[Wedge100BF-series] MAC not found Follow

Purpose:

This is the troubleshooting guide for the MAC not found on Wedge100BF-series.

Model:

• DCS800 - Wedge100BF-32X
• DCS801 - Wedge100BF-32QS
• DCS802 - Wedge100BF-65X

Procedure:

Step 0. Check the OpenBMC version

There's a known issue happened at a certain rate in old OpenBMC version before v01.10.00, the BMC detected temperature but got impossible abnormal value, like million℃, then cause the fand(OpenBMC process) shutdown COMe as protection mechanism.
In OpenBMC v01.10.00 we optimized the way to conversion the value of temperature, the issue is fixed after and including v01.10.00, please check and upgrade if the firmware not the latest one.

[Wedge100BF-series] Chip, OpenBMC, CPLD, EC Code, BIOS, ONIE, SPI version checking

Step 1. Scan the Tofino information

Method 1. Use "lspci" to scan Tofino information

Use the command "lspci" or "lspci | grep 1d1c" to check the chip.

This command can be used in ONIE, NOS, Diag mode.

root@bfn-switch:~# lspci | grep 1d1c
05:00.0 Unassigned class [ff00]: Device 1d1c:0010 (rev 10)

If it shows like the example (end with 0010), the switch using Chip B0.

Caution: If it shows end with 0001, the switch using Chip A0, and it's EOL we won't support.

If the command shows nothing(can't find the MAC), please keep check on Step 2.

Method 2. Login Diag mode to test

To login Diag, please refer to the article: Install and start up Accton Diag

Choose item 55 to Scan Tofino information.

*********************************************************
(...omitted)
55 Scan Tofino information
(...omitted)
*********************************************************
Please enter your choice (0 to quit): 55
05:00.0 ff00: 1d1c:0010 (rev 10)
The Chip is B0.


Item (55) test result: PASSED
Please enter your choice (0 to quit):

If the test result FAILED, please keep check on Step 2.

Step 2. Reboot the COMe(CPU) and scan Tofino again

If user execute following command and get "0x03" as below, it indicate the COMe had ever been shutdown by BMC due to actual reason like temperature too high, lots of fans failed, or known issue cause by old version OpenBMC firmware mentioned at Step 0.

root@bmc:~# i2cget -f -y 12 0x31 0x32
0x03

Reset Tofino and CPU and set it to normal

root@bmc:~# i2cset -f -y 12 0x31 0x32 0x00
root@bmc:~# i2cset -f -y 12 0x31 0x32 0x0f
root@bmc:~# i2cget -f -y 12 0x31 0x32
0x0f
root@bmc:~#

Then power off and power on the COMe

root@bmc:~# wedge_power.sh off
board type is Montara
Power off microserver ... Done
root@bmc:~# wedge_power.sh on;sol.sh
board type is Montara
Power on microserver ... Done
You are in SOL session.
Use ctrl-x to quit.
-----------------------

Then scan the Tofino information again as Step 1.

[Note] This register's value won't be recover to "0x0f" after reboot OpenBMC. Considering some users may not convenient to power cycle the whole switch, this procedure can make sure to reset Tofino and CPU and instead of power cycle.

Step 3. Check SPI flash and upgrade.

Article : SPI Flash Firmware version checking and upgrade

If the SPI flash has already upgraded to latest version and power cycle several times but the "lspci | grep 1d1c" still shows nothing, please contact us for RMA.