Diagnostic Flowcharts

Check dmesg/syslog output and look for error messages pertaining to gelan/igelan. Also, check nettl log messages. If errors, check card installation and reset and/or reseat card.

Verify LEDs. If 1000Base-SX link LED = OFF or for gelan, 1000Base-T all speed LEDs = ON, check card installation and reset and/or reseat card. If LEDs are then displayed correctly, continue to Link Level Test.

If 1000Base-T all speed LEDs = OFF or for gelan, 1000Base-SX Link LED = Flashing, check for incorrect or faulty network cable or connector. Ensure that switch is capable of 1000 Mbps operation. Ensure that the switch (or immediate link partner) and card are set to same autonegotiation settings. Then go to Configuration Tests. Otherwise, if 1000Base-SX link LED or one of 1000Base-T speed LEDs is on, continue to Link Level Test.

Execute linkloop(1M) to remote host. If linkloop is successful, continue to Network Level Tests. Otherwise, note which error was returned.

If loopback failed error = "Address has bad format" or "Not an individual address," correct link level address with proper station address format/value and repeat Link Level Test.

Otherwise, loopback failed because remote host did not respond. Double check remote host address or choose another remote host and re-execute linkloop(1M). If linkloop is successful, continue to Network Level Tests. You may also want to call the node manager of the remote host that did not respond (if this was the case). If linkloop fails, go to Configuration Tests.

See Flowchart 3a to validate arp(1M) entries and remote host availability.

See Flowchart 3b to check communication between network layers on source and target host using ping(1M).

Execute ping(1M) to remote host so that ARP entry is added. Whether or not ping is successful, proceed to the next step.

Use arp(1M) to verify that an entry exists for the remote host in your system's ARP cache, executing arp hostname.

If there is no ARP entry for the remote host, check to see if the remote host is up. If not, bring up remote host and continue to ping Test.

If the ARP entry is correct or complete, continue to ping Test. Otherwise, use arp(1M) to enter the correct station address of the remote system and continue to ping Test.

Execute ping(1M) to remote host. If ping is successful, continue to Transport Level Test.

If ping is not successful, execute netstat -in to verify MTU size. Ensure MTU size is the same on both local and remote hosts (9000 for jumbo frames and 1500 for standard frames) by executing lanadmin -M new_mtu nmid/ppa, and repeat ping Test.

If ping is still not successful and you are either (1) not using jumbo frames or (2) using jumbo frames with the correct speed setting, continue to next flowchart to validate network, remote host, and configuration settings.

If link speed is not 1000 Mbps, set it with lanadmin -x speed nmid/ppa,and repeat ping Test.

If network unreachable error, go to Configuration Tests.

If there is no response from ping, and using jumbo frames, validate that switches in the path support jumbo frames, making sure path MTU is 9000 from source host to destination host. Otherwise, reconfigure network path and repeat ping Test. If not using jumbo frames, or switches and path MTU are set for jumbo frames (9000 bytes), go to Cable and LED Test.

If you receive an unknown host error, add missing host name and repeat ping Test.

If you receive "error=SendTo: No route to host," use route(1M) to add route table entry for missing host and repeat ping Test. Otherwise, call your HP representative.

Execute telnet(1M) to a remote host. If successful, stop.

If not successful, try to establish an ftp to a remote host. Unlike telnet, ftp does not use a pseudoterminal (pty) driver on your system. This will determine if pty is why telnet failed. If ftp is successful, call your HP representative to determine why you have a problem with pty.

If ftp fails, check to see if TCP is configured on both hosts by verifying the /etc/protocols file. Telnet and ftp work at transport layer and require TCP. If TCP is not configured, configure now and repeat Transport Level Test.

If TCP is installed on both hosts, telnet to another host and/or use netstat(1M)to check for lost packets. If network is congested, you may need to reconfigure network. If network congestion is not the cause, more detailed network diagnostics are required. In either case, call your HP representative.

Execute ping(1M) from a known good host through a gateway to another known good host. This will test connectivity through bridge/gateway level. If successful, execute netstat -r and examine route table on problem host and all hosts in path. If necessary, correct routing table and go to Network Level Tests.

If ping fails, examine gateway to see if HP 9000 or non-HP. If non-HP, refer to networking documentation for that product. If HP 9000, execute ifconfig(1M) for all interfaces on gateway or host (see Configuration Tests for more details on ifconfig).

If ifconfig does not show parameter as UP in output for gateway, execute netstat -i to check status of network interfaces. An asterisk (*) indicates interface is down. If network interface is down, configure interface up and repeat Bridge/Gateway Test. If all interfaces are up, continue to Configuration Tests and test all interfaces on gateway.

Execute ioscan(1M) as follows:

ioscan -kfd drivername, where drivername is either gelan or igelan (refer to Table 1-1 “List of Gigabit Ethernet Products”).

Verify output from ioscan shows card "CLAIMED" by the system.

If card is claimed, execute lanscan(1M) and check if the hardware state display shows "UP." If so, go to Cable and LED Test. If not, continue to netfmt and lanadmin Test.

If card is not claimed, execute what /stand/vmunix | grep drivername, where drivername is as mentioned above. Verify if the output is similar to the output documented in the Release Notes for your Gigabit Ethernet product. Use the name of the running kernel image file in place of "/stand/vmunix" as appropriate.

If driver is displayed, check if dmesg/syslog output shows error messages pertaining to gelan/igelan. Also, check nettl log messages. If errors, check card installation and reset and/or reseat card, then repeat ioscan and lanscan Test. Otherwise, call your HP representative.

If driver is not displayed, install driver using swinstall(1M) and verify or edit /stand/system to contain gelan or igelan keyword. (See Chapter 3 of Installing and Administering LAN/9000 Software for instructions on how to edit /stand/system and create a new kernel.) Reboot system and repeat ioscan and lanscan Test.

Execute netfmt(1M) and view error and disaster log messages.

Example: netfmt -vf /var/adm/nettl.LOG00

It will help to use the time stamp to find proper logs. Ensure you are looking at 1000Base-SX/T information.

If problem is solved, continue to ifconfig Test.

If problem persists, run lanadmin(1M) to reset card.

If reset is successful, go to Link Level Test. Otherwise, reset the card once more; if still not successful, call your HP representative.

Execute ifconfig(1M) on the interface you want to configure to ensure that interface is enabled. Example:

ifconfig lan1 192.6.1.17 netmask 255.255.255.0 up

Next, execute ifconfig interface to test and verify flag setting is UP and correct IP address is displayed. Example:

ifconfig lan1<UP,BROADCAST,RUNNING,MULTICAST,CKO>

If IP and flags are correct, verify there is an entry for card interface in /etc/rc.config.d/netconf. If so, go to Network Level Tests. Otherwise, add correct interface parameters to /etc/rc.config.d/netconf file and reboot. If flags are incorrect, correct them with ifconfig and repeat ifconfig Test. Otherwise, if ifconfig is not successful and error messages appear, correct them accordingly and repeat ifconfig Test.

If you cannot correct errors, call your HP representative.