Troubleshooting

The following table describes various symptoms and corrective actions for the HP EISA or HSC FDDI adapter.

Table 5-1 Symptoms and Actions

Symptom	Action
Card fails power-up test.	Inspect the LEDs on the back of the adapter. A constant, solid LED (4) indicates a failure in the power-up diagnostics. For HSC FDDI, solid LEDs (2) and (3) indicate power-up diagnostics succeeded. For EISA FDDI, solid LEDs (1) and (3) indicate power-up diagnostics succeeded. If power-up diagnostics fail, turn the power off, reseat the adapter, then turn the power back on. Try another EISA or HSC slot. Swap adapter with another which is known to operate.
Adapter does not respond.	Use the `ifconfig` command to check the existence and status of the device. For example: `ifconfig lan0` Verify that the host to IP mapping is known to the local system by either doing an `nslookup` on the host name or by checking the `/etc/hosts` file. Verify that the FDDI cable is properly connected at both ends.
`ioscan` output shows that the adapter is unclaimed.	Check that the driver software is loaded.
Adapter cannot communicate with other hosts on the local network.	Use `fddilink` to verify that upstream and downstream neighbors are correct. Examine the LEDs. Check the FDDI cable. Make sure the FDDI media is correctly installed. `ping` the failed system from another host on the network. Distinguish between an unknown host, which indicates a `/etc/hosts` file problem; and a non-response, which usually indicates a routing problem. Check the `arp` table with `arp -a`.
(Dual-ring only) Adapter can only communicate with a subset of hosts on the local network.	See above. Check for "twisted ring" (port A cabled to port A or port B cabled to port B). Visually inspect the cabling. Or, check the syslog file (default is `/var/adm/syslog/syslog.log`) for a message like the following: Undesired Connection Attempt: A ->A. Check for link failures. Run `fddilink` on local and remote systems and check for "WRAP_A" or "WRAP_B" CFM states.
Cannot reach a host on a remote network.	Use `ping` to test connectivity to stations on your local ring. Distinguish between an unknown host, which indicates a `/etc/hosts` file problem; and a non-response, which usually indicates a routing problem. Use `fddilink` to see if you are communicating with your upstream and downstream neighbors on the ring. Check the `arp` table with `arp -a`. Use `netstat -r` to check routing tables. Refer to the man page for expected output.
Cannot connect to ring.	For dual attach stations, check the cables for PHY A and PHY B to verify appropriate connections to neighbor stations.
Ring state is unstable.	Use `fddilink` to check for abnormal statistics. If the attribute LER is near the Alarm value, check for poor connections. Verify optical power loss does not exceed 11 decibels between transmitter and receiver pairs.

LED Diagnostics

The HP EISA and HSC FDDI adapters have built-in self-test diagnostics that are executed when the system probes the bus adapter/slot. This occurs at power-up or reset of the card. These self-test diagnostics test both the card's hardware and firmware and the card's functionality. Observing the adapter's LEDs will indicate at what stage the card is functioning. These LEDs exhibit various illumination, dependent upon the state of the card.

The LEDs are located on the faceplate. Figure 5-1 “LED Configuration (Port B)” shows the configuration of the LEDs for port B for both EISA and HSC FDDI. Table 5-2 “HP EISA FDDI Adapter LED States (Port B)” defines HP EISA FDDI card states. Table 5-3 “HP HSC FDDI Adapter LED States (Port B)” defines HP HSC FDDI card states.

Figure 5-1 LED Configuration (Port B)

Table 5-2 HP EISA FDDI Adapter LED States (Port B)

LEDs by Colors				Card States
Yellow (1)	Yellow (2)	Green (3)	Green (4)	Card States
Solid	Off	Off	Off	Power on
Solid	Solid	Off	Off	Power on diagnostics
Solid	Off	Solid	Off	Passed power-on diagnostics
Solid	Off	Solid	Blink	CB running
Solid	Off	Blink	Off	RC running
Solid	Off	Blink	Solid	Ring connect
Solid	Solid	Off	Off	Failure

Table 5-3 HP HSC FDDI Adapter LED States (Port B)

LEDs by Colors				Card States
Green (1)	Green (2)	Yellow (3)	Yellow (4)	Card States
*	Solid	Solid	Off	Hardware power-on diagnostics completed; driver not yet initialized
*	Blink	Off	Off	Connecting to ring (local cables may not be connected or cables may not be connected at remote end)
Solid	Blink	Solid	Off	Ring connected (normal function)
*	Off	Off	Solid	Failure

* indicates that this LED will be Solid if a link is detected, otherwise it will be Off. The state of this LED should be ignored if the card is in the Failure state.




	NOTE: Port A of the HP HSC FDDI adapter has only one LED. This LED is Solid if a link is detected, otherwise it is Off.

Performance Troubleshooting

This section is intended to provide system administrators or advanced users with detailed information on how to troubleshoot performance-related problems with the HP EISA FDDI/9000 or HSC FDDI/9000 products. A few key terms are defined below to help in understanding the troubleshooting information:

Key Terms

Transmit Threshold: This value determines how many bytes must be in the FDDI transmit FIFO before transmission of the bits onto the FDDI cable will begin.

Transmit Underrun: A transmit underrun error occurs when the FDDI transmitter encounters an empty transmit FIFO during the transmission of bits onto the FDDI cable.

Memory Subsystem Latency: The time it takes to move data from system memory to an I/O device. This time includes the arbitration delay for the I/O device and for each bus bridge between the system memory controller and the I/O device.

Arbitration Delay: The time it takes an I/O device or bus bridge to acquire the I/O bus for data transfer.

Transmit FIFO: A buffer on the FDDI card used to hold data transferred from system memory to the FDDI card.

The EISA FDDI/9000 and HSC FDDI/9000 products are currently optimized to achieve the best single card performance. While the current Transmit Threshold value allows the FDDI products to achieve their best performance, it also increases the probability of Transmit Underrun errors. A large number of Transmit Underrun errors (more than 1 out of every 10,000 packets) can cause a noticeable drop in networking performance.

Transmit Underrun errors may occur in some HP systems that have high memory subsystem latencies combined with sufficient bus contention from competing I/O devices. These errors can be monitored in two ways:

Examine the output from netstat -I interface
Examine the output from fddiif interface

If the number of output errors is high (more than 1 out of every 10,000 packets), then the system is most likely suffering from transmit underruns on the specified network interface and corrective action must be taken to resolve the problem.

EISA FDDI/9000

For the EISA FDDI/9000 product, use the lanadmin -S command to toggle the use of onboard memory for transmit on or off. (By default, use of onboard memory for transmit is off.) For example:

lanadmin -S 1 PPA

Note that the value specified after the -S parameter is not important, as the command is simply a toggle. After issuing the above command, the current speed of the interface is echoed; this output may be ignored. You must then wait at least 5 seconds before attempting to use the specified network interface.

HSC FDDI/9000

In order to achieve the best single card performance, the HSC FDDI/9000 product has set the Transmit Threshold to an aggressive value. The Transmit Threshold is set so that transmission will begin after 1024 bytes are in the transmit FIFO.

The HSC FDDI product supports four levels of Transmit Threshold. These four levels are modified via the -S option of the lanadmin command, as follows:

lanadmin -S TransmitThreshold PPA

where TransmitThreshold can be one of the following values:

1024 is most aggressive (this is the default value)
2048 is somewhat aggressive
3072 is somewhat conservative
4096 is conservative

In all of the cases above, after setting the TransmitThreshold mode as specified, the lanadmin command will echo the current speed of the interface; this output may be ignored. The following example shows output for 100 Mbits/s operation:

old speed= 100000000 new speed= 100000000

After issuing lanadmin -S, you must wait at least 5 seconds before attempting to use the specified network interface.

HSC FDDI transfers data from memory on the card to memory on the host via Direct Memory Access (DMA) operations. By default, the data transfer burst size for DMA operations is 128 bytes. If the HSC FDDI card is installed in an HSC bus that is shared with other HSC devices (for example, an HSC SCSI card and other network cards), the bus competition among those devices may be increased. If setting the Transmit Threshold value with the lanadmin command results in increased "receive overrun" errors, it may be because of the HSC bus competition with other devices. To reduce the receive ovverun errors, change the data transfer burst size from 128 bytes to 32 bytes with the special command lanadmin -S 0. Note that if you issue the lanadmin -S 0 command, any subsequent setting of the Transmit Threshold value with the lanadmin command restores the data transfer burst size to 128 bytes.

Troubleshooting

Technical documentation

» Table of Contents

» Glossary

» Index

LED Diagnostics

Performance Troubleshooting

Key Terms

EISA FDDI/9000

HSC FDDI/9000