HI all, I have a Gigabyte MZ73-LM0 board which worked fine for some time, but would not boot again after the system having been shut down overnight. I can see from the logs that the Management NIC remained alive and connected to the switch for some time (around 12 hours) before dropping off for good.
I shall be grateful if anybody has any idea/suggestions as to what is going on… As far as I can tell the system remained in the power-off stage so the only thing flowing through the board would be the 5V standby.
Symptoms:
When board is powered, BMC LED is on but would not blink after 30 minutes (solid = booting, blinking = booted); BMC_CPU LED never turns on
The 2 Broadcom built-in NICs are alive (link and activity lights on once connected them to switch + they were able to negotiate the link speed etc with the switch), blinking every second or so, but the Management NIC is never alive (link and activity lights off, despite having been connected to switch)
Shorting the pins for power switch (on the front-panel header) doesn’t boot the system, connected fans do not spin
Board seems ok (visually) there are no burn marks
There is some kind of UART output (I tried to read this with a Raspberry Pi) from the BMC when there is power, but it was garbled
CPU pads look fine, and the sockets look fine
The NVMe drive is good (tested in external enclosure)
Testing Done (all steps have no effect):
Power cycle (unplug and replug power cable)
Removing all PCIe devices
Moving board to outside the case
Removing CMOS battery + reset CMOS via jumper
Sending Wake-on-LAN packets to the 3 NICs
Changing PSU and cables
Re-seating CPU/RAM
Exchanging CPU0 and CPU1
Removing all RAM and leaving only 1 RAM stick per CPU
Removing both CPUs and all RAM sticks
I filed a service request with Gigabyte but there has not been a reply after 24 hours. Same with their UK service company.
I have seen this happen before but never had a reproducible resolution, however some mild success with:
in two parts
I will assume you can’t connect to BMC and you have a cable plugged into strictly the BMC lan?
Use angry ip scanner and see what ip address it uses, login if you can , see logs, power up
Strip motherboard down to 1 socket (0) and the matching 2 dimms of ram, and nothing else, no drives, no video cards, no peripherals , not the other cpu, and no other ram
Plug monitor into board vga,
Attempt power on
Requires about 15-18 minutes of self cycles, and see if it comes back up to a bios screen
Would you be able to elaborate what is “mild” success (was this with one of the boards that you returned) compared to total success?
The fans don’t spin when I attempted power on, so I suppose the board is dead / something is preventing it to even proceed to power on. I have checked and there is nothing on the board that would have made a short & there is nothing that looks burned, etc.
BMC LAN port has no link nor activity light after a few hours powered on with BMC LED on, so basically, no traffic. BMC CPU LED never turns on. Same with 1 CPU and 2 RAM sticks.
sure
~mild success, able to boot, shutdown, install remaining peripherals, boot, problem returns after a while, despite bios updates to that particular board, do a return at that point
~success, able to boot, shutdown, install remaining peripherals, boot, update bios or refresh bios, problem does not return
The answer is not straight forward.
On MZ73-LMO, with any version less than Rev 3, one will have issues on first boot unless newer bios is installed and BMC updated. So to answer your core question - it must be Rev 3
The more correct answer is:
if one is buying a board from scratch on this, then it is best to buy the MZ73-LM2:
~no issues on boot
~hot running lan chip is resolved
~larger vrm heatsinks
~Gen 5 on the onboard nvme m.2
~more socket power
~additional pcie power
~able to handle greater variety of CPU variants and Ram
So given a choice of initial purchase, I would select MZ73-LM2
IF (for availability or price point) one is constrained to MZ73-LM0, then please acquire Rev 3x
Surprisingly, after 5 days of not wanting to boot, I changed the heatsinks back to the original ones and left the board without any CPU/RAM/PCIe cards installed, but with the 3V battery in place and left the board for 24 hours connected to power. This afternoon the BMC suddenly booted. The Web UI was also working. As far as I can see everything is working. There were no adverse entries in IPMI logs, as if nothing ever happened.
Then, the power on command would work immediately on CPU0 + 2 sticks RAM immediately, and worked again immediately after power off + on with both CPUs installed with 2 sticks RAM each, and I was able to POST immediately…
Supposedly, this is exactly the kind of “Mild Success” you spoke about, but it is a great success for me indeed. I shall get the LM2 soon
