variables.

LSCle01355 Specify Valid UNIX File Names With File Operations

Under certain circumstances while using the StreamView Save As operation, it may be possible to specify strings containing white space and nonprintable characters as file names.

While such file name strings may be accepted by the tools, they do not form valid UNIX file names. Creation of files so named may fail, or files so named may be difficult to manipulate later using standard UNIX mechanisms.

Workaround

When working with the StreamView tool set, do not use white space characters, non-printing characters, or control characters in file names.

LSCle02020 Avoid Concurrent Modifications to a Single Database

The StreamView configurator (CFG), permanent virtual circuit (PVC), and virtual LAN internetworking (VLI) tools should not be used concurrently on the same database. When one StreamView tool is directed to use or modify a database that is already being used or modified by another tool, it issues an alert that the database is currently in use and is locked. However, it is possible to dismiss this alert and continue concurrent modification of the database. Caution Doing so may cause loss of updates to the database from either or both tools.

Workaround

Avoid simultaneous modification of a single database using the CFG, PVC, and VLI tools. If you find yourself in the position of editing a single database with two tools concurrently, complete your edits with the second tool invoked, save, and exit. Then reopen the database file with the remaining tool before editing the file.

LSCle02025 Prompt to Save Changes Displayed Prior to Exit

Under certain circumstances, after you have asked to exit the StreamView Configuration tool, it displays a dialog box inquiring whether you wish to save changes to the database, even when no changes have been made to the database since the last save.

Workaround

If you are certain that no changes have been made since the last save operation, you may provide a negative response to the prompt. However, saving the database again is harmless.

LSCle02483 Avoid Storing Database Files in the /tmp Directory

Under certain circumstances a database file cannot be saved in the /tmp directory.

Workaround

When using the Configuration tool Save As function, avoid specifying the /tmp directory as the directory in which the database should be saved.

LSCle02620 Verify and Save of Many Nodes May Cause System to Exceed Available Memory

An attempt to save the configurations of many switches (more than 8) which have been acquired by the Verify operation may fail.

Workaround

When you use the Verify operation to acquire node configurations, save after verifying every eight nodes.

LSCle02715 Network Management Station May be Attached to a PLC Port

Network operations on the LS2020 NP are available from three connection sources:

• The test and control system (TCS) modem and console ports
  

• The NP Ethernet port
  

• A port on a packet line card (PLC)
  

Connections through TCS and the NP Ethernet port are well described in the documentation for this release. Connection through a PLC port is not.

An NMS connected to a port on a PLC is able to connect to an NP in the network if two configuration tasks are performed:

Step 1 Configure the IP address of the NM workstation.

Step 2 Enable the NP Traffic Filter on that port to forward connection requests to the NP.

The IP address of the NMS must be configured with a unique host number on the same network as the LS2020 Chassis IP network. For instance, if the Chassis IP address in an LS2020 network is configured as network address 123.45.6.0, the NMS must also have 123.45.6.0 as the network portion of its IP address. Secondly, if host addresses 1-10 are configured for Primary and Secondary NP Chassis IP addresses in the LS2020 network, the NMS must be configured with a host address other than in the range 1-10, i.e. 123.45.6.11.

The default configuration of each PLC port in this release is to block all traffic intended for an NP. The LS2020 private MIB variable that defines the action (forward or block) for PLC port traffic with an NP as its destination is lsLanPortNpTrafficFilter. The values to which this variable can be set are 1 (forward) and 2 (block). This release provides two ways of setting lsLanPortNpTrafficFilter:

• By using the CLI set port c.p npdeliver command
  

• By configuring and downloading the port attribute "forward" from StreamView
  

Use the following CLI commands to set port 1 of Ethernet card 6 to forward received traffic destined for an NP in the network:

cli> set config lock

cli> set port 6.1 np-deliver forward 

cli> set config unlock

Use the following CLI commands to set that port to block the same traffic:

cli> set config lock

cli> set port 6.1 np-deliver block

cli> set config unlock

The StreamView cfg program can also be used to control whether a PLC port forwards or blocks traffic destined for an NP. In the Port Configuration window, configure the NP Traffic parameter by selecting either the Forward or Block button, and then saving the configuration.

LSCle02769 No ifPhysAddress Object from ifTable Entry for NP Interfaces

The instance of the ifTable for the Network Processor (NP) does not provide an instance of the ifPhysAddress object.

LSCle02855 The Clear Button Does Not Terminate an Update

A user selects the option to update a node with new information from the database as follows:

1. Enter the appropriate parameters in the 'SendUpdate...Update Time' Menubar pulldown menu.
   

2. Select 'Save' from the 'File' pulldown menu.
   

Once the user does this, the process of updating the local node database becomes essentially irrevocable, unless the PVC StreamView process and the currently running cfg_a child process are both terminated.

• The PVC StreamView process can be terminated in the normal manner by selecting 'Exit' from the 'File' pulldown menu.
  

• The "cfg_a" child process, which is the independent process that actually performs the node update, can be terminated by using the Unix ps command to determine its Process ID number (PID), followed by a Unix kill command on that PID number. For more information on how to terminate this process, see the LSCle0111r1/LSCle01540 workaround in the "Other Network Management Caveats" section.
  

Do not misconstrue the Clear button on the Update Time Screen as an alternative method by which you can abort a pending node update. The Clear button has no bearing on the process of updating a node configuration from the global database. It is simply used to clear an entry from the scrollable list box of user-defined chassis-pair entries on the 'SendUpdate...Update Time' screen.

LSCle02857 Default Parameter Values Represented as "Unconfigured"

When a node is updated with a new PVC (particularly a virtual channel identifier (VCI) PVC), with most of the PVC values left blank, accepting default values, subsequent Verify operations may show something like the following:

Attribute Differences - Database   /  Switch

------------------------------------------------

Circuit chicago7.4.0,1-chicago7.4.1,1

Source IR/TD            unconfigured  109

Source Insured Burst    unconfigured  128

Source MR/MD            unconfigured  96000

Source Max. Burst       unconfigured  128

Dest. IR/TD             unconfigured  109

Dest. Insured Burst     unconfigured  128

Dest MR/MD              unconfigured  96000

Dest Max. Burst         unconfigured  128

The "unconfigured" value under the "Database" column is a reminder that these values were left blank, when the PVC was sent to the node during a SendUpdate operation.

LSCle02880 Card Parameters Not Resent When Card Reboots

During the course of some configuration activities, a card may reboot (intentionally) during a SendUpdate process. This occurs, for example, when the card type is changed from Trunk to Edge. Because the card reboots, parameters that are sensitive to whether or not a card is configured as a Trunk or Edge do not get set on the target node unless a second SendUpdate process is applied after the first.

To analyze a specific example, consider a card that is initially in Edge mode, where Peak (Port) Cell Rate, a Trunk-only parameter, does not apply. When the card is re-configured to Trunk mode, and a new Peak Cell Rate is entered for the card as well, the first invocation of the SendUpdate process only sends Edge card parameters, because that is the initial state of the card, and Trunk parameters do not apply. The card function is not known until the first configured port is set to Trunk (clc1InfoAdminProtocol = 1). The first Sendupdate sets the Peak Cell Rate to 0 rather than the configured value.

Workaround

Once the card has rebooted and switched to Trunk mode, a second SendUpdate process must be invoked to send the newly configured Peak Cell Rate.

LSCle02927 StreamView CFG: Deleted Filter Assignments may not be Reassigned

Under highly unusual circumstances, if a filter assignment or a PVC has been deleted from a port, but a download operation has not been performed, it may not be possible to reassign a filter to that port using the same filter ID, or a PVC using the same PVC identifier (DLCI or VCI number).

Workaround

In general, there is no need to delete a filter assignment and then reassign the filter. If you wish to modify a filter assignment, you may do so using the filter assignment Edit operation. However, if you do encounter this problem, perform a download after deleting the filter assignment. This allows the filter to be reassigned.

If you must reuse the identifier of a deleted filter assignment or PVC, download the switch configuration before you assign the identifier to a new object of the same type.

LSCle02943 E3 Framing Cannot be Changed from G.804

Attempts to change framing type for E3 ports from G.804 to PLCP appear to succeed, but in fact the framing type is not changed.

Hardware and Diagnostics Caveats

LSCle01539 MS1-CP and E3-PLCP Cards Wrongly Report Receive Error Rate

MS1-CP and E3-PLCP cards report a receive error rate of about 1/second when there are no errors. This is only a problem with the E3-PLCP card. It is not a problem with T3 or E3-G.804 card.

Workaround

A hardware solution is in process; contact Cisco customer support.

LSCle01956 DS3 UNI Rate Enforcement is Not Precise

The hardware and algorithms used to limit incoming data at an ATM UNI edge can allow more data than configured. The deviation is usually less than 1%, but can be greater for very high enforcement rates (e.g. greater than 1/2 of a T3).

Workaround

Allow for this deviation when configuring ATM-UNI edge ports.

LSCle02440 8-Port T3 Card Does Not Report Loss of Signal (LOS)

The hardware is unable to report LOS because of the line interface unit (LIU) chip used in the 8T3 front end circuitry. The LIU chip has a very sensitive receiver capable of detecting signals down to 35mV. This allows the card to detect signals over long distances or very poor grade coax.

Because of this receiver sensitivity, the LIU picks up noise when there is no signal on the line, and continuously resets the LOS detection circuit, with the result that LOS status is never reliably set.

A newer version of the LIU device will be available in the November timeframe. Until then when the hardware detects a loss of frame (LOF) condition, a LOS condition will also be reported.

LSCle02703 FDDI Card May Fail to Come Up at System Boot Time

When an LS2020 containing an FDDI card is booted the card may not come up completely. As far as we know, this is a very rare occurrence. The one time it occurred, the CLI show port <port> statistics command showed very large deltas for the Octets Rcvd and Normal Packets Sent statistics.

Workaround

If, on booting a node, an FDDI card fails to come up to the point that it passes data, disable and then re-enable the card. This should clear the problem.

LSCle02741 Diagnostics May Fail Test 70 on a Medium-Speed Line Card

The diagnostics software does not set the card up properly for this test, and it fails intermittently.

Workaround

Either deselect test 70 when running the diagnostics on a medium-speed line card, or ignore failures of test 70.

LS2020 Congestion Avoidance Caveats

Congestion Avoidance (CA) Not Enforced for Point-to-Multipoint Circuits

Traffic entering the network on point to multipoint circuits is not throttled at entry ports when congestion starts to build within the network. A consequence of this is that CA more severely throttles unicast traffic at entry ports. The effect is to give multicast traffic priority over the unicast traffic.

Switch Bandwidth is Not Taken Into Account for Call Admission

Switch bandwidth is not taken into account in call admission decisions.

Each line card has a 200 Mbs path (called its switch path) into the node switch. The cards are designed so that all port cards can run at line rate without overrunning the card switch path when all circuits through the ports are unicast circuits. Prior to the introduction of the Release 2.1 multicast feature it was unnecessary to take switch bandwidth into account in making call admission decisions because it was not possible to overcommit the card switch path.

When a multicast circuit branches at a node, incoming data is replicated for each branch at the card's switch path. Because Release 2.1 call admission does not take card switch path bandwidth into account, the following abnormal behavior is possible:

• A multicast circuit can be established that cannot possibly carry the requested bandwidth. Consider, for example, a 40 Mbs guaranteed circuit over an FDDI LAN interface branching to 5 other FDDI LAN interfaces (requiring 240 Mbs to the switch). This circuit would drop cells at the card switch path when attempting to carry sustained traffic at the allocated rate.
  

• The 200 Mbs switch path can become a congestion point if its bandwidth is overallocated; that is, if each individual circuit fits but they all do not fit together. Unlike trunks and exit ports, the card switch path is not protected by the CA mechanism. Therefore, the onset of congestion on the switch path does not force drops at the entry edges to eliminate the congestion. As a result, individual cells from packets would be dropped at the congested card switch path.
  

• When there is congestion on a card switch path, there is no mechanism to drop best effort traffic ahead of insured traffic. The cell CLP bit, the XCLP bit, and cell transfer priority are all ignored in the identification of cells to drop. This affects Unicast insured traffic as well as multicast insured traffic. No impact is likely on guaranteed traffic and higher priority traffic (such as CBR) in an operational setting.
  

These limitations will be addressed in a future release.

LSCle02064 LAN Packet Flooding Performance

An LS2020 node can flood between 70-100 packets per second (depending on the node and network configuration) by means of its general purpose flooding mechanism. If higher packet flooding performance is required, configure a multicast group for the flooding. With adequate network capacity the LS2020 multicast feature can support flooding at up to LAN port rates.

LS2020 Trap Caveats

LSCle02530 - No Trap for Invalid Receive Clock

The 8T3 card does not send traps to report that it has an invalid receive clock.

Workaround

None.

LSCle02638 Trap nptmm_2010 Reported When Line Card Removed or Reset

Whenever a board is removed or reset the following benign INFO trap may be reported:

(INFO) NPTMM_2010 at <local date time> (<GMT date time>) 

ERROR: Slot <n> TCS Action Register 23

Read Error (UNIX error 15: Address fault detected)

Workaround

This trap can safely be ignored.

LSCle02913 Trap Indicates Possible Card Malfunction

An LC_2000 vector 30 INFO trap like the following means that a parity error occurred on the specified line card:

(INFO) LC_card# at 8/12/95 12:24:22 EDT (8/12/95 16:24:22 GMT) Slot port#: Reason:
CP_CRASHED -- vector 30, pri 0x0, pc 0x002043fc

Workaround

If this trap appears, monitor the card carefully for additional parity errors and other signs of malfunction. Consider replacing the card if additional parity error crashes occur.

LS2020 Resource Allocation Caveats

LSCle01369, LSCle01791 Restrict MIB Variable Polling to 20 per Second or Less

The rate at which the LS2020 SNMP agent processes SNMP requests is currently limited to approximately 20 requests per second. For optimum system and SNMP monitoring performance, you should seek to limit the rate at which SNMP requests are delivered to the agent to 20 requests per second or less.

Note that SNMP requests may be originated by an external NMS (such as HP OpenView), the StreamView tool set, the LS2020 CLI, and the LS2020 Collector utility.

The NP software does not throttle excess traffic from external SNMP devices. This can consume CPU resources and can cause spanning tree timeouts and other timeouts, degrading system performance. When this happens, you must reduce the SNMP traffic from the external device.

Workaround

To reduce the SNMP request rate, you might reduce the number of variables being tracked, increase the polling interval, reduce the number of concurrent network management processes that are requesting data, or delete unused collections defined and running on the switch.

LSCle01701 PVC Setup can Deadlock when Resources Do Not Allow Creation of All Configured PVCs

In Release 2.1, PVCs are implemented as two unidirectional circuits, rather than as a single bi-directional circuit. The LS2020 at each end of the circuit establishes the transmit circuit for the PVC from its end.

In a situation where there is insufficient bandwidth between two nodes it is possible for several PVCs to get stuck in a half-open state. For example, suppose there is sufficient bandwidth between LS2020 A and LS2020 B to support one PVC (PVC 1 or PVC 2, but not both). It could happen that PVC 1 between A and B has its A-to-B circuit established, but not its B-to-A circuit, and that PVC 2 has its B-to-A circuit established, but not its A-to-B circuit. PVC 1 and 2 will stay in this state indefinitely because there is insufficient bandwidth between A and B to support the additional unidirectional circuits required to fully establish PVC 1 and PVC 2.

This does not occur in a network with sufficient capacity to support the PVCs configured for it. However, it could become a problem if trunks fail, so that existing PVCs need to be re-routed, and there is insufficient trunk bandwidth in the trunk-reduced network to support all of the PVCs.

Workaround

To recover from this state to the point that the trunk-reduced network supports the PVCs capacity permits (as opposed to far fewer, due bandwidth wasted by half-open circuits), do the following: make a priority list of PVCs, and temporarily disable low-priority PVCs. After the high-priority PVCs have been re-established, re-enable the low priority PVCs. When the failed trunks are restored and lost network capacity is recovered, the LS2020s will reestablish the remaining low-priority PVCs.

LSCle01936 Number of Circuits Supported on the LS2020

The number of circuits that can be created on an LS2020 is limited by the amount of memory on the NP.

Workaround

Users should take extreme care when over-riding the default limit of the number of VCs allowed on each card. The total number of VCs should not exceed a chassis-wide total of 4000.

LSCle02399 - Cell Packing Factor and LAN Traffic Profile Bandwidth Settings

The bandwidth parameters of a traffic profile are for managing network bandwidth. When a LAN frame arrives at an LS2020 port, the software appends an 8-byte AAL5 trailer and segments the result into 48-byte cells for transmission into the ATM network. The last cell is usually partly empty (unless the result of adding the AAL5 trailer to the frame is a multiple of 48 bytes long). For example, a 64-byte Ethernet frame is segmented into two cells, with only 16 data bytes (and the 8-byte trailer) in the second cell.

Traffic Profile rate enforcement applies no factor to account for cell packing when determining the number of cells worth of bandwidth to allow on the network. Therefore, the rate is divided by 384 bits per cell and that number of cells are allowed on the network. All other LAN flows are allowed a 20% overage to cover the inefficiencies of packing frames into cells.

The following example shows how to take into account segmentation overhead---the empty portion of the last cell---when you determine the circuit bandwidth required to support the desired throughput:

Suppose an application generates 64-byte Ethernet frames, and you want to allocate enough network resource to support 10,000 packets per second. (A 10Mb Ethernet connection can support 14,880 64-byte packets per second.) The network bandwidth required to support 10,000 packets per second is calculated as follows:

(desired packets per second) * (cells per packet) * (bytes per cell) * (bits per byte)
= 10000 * 2 * 48 * 8
= 7.68 Mbs

Compare this with the bandwidth required to support 10,000 packets per second if there were no segmentation overhead:

(desired packets per second) * (packet size in bytes) * (bits per byte)
= 10000 * 64* 8
= 5.12 Mbs

The difference is due to what is termed a cell packing factor, in this case a factor of 1.5.

• With no traffic profile specified, the software uses a cell packing factor of 1.2.
  

• With a specified traffic profile, the software uses a cell packing factor of 1.0.
  

This means that you must take segmentation overhead into account when specifying the bandwidth parameters of a traffic profile.

Before establishing a circuit, the connection management software checks the available bandwidth of each segment of the circuit, including the entry port and the exit port. In order to take segmentation overhead into account, the software calculates the capacity of a LAN port as follows:

(LAN capacity) * 1.2

Due to rounding in the conversions between cells per second and bits per second, the maximum bandwidth request that can succeed is 11,999,999 bps for Ethernet (rather than 12 Mbs), and 119,999,999 bps for FDDI (rather than 120Mbs).

LSCle02757 No More than 512 Traffic Filters Supported

If you attempt to configure more than 512 traffic filters on a given switch, the 513th filter is not configured, the switch's SNMP agent (MMA) ceases to respond to SNMP requests, and subsequent attempts to boot the switch may fail.

Workaround

Do not configure more than 512 traffic filters per switch.

LSCle02864 LCC Process May Exit if there is Insufficient Memory

An LCC process requires enough memory for the circuits configured for its line card. If there is insufficient memory for an LCC process, it exits during its startup sequence.

When this occurs the following sequence of traps may appear in the MMA trap log (in /usr/tmp/mma/mma.traplog) repeatedly:

(INFO) NPTMM_2020 at 08/03/95 20:02:43 EDT (08/04/95 00:02:43 GMT)
Slot 5 State Changed From DOWN To UP
(OPER) NDD_7 at 08/03/95 20:02:45 EDT (08/04/95 00:02:45 GMT)
Line Card Control Process for lstb5:5 exited unexpectedly (status -1).
(OPER) NDD_5 at 08/03/95 20:02:46 EDT (08/04/95 00:02:46 GMT)
Line Card lstb5:5 (8T3E3-EDGE) down (ERMP failure 0x4).

(INFO) NPTMM_2020 at 08/03/95 20:02:47 EDT (08/04/95 00:02:47 GMT)
Slot 5 State Changed From UP To DOWN

Concurrently, the following sequence of lines appears repeatedly in /usr/tmp/apps.log:

PROGRAM: lcc: (ls_main) compiled Jul 19 1995 @ 02:53:47 [pid:71]
LCC [PID 71  EIA lstb5:5]|

TRAP (FATAL) 4005: PID.71:
LCC slot 5 error 60 initializing LCC-ATM interface

ERROR [ndd.55]: Line Card Control Process for lstb5:5

exited unexpectedly (status -1).

Workaround

Make sure that there are not too many circuits configured for the chassis, and that the sum of the cardMaxVCs parameters for the cards is not too large. The chassis maximum is 4000 circuits. Reduce the number of concurrent instances of the CLI running on the chassis. The first CLI consumes about 1.5 MB and each additional CLI consumes an additional 0.77 MB.

LSCle02860 Limitation of Size of Multicast Groups

For R2.1, the size of a multicast group is limited as follows.

The number of ports plus the number of trunk links required to build a tree for a point to multipoint circuit for a multicast group must be less than 100. That is:

#ports + # trunk links < 100.

When this limit is exceeded, messages appear of the following form:

GIDD.58: ERROR (gid_pmap.c line 1979): Link_Add: Link

          table(100) too small for path(100).

        Failure of ir_generate_path for <chassis_id:slot>:

          IR error 10: Insufficient memory

(INFO) ATMM_2090 PID=71 (<chassis_id:slot>)

Workaround

If you build a circuit with a large number of endpoints, make sure that the sum of endpoints plus trunks in the multicast group topology is fewer than 100.

LS2020 Installation, Upgrade, and Initial Configuration Caveats

LSCle00645 The ckswinstall Utility Can Give False Errors

When the ckswinstall utility is applied to update distributions, it may give false error messages.

• The ckswinstall utility reports false permission mismatches on directories under /usr/app.
  

• When using ckswinstall to verify an update distribution (for instance 2.0.8), ckswinstall only verifies the software in the update. On a system running an update, most of the software in use is from the original release, so ckswinstall should also be run on that original release (for example, 2.0.7). When running ckswinstall on the underlying release, false errors are reported against software that was upgraded as part of the update.
  

Workaround

Ignore these messages.

LSCle00710 Procedure for Checking Software on the Standby NP

When verifying a software installation using the ckswinstall utility on a redundant-NP system, the software installation on each of the two NPs must be checked explicitly.

Step 1 Run ckswinstall on the current primary NP.

Step 2 Run ckswinstall on the current backup NP.

LSnode:2# rsh other-np ckswinstall ...

LSCle01015 Fallback from swchgver Does Not Copy Configuration Files Back

LS2020 configuration information is stored on disk as part of a specific release. When a new release is installed, configuration information from an old release is copied forward to the new release as part of the installation process. Configuration information, however, is not automatically copied between releases when swchgver is used to change the current release.

For instance, if a node is upgraded to run new software, configuration changes are made, and then the node is downgraded to a previous release, the node is running with the configuration information that was cached at the time of the upgrade, which does not include the subsequent changes.

Workaround

After falling back to an old version, download the current configuration to the node from the NMS.

Alternatively, before falling back to an old version, copy the files in /usr/app/base-newrel/config to /usr/app/base-oldrel/config before the fallback (replace newrel and oldrel with the appropriate release numbers, such as 2.1.0 and 2.0.8).

LSCle02646 swinstall Needs Free Memory

The swinstall program can fail due to lack of free memory if extra processes are running.

Workaround

Make sure processes which are not needed during the installation, such as CLI, are not running.

LSCle02863 The setsnmpconfig Script May Fail if Configuration Database is Locked

When a switch is booted, if it detects that the minimum required configuration information is missing, it runs the setsnmpconfig script and prompts you for configuration information. If you supply the minimum node configuration information, and later discover that this information is not in the configuration database, it may be because the configuration database lock was set when setsnmpconfig was started.

Workaround

While logged in as root, use the following commands to delete the configuration database lock and run setsnmpconfig manually:

LSNode:1# rm /usr/app/base/config/configure.netdb.lock

LSNode:1# setsnmpconfig

LSCle02874 Running the trunkconfig.cli CLI Script

The node runs the setsnmpconfig script at boot time to collect configuration information if the node detects that the minimum required configuration information is missing. If you configure trunks it creates a CLI script named trunkconfig.cli and instructs you to have the CLI run it to complete configuration of the trunks. However, before you run the trunkconfig.cli script, you must wait for all the trunk cards which you have configured to come up.

Workaround

Use the CLI show card command to determine when each trunk card is up.

LSCle02869 Cannot Execute swinstall After it Fails Once Due to Disk Read Error

The swinstall utility fails if the /mnt mount is in use. This may come about because of a disk read error in a previous invocation of swinstall. For example, due to a media error.

Workaround

Unmount the floppy drive with the command umount /mnt, and then run swinstall again.

LSCle02871 "Executable file in use" Failure of the swchgver Utility on Dual NP Systems

On dual-NP systems the swchgver utility may fail with the following error message:

bash# swchgver

swchgver: checking backup NP /dev/sd0b disk space for

<Release name> (/usr/app)

/bin/rsh: Executable file in use

swchgver: Error: attempt to contact other-np failed.

bash# 

This failure (reporting that the executable file is in use) is due to interference between swchgver and the mechanism that keeps critical files on the two NPs in synch. It is of low probability.

Workaround

Wait a moment and then invoke swchgver a second time.

Other LS2020 Software Caveats

Loss of Carriage Return or Echo

A terminal may lose carriage return or echo functionality.

Workaround

Execute the bash tset command to restore the lost property. Do this by entering the tset command at the bash prompt and pressing Return. Note that when you enter the command, the text does not appear on the screen.

Changing Primary Switch Causes Chassis to Reboot

On systems with at least one release 1 switch card (SC1), changing which switch is primary with the CLI command set chassis primaryswitch causes the chassis to reboot.

Workaround

If you change which switch is primary, wait for the chassis to reboot.

Out of Range Bit Rates Permitted on Low-Speed Card

With the set port c.p dte-bitrate command, values as high as 6,000,000 bps are available, but values over 3,840,000 bps are not supported on low-speed cards.

Workaround

Do not set rates higher than 3.84 Mbits/sec per port on low-speed cards. With the set port c.p dce-bitrate command, the values 4000 and 5376 (Kbits) are available but not supported. Rates higher than 3.84 Mbits/sec per port may work for large packets.

Internal Looping of a Frame Relay User-Network Interface (UNI) Port

Looping of a Frame Relay UNI interface provides no useful diagnostic information because the UNI protocol is asymmetric.

Workaround

LS2020 supports internal looping of such a port by first converting it to an NNI interface. A successful loop sets the administrative state of the port to testing and the operational state to up.

Looping of FDDI and Ethernet Ports Not Supported

In release 2.1.0, FDDI and Ethernet ports cannot be looped.

Workaround

Do not attempt to loop FDDI and Ethernet ports.

No Error Checking with setsnmp

The CLI setsnmp command allows you to directly modify variables in the MIB. However, this command does not validate its arguments and does not prevent you from setting a MIB variable to an inappropriate value.

Workaround

The preferred approach is to use high-level CLI set commands to change MIB variables and avoid setsnmp unless specifically instructed otherwise. These commands give the CLI enough context to validate the new value for the variable before changing the MIB.

LIGle00038 Defective Floppy May Cause tar Process to Hang

When attempting a tar operation to the floppy drive (for example tar xvf /dev/sd1) when the floppy disk has media errors, the software process using the device may hang, preventing future access to the floppy or to the terminal where the tar command was being run.

Workaround

If the process hangs, reboot the network processor.

LSCle00404 FDDI PATH Configuration Table is Not Supported

The RFC 1512 fddimibPATHConfigTable MIB object is not implemented.

LSCle00711 Periodic rdist to the Secondary NP Does not Check for Available Disk Space

On a redundant NP system, an automatic mechanism updates software and configuration information from the current primary NP to the current secondary NP. If the current secondary NP runs out of disk space, the secondary NP is not kept consistent with the primary NP. Furthermore, no error is reported.

Workaround

Establish a regular maintenance procedure of checking disk space availability on both NPs.

LSCle00720 FDDI PathTestFeature is Not Supported

The command set port c.p fddi smt station path test has no effect.

LSCle00961 Cannot Connect to Backup NP in Diagnostic Mode

You cannot load NP diagnostics and then connect to the backup NP.

Workaround

Use the CLI test command to run network processor (NP) diagnostics. The test command may be used on any card except the active NP. (The command test -m is not supported on the backup NP.)

LSCle01044 STP - External Bridge Loops and LS2020 VLI Functionality

The presence of external bridge loops may result in loss of LS2020 VLI functionality.

Workaround

When configuring LS2020 switches with external topology loops, configure the bridge priority for the switches such that one of them will become the spanning tree root.

LSCle01096 An ATM UNI or Frame Forwarding Port Reports Operationally Up When Disconnected

Currently the frame forwarding and ATM-UNI interfaces use no protocol to provide a link-level reliability indication such as the LMI for FR or Trunk-Up-Down (TUD) for LS2020 trunks. The Operational Status on these ports reflects the Administrative Status of the port. The operational status does not indicate physical layer or data link layer status.

Workaround

None.

LSCle01318 OC-3 Card Continuously Bounced When Connected to SONET MUX

When an OC-3c port was connected to a SONET multiplexor with internal clocking the connection continuously bounced.

Workaround

When connecting an OC-3c port configured for internal clocking to a SONET multiplexor, a clocking type of external for the LS2020 port should normally be used (rather than the default of internal). Use of internal clocking may prevent the port from operating correctly.

LSCle01560 Console Messages Regarding NP Ethernet Port

When the node NP OS kernel detects errors on the NP's Ethernet port, it may output one of the following messages:

oblan: ethernet transmit error (status=0xnumber): string

oblan: receive error (0xnumber): string

In the transmit error message, if string is "lost carrier", the Ethernet cable may be disconnected from the NP's Ethernet port; if it is "late collision", there is a network problem, such as noise, poor connectors, or a broken station. In the receive error message, the string "missed packet" may occur under extreme load. The number in these messages represents hardware information which may be useful to Cisco Support.

These errors appear on the console only if they occur before starting the ATM switching software on the NP (for example, while the system is being installed or during system startup). After the ATM switching software has started, these errors go only to the traplog, where they appear as KERN_2001 traps.

Workaround

Verify connectivity to the NP's Ethernet port.

LSCle01874 Long Term PVC Statistics Lost when PVC Re-established

Whenever it is necessary to re-establish a PVC (either because of re-routing due to network service disruptions, such as a trunk failing, or because of changes to PVC parameters by an operator) the statistics for that PVC are cleared and all previous statistics for that PVC are lost.

Workaround

If maintaining long-term counts is important, poll these variables and save them externally to reduce the effect that the loss of counter values has on your data.

LSCle02174 - Loopback - ATM-UNI Status Wrongly Reported Up

When set to remote loopback, the correct operational status for an ATM UNI port should be DOWN, but the reported status is UP. This is because the operational status of the ATM UNI interface does not take into account the physical layer indications.

LSCle02211 DSX1 Statistics Tables Not Implemented

The current, interval, and total tables of the DSX1 MIB (RFC 1406) are not supported for CEMAC cards in Release 2.1.

LSCle02228 - Old cardMaxVCs Value is Retained After Card Swap

The cardMaxVCs parameter applies to the card slot. As a result, when a card of one type replaces a card of another type, the value of cardMaxVCs set for the first card may be inappropriate for its replacement.

Workaround

When you replace a card with a card of another type, verify that the value set in cardMaxVCs is appropriate for the new card type, and change it if necessary. From the CFG tool, delete the original card type and add the new type. Use the Sendupdate Changes only operation to send the new configuration to the chassis.

LSCle02656 IP Filtering Not Supported for 802.3 (SNAP) Encapsulated IP Frames

Custom filters do not work for 802.3 encapsulated IP frames.

Workaround

Do not attempt to filter 802.3 (SNAP) encapsulated IP frames.

LSCle02709 dot1dStpPortEnable Not Set When LAN Port is Down

When a LAN port is operationally down (unplugged), attempts to set the dot1dStpPortEnable object for the port appear to succeed but in fact have no affect.

Workaround

Do not attempt to set the dot1dStpPortEnable object when a LAN port is operationally down.

Software Upgrade Procedures

This section provides information about upgrading the LS2020 enterprise ATM switch to release 2.1.01 of LS2020 software.

Requirements for Upgrade

For release 2.1.01 to be installed, the network processor must have 32 Mb of memory.

For release 2.1.01 to be installed successfully, you must currently be running base Release 2.1 node software.

You must have a modem in each node in which the software upgrade is to be made remotely. Please refer to the section entitled "Modem Recommendations" in the LightStream 2020 Site Planning and Cabling Guide. Caution The LS2020 NP is a special-purpose communications processor. It should not be used as a general-purpose UNIX host. If any files have been copied or placed on the disk (especially in the root partition), they should be removed before upgrading to release 2.1.01. If the names of any Cisco-provided files have been changed, the original file names should be restored.

Distribution Media

Below is a list of the LS2020 release 2.1.01 node software distribution diskettes.

LS2020 Release 2.1.01 Distribution Diskettes	Version Listed on Diskette Label
Boot Disk	2.1.01
System Disk 1	2.1.01
System Disk 2	2.1.01
System Disk 3	2.1.01
Application Disk 1	2.1.01
Application Disk 2	2.1.01
Application Disk 3	2.1.01
Application Disk 4	2.1.01
Application Disk 5	2.1.01
Application Disk 6	2.1.01
Application Disk 7	2.1.01
Diagnostic Disk 1	2.1.01
Diagnostic Disk 2	2.1.01
Firmware Disk 1	2.1.01
Firmware Disk 2	2.1.01

Outline of Upgrade Procedures

The following procedures are used to upgrade a network to Release 2.1.01 software:

• Procedure 1, Copy New Software to the Distribution Node
  

Use the swinstall command to copy the new software from the installation diskettes to the node being used as the software distribution node. (If swinstall reports insufficient disk space, you will be directed to perform Special Procedure A, Freeing Up Disk Space.)

• Procedure 2, Copy New Software to Remote Nodes
  

Use the swremoteinstall command to copy the new software from the distribution node to other nodes in the LS2020 network. (If swremoteinstall reports not enough disk space, you will be directed to perform Special Procedure A, Freeing Up Disk Space.)

• Procedure 3, Change the Running Software Version
  

Use the swchgver command to change the version of software running on each node after copying the new software to the node.

Outline of Special Procedures

These procedures are not normally needed. It is possible that one of the upgrade steps might send you to perform one or more of these procedures.

• Special Procedure A, Freeing Up Disk Space
  

If the swinstall command or the swremoteinstall command reports that you do not have enough disk space, delete files for obsolete releases of software.

• Special Procedure B, Falling Back to the Prior Version
  

You have the option of falling back to the prior version of software if you wish to for any reason.

• Special Procedure C, Verifying Connection to Backup NP
  

If the swchgver command fails to contact the backup NP on a redundant system, this section describes some of the common causes and how to fix them.

• Special Procedure D, Getting rsh to Work Successfully on a Remote Node
  

If the swremoteinstall command (which uses the rsh command) reports that it does not have permission to copy the files to the remote node, this section describes some of the common causes and how to correct them.

• Special Procedure E, Backing Up the Distribution Diskettes
  

Some users may wish to back up the distribution diskettes.

Procedure 1, Copy New Software to the Distribution Node

With this procedure you will copy the new software to a local LS2020 node from the distribution disk set. This local node will be referred to as the distribution node. In Procedure 2, Copy New Software to Remote Nodes, you will use the distribution node as the source from which you will copy the software to other nodes in the network.

bash# swchgver

Checking and downloading FLASH memory for all function cards

Checking and downloading files for standby network processor.

Rebooting standby network processor.

....

Forcing reset of line cards



==> (OPER) NDD_5 at 10/20/94 18:11:24 EDT (10/20/94 22:11:24 GMT)

      Line Card lsnode8:3 (LS-EDGE) down (ERMP failure 0x401).



==> (GENERIC) at 10/20/94 18:11:28 EDT (10/20/94 22:11:28 GMT)

Link Down Trap at 10/20/94 18:11:28 EDT (10/20/94 22:11:28 GMT)                

      Port 3002



==> (GENERIC) at 10/20/94 18:11:28 EDT (10/20/94 22:11:28 GMT)

Link Down Trap at 10/20/94 18:11:28 EDT (10/20/94 22:11:28 GMT)                

      Port 3003        



==> (GENERIC) at 10/20/94 18:11:28 EDT (10/20/94 22:11:28 GMT)

Link Down Trap at 10/20/94 18:11:28 EDT (10/20/94 22:11:28 GMT)                

      Port 3004

==> (GENERIC) at 10/20/94 18:11:28 EDT (10/20/94 22:11:28 GMT)

Link Down Trap at 10/20/94 18:11:28 EDT (10/20/94 22:11:28 GMT)                

      Port 3005



Rebooting the network processor



NP040 POST Version 0.225 Feb 21, 1995

4Meg Bit value = 1

Configuring Main Memory for 32 Megabytes

Clearing memory (32 Megabytes)...



booting: drive:0, partition:0, kernel:"lynx.os", flags:0x4308

Resetting SCSI bus

Kernel linked for 0xea010000

LOAD AT 0x10000

        483328+49152+262504[+62736+51815]

TOTAL SIZE: 909536 at 0x1001c

START AT 0x10020

NP memory size: 32 MB

ILACC: EEPROM enet addr:8:0:8:0:14:6f, Silicon Rev:0x5, IB:0xea1dfce0

Old-style NP detected

virtual console: IB: 0xea1dfe68

NCR 53C710: Chip Revision: 0x2, IB: 0xec18e000



LynxOS/68040-MVME167 Version 2.1.0

Copyright 1992 Lynx Real-Time Systems Inc.

All rights reserved.



LynxOS release 2.1.0, level 1: NP-LynxOS #107: compiled Apr 17 1995 14:50:57



LynxOS Startup: ma

fsck /dev/sd0a

(all sizes and block numbers in decimal)

(file system creation time is Mon Apr 11 08:57:19 1994)

checking used files

recovering orphaned files

making free block list

making free inode list

40518 free blocks 3314 free inodes

fsck /dev/sd0b

(all sizes and block numbers in decimal)

(file system creation time is Mon Apr 11 08:57:52 1994)

checking used files

recovering orphaned files

making free block list

making free inode list

17838 free blocks 3426 free inodes



fsck /dev/sd0c

(all sizes and block numbers in decimal)

(file system creation time is Mon Apr 11 08:58:25 1994)

checking used files

recovering orphaned files

making free block list

making free inode list

8645 free blocks 3534 free inodes



fsck /dev/sd0d

(all sizes and block numbers in decimal)

(file system creation time is Mon Apr 11 08:58:58 1994)

checking used files

recovering orphaned files

making free block list

making free inode list

26597 free blocks 3601 free inodes



Mounting all filesystems

Starting VM system ... Virtual Memory Engaged!

inetd started

Starting crond ...

Initializing the switch hardware interface ...

Using switch A, cards are NOT synchronized, fast cutover is supported

PCP version: 0x410, CMP version: 0x12, FSU version 0x109

Starting the switch software

LightStream 2020 Version 2.1.01

Copyright 1993 LightStream Corp.  All rights reserved.

Portions copyright 1992 by Lynx Real-Time Systems Inc., 1983 by the Regents

of the University of California, 1988 and 1990 by Paul Vixie, and 1991 by

SNMP Research Inc.



This software contains unpublished proprietary and trade secret information

of LightStream Corp.



LightStream 2020 Software provided to the U.S. Government is subject to the

notices on the software and on the LightStream user documentation copyright

page.

PROGRAM: cbuf: (ls2_0) compiled Apr 26 1995 @ 21:49:18 [pid:48]



user name:

bash# swdelete 2.0.7

Deleting version 2.0.7

bash#