PWR-C45-6000ACV in Cisco 4510R+E

[Matt Reader]

Strange question regarding power provision.  I have 4510R+E with 2 PWR-C45-6000ACV supplies with 4 identical feeds.  The switch is configured for power redundancy mode. (We have many of these - all exhibit the same problem - identical hardware/IOS/config).

When one of the power feeds is disconnected, the whole PSU goes into error disable.   I wasnt expecting this to be the case and we are drawing no where near any capacity.

Is this correct behaviour, bug or misconfig?  Do we need to specify max psus (not that I was expecting to do so?)

Thanks

Matt

Power                                             Fan      Inline

Supply  Model No          Type       Status       Sensor   Status

------  ----------------  ---------  -----------  -------  -------

PS1     PWR-C45-6000ACV   AC 6000W   err-disable  good     good

PS1-1                         220V   good

PS1-2                                off

PS2     PWR-C45-6000ACV   AC 6000W   good         good     good

PS2-1                         220V   good

PS2-2                         220V   good

*** Power Supplies of different type have been detected***

Power supplies needed by system    : 1

Power supplies currently available : 1

Power Summary                      Maximum

(in Watts)              Used     Available

----------------------   ----     ---------

System Power (12V)       1200        2200

Inline Power (-50V)       497        4760

Backplane Power (3.3V)     40          40

----------------------   ----     ---------

Total                    1737 (not to exceed Total Maximum Available = 6000)

                             Output Power To Chassis            Input Power

                    =======================================  ==================

Power Measurement   System Power (12V)  Inline Power (-50V)

(in Watts)                (+/- 12W)        (+/- 50W)              (+/- 5%)

------------------  ------------------  -------------------  ------------------

PS1                        171                  110                  412

PS2                        372                  250                  828

------------------  ------------------  -------------------  ------------------

Total                      543                  360                 1240

Note: System power budget is 1200 Watts to accommodate peak power utilization.

[Reza Sharifi]

No completely sure, but I think this is what is happening:

When you have a 6000W power supply with 2 feeds, that is essentially 2 3000W power supplies each with one feed.

So, when you disconnect one of the 2 feeds, the system thinks, the power supplies mismatch now since one is 6000 and the other 3000 and so it puts the 3000 one in err-disable.

Also, have a look at this link for more info:

http://www.cisco.com/en/US/docs/switches/lan/catalyst4500/12.2/31sga/configuration/guide/pwr_envr.html

When you insert power supplies in your switch, use power supplies that are of the same wattage. Multi-input power supplies such as 1400 W DC triple-input and 4200 W AC have additional restrictions. Read the sections on special considerations for these power supplies. If you mix power supplies, the switch will use the one it recognizes first and ignore the other power supply. The power supply status displays as err-disable and the summary displays as all zeros (0) for wattage values in the output for the show power command.

HTH

[Matt Reader]

Thanks, I've seen that link.  Also found another document that states if only one input is connected to the 6000w supply, the total output power is 3000w.

So I'm wondering what happens if we disconnect 1 or 2 feeds of the primary PSU - what will happen?  Unfortunately it's a live switch, but it may have to be control tested to find the answer

[Reza Sharifi]

So I'm wondering what happens if we disconnect 1 or 2 feeds of the primary PSU - what will happen? Unfortunately it's a live switch, but it may have to be control tested to find the answer

That exactly what I was thinking.  In theory nothing should happen since you are not drawing 3000w, but the reality is that this is a live network and you probably don't want to risk bringing down the entire switch to test the PSUs.

Be nice to have a test switch...

[Matt Reader]

Done a controlled test on this now.  By swapping to any 1 or combination of the 4 supplies the switch continues to work and adjusts accordingly without any visible problems.  Didnt see any packet loss on a test I was running at the same time. (although the results would be very different if I was drawing over 3000w).

The err-disable and syslog messages are probably a bit misleading.  Only when the 1 + 1 or 2 + 2 inputs are in and working does it remove the err-disable on the supply.