Power usage for BL465 G5 increased with OA 2.32

[conradsjm]

I am seeing a higher amount of power usage on my 465 blades since I upgraded from OA 2.31 to 2.32. My blades that are in 2.31 chassis' and below report roughly ~220 while the ones on 2.32 report ~340.

Has anyone else noticed this increase in power consumption and has HP given you any answers to why.

I was told it was mis reporting in previous versions and is now reporting correctly which I feel is a real issue.

This should have been disclosed publicly if this is true.

[Neal Bowman]

Conradsjm,
have you confirmed that your ILO2 is a V1.70? This firmwware also upgrades the Power Management Controller firmware on the device

[conradsjm]

Neal,

Yes, we upgraded the ILO to version 1.70 and the bios was upgraded to the latest version as well.

Funny thing is if I downgrade the ILO and bios I get readings in the ~220 range.

HP is trying to tell me that those readings are wrong and upgrading that chassis to 2.32 will modify the reading. They told me HP made a change to some algorithm that calculates the power consumption.

I'm really not buying this so I'm trying to see what readings others get.

What do your blades report? If you don't mind disclosing.

[Neal Bowman]

Conrad,
I have upgraded my chassis in one site, and am recording ~348-353W for each of the 10 blades installed. OA firmware is 2.32. All BIOS/ILO2/E200 firmware upgraded to latest available on Bladesystems Firmware Deployment Tool 1.41 ISO.

At a second site that has not been upgraded yet, I am seeing values of 285-303W. OA firmware is 2.25. ILO2 is V.160, BL460c is 4/1/2008.

All hardware is identical between the two sites. BL460c, 32G memory, 2x E5450 CPU, Emulex HBA, and NC325 quad NIC in each blade. All are running ESX 3.50 Update3, I believe.

[conradsjm]

okay, cool.

So the question is, which is correct? If what HP is telling me is true, the all of my other chassis', a total of about ten, all have power misreading.

So when im building out systems im not properly calculating my power needs.

All of my servers are identical as well, so I know that this new power consumption is due to the change in the OA.

But what I want to know is which is correct? And why aren't they reporting to their customers that the power reading is incorrect?

That's a big issue in my opinion.

[Neal Bowman]

That is the $1M question. I did not notice the change in my power readings until you pointed out your issue. I'll make a notation of my readings before I start my next round of firmware upgrades to 2.32 and VC 1.34. I'll also confirm the readings with what the blade power configurator application provides as well.

[conradsjm]

I don't think most customers are noticing it, the only reason I noticed was because I had a blade reported it was using 10,000 watts and prevented all of the blades from powering on.

Then after that system board was replaced, I noticed they blades were using >300 when previously they only used ~220.

So I logged a call, at first HP told me this was an issue, and then they revised their numbers and told me this was in the normal range.

I'm going to use the calculator too and see what it says. I hope more people open cases with HP on this so we can get some answers.

At least be aware of the new power usage because it costing us more money.

The other strange this is that this chassis is running 2.31 and ILO 1.60 which is why HP first told me it was an issue and to upgrade the OA. When I upgraded another chassis and they reported the same numbers, they then revised their numbers to make this usage normal.

I'm making a big stink about it because I don't think they have properly communicated or documented this change. I also don't feel I am getting a straight answer.

[Neal Bowman · Accepted solution]

yeah, I had one chassis go into alarm mode when one blade was consuming 43000W (not a typo) of power and the chassis knew it could not deliver that amount. I have dynamic power enabled on all chassis and blades, and all 6 psu were showing power consumption. After a couple of hours and A blade firmware upgrade, OA is still reporting that blade at 504W, while all 9 others are pulling 190W. 9 of the 10 are configured w/1 CPU, 4G, and 1 dual nic mezz card, and the 10th has 2 CPU and 12G of memory. #10 is reporting ~300W.

[conradsjm]

Do you plan on contacting HP on this "issue". I really want to get some answers on what is going on and so far through my channels I am not getting anything.