C6100 bios / CMOS issue

[Sielbear]

Based on the wealth of info here, I purchased a C6100 for our small business. I installed the disks and started working in the bios. It appears I'm on bios 1.70 FYI. Here's what I noticed:

When I make changes to the bios (all 4 nodes behave identically) the changes don't appear to save. I first changed the ipmi settings with a static / dedicated nic. I noticed I could not load the web page, so I went back in the bios. This time I spent more time customizing things- enabled virtualization / vt-d, etc. Went to save, rebooted. This time I caught a glimpse of a message before entering bios: CMOS checksum bad.

I'm guessing default values are being reloaded. Looking in the bios, I can read the bmc error log. I see the same pattern of:
Start post

Post error

Post error
CMOS checksum bad

I see reference to a set of jumpers to clear the CMOS as well as reset system defaults. If clearing cmos is a good step, do i set the jumper, reinstall the cpu node, power up, power down, remove jumper? or do i just touch the jumper briefly to clear?

Additionally I've seen indications that a bad CMOS battery could cause this. I've not yet pulled the nodes out as it was so late when I got this far.

Anyone seen this? All system components / all ram appears to be properly detected. Again, all 4 nodes are behaving the exact same way.

 

[PigLover]

Your symptoms are consistent with a dead CMOS battery. Replace one and see if that node behaves better.

 

[Sielbear]

I'll do this tomorrow. I hope that's the case and an easy fix sets things right!

Next I'll need to look at this new bmc firmware that was posted today.

 

[Sielbear]

I've pulled the first battery from the chassis, but on a volt meter it's measuring right at +3.04 V. Looks like it's not going to be a battery problem.

 

[Sielbear]

Pulled battery and re-seated. I also placed a jumper on the clear cmos pins, however, I don't know if I need to connect to power to dump the cmos or if just jumpering the two pins should send the charge to ground pretty quickly. After performing those steps, I have the same issues. Default values are loaded from CMOS upon each boot. I did confirm that if I modify the virtualization setting in BIOS, the entire chassis powers off (when 1 node running) to make the change, so it appears it's acting like it is supposed to, however, something is wonky with the cmos save / store process.

I'm showing BMC running firmware 1.04, but BIOS is definitely 1.70. Is it possible old BMC firmware with the newer BIOS could cause this issue?

 

[Sielbear]

I picked up a battery today and replaced, however, same symptoms as before.

 

[Sielbear]

Another update for you - I have been in contact with the seller. They indicated some issues with BIOS after updating some of the (presumably DCS) boards from one batch of purchases. They are updating the bios on 4 replacement nodes and will be sending my way. I'll install RAM, CPU, and SAS card off the existing nodes to see how it goes. I'm hopeful this will resolve the issue.

 

[PigLover]

You care to share who your seller is? Honorable dealers who give this level of service deserve recognition...

 

[Sielbear]

Mrrackables / unixsurplus
They had the 24 drive chassis I wanted. Would have liked the 5639 procs, but I can upgrade later as the prices fall. Pretty good to work with. Sending me replacement nodes - have 30 days to return the originals.