DL380-G5 power supplies overheat when plugged in but off

[Jian_2]

Hi,

We bought 5 Proliant DL380 G5 servers last November 2006, but only started using them January 2007. All power supply of each server (5 servers), starting from the top of the rack to the bottom failed. The first one failed 2weeks before the second and the the rest with a gap of 2 to 3 days only. We did have some prower failures in our area but the servers are attached to a 5KVA APC UPS with 2 external battery packs.

[mhutchesson]

Well guess what I can confirm there is a problem HP told me they had 1500 on back order into Australia to meet the need, I have had around 8 die since Sept 2006

When I have had them fail, the normal response has been none available in the country but escalation to higher levels normally finds one for me

They talk about the 03M version being good but I have had them all fail, theres one model around thats 300 grams lighter than the rest ( dont know why , but this model has also died )

Michael

[c.wolf]

Hello,
a HP Technican had told us that the Charge of defect PS all have Sparepartnumber SPN: 403781-001

The bad one have Partnumber
P/N: 379123-001 REV 03M (without green Point)

following Partnumbers are ok
- P/N: 379123-001 REV 03M (with a green Point)
- P/N: 379123-001 REV 04M (with or without green Point)
- P/N: 379124-001 (alle Revisions)

good luck

[Jian_2]

Thanks for the info guys. I'll check the part numbers of the 2nd PS that have not failed yet and the new ones that I got as a replacement.

With regards to replacement, we don't seem to have any problem or long delays in our location here in U.A.E.

[Ivan Kuznetsov]

Hello!

I measured current which server takes from mains in different modes.

Hardware is DL380 G5 with 2 CPUs, P400 RAID controller and 8 SAS SFP 10K HDDs, two redundant PSU (p/n 379123-001 REV 04M).

Mains is 230V AC 50Hz

Just 1 PSU is connected:
stand-by: 0.25-0.27A
Startup: 1.48A max
Working (running tests): 1.24-1.38A

Both PSUs are connected:
stand-by: 0.25-0.28A
Working (running tests): 0.72-0.77A

Therefore server takes ~60VA of power for each of connected PSU while in stand-by mode.
I don't know and can't measure power factor for PSU, let's consider it 0.7. PSU dissipates most of the power as heat (~40W). It is sufficient to make some parts of PSU (transistors, transformer and coils) wery hot if there is no air flow trough.

Total power taken is ~120VA (~80W) if two PSUs are connected and server is standby. I'm surprized

[Jun.K]

Hello.
Could you teach Difference point
of Measures parts and parts before Rev03M
(measures Electric element)
My best regards.

[Ivan Kuznetsov]

Hello!

No, I can't. Now we do not have DL380 G5 with old different power supply revisions

I measured temperature inside a PSU which was in standby mode for ~12h. It's 60-65C. The hotest part is the small transformer which is about 80C

Regards, Ivan Kuznetsov

[George Jires]

Hi,
our customer bought five DL380G5 servers last year and now the customer notice,that five of redundant PSs are failed. I don`t know if those PSs failed at one time (and customer didn`t know too) . Customer also complained about strong overheating in PSs area.All failed PSs are PN 379123-001,REV. O3,serial number ATZD0638153xxx (xxx= 320,323,229,319,960) ,without any green or yellow points on the label.
(Nobody know,what those points does mean ??? I have already seen green,yellow and red points on different spare parts which came from HP,and I remember that SystemBoard-DL360 R01- with the red point has got the same fault as the bad one. Are those parts recycled ?)
So,new PSs from HP are same PN,but REV.06,all with yellow point. Serial Number ATZD0726878xxx . Should be rest of original PSs replaced ?
George

[David Stavert]

I searched this forum looking to see if anyone else was concerned about the lack of redundant fans. In February HP dropped redundant fans. No longer an option although the marketing literature now claims that all the G5s have redundant fans as standard equipment. With fewer fans (6 rather than 12) and no fan in the power supply it might explain the failures. Power supply failures have been very rare in my experience. Certainly the lack of redundancy and heat issues.

I have always liked Compaq equipment since it first came out and perhaps mostly for its engineering and thoughtful design. I have purchased ProLiant servers because I believed they were building good reliable equipment. It helped me sleep at night. Even after ProLiant acquired an HP logo I still felt that comfort. They still felt the same looked the same and sounded the same. I could still see the same design thoughtfulness.

I don't have hundreds of servers in my racks but like everyone they need to be reliable. My perception of security has been shaken after twenty some years I am sorry to say. My sleep is a little less easy today.

[Brian Akey]

We also purchased our first DL380-G5's. I got 5 of them for a specific project, all with redundant power. We had a data center test with a power outage and these servers didn't get turned back on for a decent period of time after the power was turned back on. I had 2 of the RPS fail immediately and now a 3rd has failed.

After reading the other posts here, it makes sense that they were damaged by having the power applied but the server not turned on to have any heat dissipation. Sounds like a serious design flaw to me. I can understand trying to save on power by removing the fans from the PS, but then they should run in a reduced power mode. I'm sure they still need to be active because of the ILO and management agents, etc...but not a good plan to have them cook themselves.

I'm sure that HP will gladly replace the PS' now, but what about when they come out of warranty? Do I have to run around and unplug my servers when they are powered down but still have power available to them?
So much for remote administration...

Brian