UCSB-PSU-2500ACPL versus N20-PAC5-2500W

[MIKE MATHENY]

Our older chassis have 5108 power supplies.

N20-PAC5-2500W  

The newer chassis have 6508 power supplies.

UCSB-PSU-2500ACPL 

We just plugged in the two new B420 blades and within two hours we are getting thermal faults.

What is the difference between these two power supplies?

[Walter Dey]

See

http://www.cisco.com/c/en/us/td/docs/unified_computing/ucs/hw/chassis-install-guide/ucs5108_install/ucs5108_install_appendix_0100.html#ID256

Specifications for the Cisco UCS 5108 Blade Server Chassis Power Supply Units

Table 5 AC-input Gold Power Supply (N20-PAC5-2500W) Specifications

              Table 6 AC-input Platinum Power Supply (UCSB-PSU-2500ACPL) Specifications

[Keny Perez]

The thermal problems might not even be related to the PSUs themselves but instead a congestion in the I2C bus of the chassis.

Open an ssh session to UCSM and run the commands:

1-connect local a

2-connect iom #   <<< # is the chassis number where the IOM is affected

3-show platform software cmc showi2c

4-show platform software cmcctrl ohms all

5-Exit and try step 1 with "b" at the  end and use the same chassis # in step 2.

6-Post outcome here.

-Kenny

[Walter Dey]

Please post as well the UCS firmware version ! a lot of this I2C bus problems should actually be fixed, at least in the newer versions.

I2c Software Fix

•Multiple symptoms across the system

•CSCue49366

•Fixed in 2.0.5b and 2.1.1f

•Fan/Thermal

•CSCtx52556

•CSCtx49686

•CSCtl43716

•CSCua10675

•PSU

•CSCtx90410

•CSCuc52899

•I2C/HIF Flap

•CSCtq10987

•Blade loss

CSCto6853

I2c Hardware Fix

Engineering Change Order was implemented in August 2011, additional HW changes to improve margins for i2c are in the following part/part numbers

2104 IOMs with fix PN 73-11623-08 and later have modification

22xx IOMs are unaffected

2500W PSUs with fix PN 341-0293-10 and later have modification

Platinum PSUs have I2C reset circuitry after timeout event