ITEM: I2646L
Problems pinging other RS/6000 machines from a brand new install.
Question:
I have a brand new RS/6000 model 365 running 3.2.5. I have configured
this machine on my ethernet network. I have two other RS/6000 boxes and
two non IBM Unix machines on this ethernet network. The ethernet consists
of thick net. These machines are able to ping each other, but can not ping
my new RS/6000 model 365 machine. I am able to ping myself and I do have a
current /etc/hosts file. Why am I not able to ping the network?
ANS: There are multiple reasons that you might not be able to ping the
network. Some of the possible causes are bad cables, bad configuration,
integrated card not jumpered, incompleted /etc/hosts file, bad hardware.
To determine what is causing the problem follow the steps below:
lssrc -g tcpip -Verify that all necessary daemons are running,
inetd must be running. If the daemons need
to be started then type the command,
startsrc -g tcpip
vi /etc/hosts -Verify that this file is up to date and contains
a loopback entry.
127.0.0.1 loopback localhost \# loopback (lo0) address
smitty tcpip -Verify that the entries are correct. Note
that the cable type should match the following
table.
Adapter High Performance Totally Integrated Partially Integrated
Connectors DIX and BNC DIX only DIX, BNC or UTP
10Base5 DIX N/A N/A
10Base2 BNC N/A N/A
10BaseT DIX N/A N/A
netstat -i -This command will give you information on what
network interfaces that you have configured.
Since you are only on a ethernet network you
should have 4 entries. You should have two
entries for loopback and two entries for your
ethernet network, ie en0. This command showed
that et0 was also configured. This is not the
type of of interface that you are using. Note
that your tcpip problems could be caused by this.
The rest of your machines are configured with the
interface en0 and this machine has both interfaces
configured with the same IP address for each interface.
The en0 and et0 interfaces are not compatible since
they both have different packet formats. Remove the
ent0 configuration and restart the tcpip daemons with
the following commands:
ifconfig et0 down
ifconfig et0 detach
rmdev -l et0 -d
stopsrc -g tcpip
startsrc -g tcpip
Try to ping the machines on your network now. If you
are still unable to ping continue to the next step.
arp -a -this command will show the arp table.
The arp table is used to resolve the
software IP address to the hardware
address of the actual adapter. This
hardware address is known as Media
Access Control (MAC) address.
This command showed that the arp table
had "incomplete" entries. The table
showed the hostname and software address
but not the hardware address of the
machine that was pinged.
hostname.fulldomain.name (192.9.200.5) at (incomplete)
The arp table can be set with;
arp -s Type HostName Address
I do not think that this would help in this case
since the machine does not seem to be able to send
or receive any broadcasts on the network. This could
be caused by a loose cable, a bad adapter, or a bad
cable. Confirm that the cables are good and have good
connections. Check the error report to see if the
adapter is reporting any problems. This can be done
with the following command;
errpt
Note that since the machine is able to ping itself
the adapter is more then likely functioning properly.
The loopback is used when the machine pings itself.
If the hardware is correct and working properly then
continue to next step.
lscfg -l ent0 -d This command will give information on the
type of ethernet card that is in the
machine. If only one line of information is
is shown then the card is a partially integrated
ethernet. These type of cards have a jumper
that is set either for thick or thin.
DEVICE LCOATION DESCIRPTION
ent0 00-01 standard ethernet adapter
In the past if the jumper was not set properly an
error message would be echoed on the console to
check the jumper setting. At 3.2.5 and above
this might no longer be the case. Therefore it if
the card is jumpered to the wrong cable type then
you would not be able to send or receive broadcasts
from the network. If this is the case, the only
way to change the cable type is by moving the jumpers
on the riser card, a job that an IBM C.E. should
perform if the RISC is under warranty. There is no
change in the software configuration for this card.
Once the jumper is set correctly then reconfigure the
interface via smit.
smit tcpip
Minimum Configuration & Startup
et0
\
Try to ping. If the RISC is still not able to ping the network then check
to make sure that there is not an alternate ethernet address set.
smit commodev
Ethernet
Adapters
Change/Show
ent0
Verify that the ALTERNATE ETHERNET address option looks like:
ALTERNATE ETHERNET address [0x]
Support Line: Problems pinging other RS/6000 machines from a brand new install. ITEM: I2646L
Dated: January 1995 Category: N/A
This HTML file was generated 99/06/24~13:30:45
Comments or suggestions?
Contact us