Power Supply Faliure (a kind of)

[Ayman Altounji]

Well, out of sudden, server started crashing.

Then it enters some sort of loop where you hear a bzzz sound, HDD starts spinning up (for a half a second) and then stops, then again repeats the same thing (bzzz & HDD spin up/down). There is a LED turning on on the MB (close to cca 3rd PCI slot), but turns off together with the HDD. Loop has interval of about 2-3 seconds.

It is purely hardware problem, linux syslog doesn't detect anything (ok, low-level hardware problem, higher level problems would be detected) - I managed to get a look at the syslog when I managed to boot it, but after that it's completly dead.

So. there's a power to MB, main fan is working (so PSU works, gives power, it's not a burnout), but FAN ON THE POWER SUPPLY is dead (doesn't turn).

So I presume it's a PSU (fan) problem and system detects somehow that fan is dead (which is bit odd since I've seen computers working with PSU fans dead for hours before burning out, is this somesort of "clever" MB and PSU where MB can see if PSU's fan is not working). PSU is cold to touch (so it's no overheating), but system doesn't get further than bzz-spin-bzz-spin-bzz-spin.

Anybody had similar experiences? Is it safe to presume it's ONLY PSU problem (it is of 0H revision, the "troubled" one, as described here on the forum).

Zvonimir Zelenika
CARNet

[Ayman Altounji]

There may be something in the system that could have shorted and is holding down the power supply. In this case, try removing any/all PCI cards and check to see if the system comes up (note: try this first). If the condition persists, you may want to replace the main system (processor) fan. note: just because it is spinning, doesn't mean that it is spinning up to the corect RPM (note: fan speed is monitored and a fan that is spinning too slowly will be read as a failed fan by the system) and the server will keep rebooting. Otherwise, replace the power supply. The worst case would be one of the thermistors that monitor temperature zones on the system board is incorrectly reporting an overheat condition which would require system board replacement.

[Ayman Altounji]

I am experiencing the same characteristics of failure, a spare PSU is being shipped to me today, hopefully this will fix it The server will start up after failure but sometimes several minutes after doing the 2 1/2 second loop, it then only runs for 1/2 - 2 hours before crashing again. Nothing is written to NT4 event log.

[Ayman Altounji]

It will do so for a day or two (manage to boot from time to time and work for few hours), but after that it will fail totally, just spinning up in that odd loop.

My server first had 3 crashes during the night, and then two times during the following day, but after next one it failed completly, just bzz/click etc.

Also, still didn't get a spare part, have allready 3% downtime on year-basis (14 days today since the failure), but "Next Bussiness Day, On-Site, 3 Year Guarantee" means something else when you're in Croatia :op

I could have been better off buying components and building a server myself. Would have spare PSU in matter of hours, or I'd buy extra PSUs in advance. But, to get "support & service" we bought Compaqs from their main representative in Croatia.

Zvonimir Zelenika CARNet

[Ayman Altounji]

Got it repaired. After cca 20 days, but luckily it was only PSU that went wrong. Our supplier said that they ordered some extra PSU's so they'll be able to react much faster than this time. Works OK now, it's 04* revision (as far as I could see).

Zvonimir Zelenika CARNet

[Ayman Altounji]

We got the exact problem on our 6 months old ML350 this morning. With so many PUS complains posted on this server, Compaq should just acknowledge their quality control problem and post a note/fix on their support page, instead of telling everyone to go look for shorts inside the machine!

[Ayman Altounji]

What the heck is going on with these power supplies? One of my ML350 dropped 42 times before I was able to backup all of my databases.

No errors log, no reports to the Insight Manager, just power is gone, then reboots.

Must be a pretty large problem, since the PSU are back ordered. I've been down a whole day now, good thing I've got the next business day warranty, should only be down about four days according to Compaq Support. I can't even get the FE to call me to give me a approximate arrival time or day.

Anybody else having problems getting Compaq to fix their PSU?