Quadrics QsNetII Interconnect

D
Output From the qselantest Utility
Using the qselantest utility is described in Section 12.3. The output from the
test is extensive and is broken down for description as follows:
• Device information (dev_info) is described in Section D.1.
• PCI Bus information is described in Section D.2.
• Thread processor information is described in Section D.3.
• SDRAM memory information is described in Section D.4
The qselantest can fail if a faulty component is detected or if the link cable
to the interconnect network is badly seated or missing. If the failing test is one
of the qsnet2_dmatest tests, the most likely cause is the link cable connection
to the network. Remove, inspect, and reseat the cable at both ends. Make sure
that the locking catch is fully seated. If a qsnet2_dmatest run continues to fail,
replace the link cable.
If a qsnet2_regtest test fails, it is most likely due to the QM500 (Elan4) network
device driver not being loaded, or a badly seated or broken QM500 PCI adaptor.
If the qsnet2_tmemtest fails, there is a problem in the QM500 PCI cards
SDRAM memory and you must replace the card. The QM500 network adapter has
64 Mbytes of local SDRAM. Its memory system can detect and correct a variety of
single and multiple bit errors. These errors are logged by the device driver which
also outputs errors on the console in the following format:
elan0: ECC memory error
Address=001e71d0 Syndrome=13 Correctable
elan0: ECC memory error
Address=001e71c0 Syndrome=13 Correctable Multiple Errors
An uncorrectable error causes the node to panic. The multiple errors message is
displayed if another correctable error occurs before the device driver handles a
pending error. A low rate of correctable errors is to be expected; approximately 8
per day on a system with 1000 QM500 network adapters is usual. Memory errors
will be logged in the console files for each node. They can also be seen in the output
from the dmesg command. Use the elandebug command to display a summary of
device error statistics as follows:
#
/usr/opt/rms/diag/bin/elandebug device_error_stats
...
CorrectableErrors 0(0)
UncorrectableErrors 0(0)
MultipleErrors 0(0)
If you are escalating a problem report to a support organization, use the following
command to obtain information about the card:
# cat /proc/qsnet/elan4/device0/vpd
PN: QM500b
EC: XX
SN: K76A7A1BFPN471
FN: A1
MN: 0000
Z0: 686a232b
MT: 0104-023-112007
Output From the qselantest Utility D-1