ML370 G3 - Fan Failure

[pslawrence]

We have an ML370 G3 server which until this weekend just gone was working reliably and correctly. The unit is now unable to fire up all 6 of the internal fans and the internal health warning light is on.

Internal examination show that Fans 1 through 4 all show the 'failed' as indicated by the amber LED. Fans 5 and 6 are working ok. I have swapped the fans around systematically to prove that each fan does in fact work perfectly if positioned in sockets 5 or 6 only.

If I remove fans 2, 3 and 4 then fan 1 may (about 60% of the time) fire itself up at boot but this is still insufficient to allow the server to boot up. With a little coaxing I have managed to get a fan in slot 3 to fire also which boots the server but soon afterwards a shutdown will initiate due to a loss of cooling.

To avoid the shutdown temprarily I have adjusted the system time to delay the shutdown, I realise I could disbale this shutdown completely but long term didn't wish to do this.

I have used the version control manager to ensure all drivers are up to date and have used the firmware maintenance CD to check that also.

In the BIOS I have changed the default fan speed to 'High' instead of 'Normal' but this makes no difference.

This server is just outside of warranty, what are my options? Should I assume that it is either the fan/thermal logic circuit or fan power circuit both on the main system board and replace that at about £450 or could it be something else?

[Terry Hutchings]

What BIOS/FW revision was installed on the server at the time this problem occurred? The revision from 13 Aug 2004 is supposed to keep this problem from happening. If you did not have this version installed prior to the "fan failure" errors, then it's usually too late to apply this BIOS upgrade. At this point, it will most likely require a replacement system board.

[pslawrence]

The Firmware was Compaq P28 01/05/2004 at the time of the fault. The firmware is now up to date. How would the BIOS update have prevented this?

[Terry Hutchings]

The firmware speeds up the default fan speed to keep the component on the system board from becoming damaged. Generally, it's too late to upgrade the firmware on the server once this type of failure occurs.