0

For some reason (and it's possibly coincidentally suspicious that I needed to get a bit of extra slack on the SATA cable that is wrapped around/under fan module 1, so I gave it a firm pull..) my 1950 PE server is now reporting E1311 and E1313, "Fan RPM 1C" and "Fan redundancy". OSMA reports 0 RPM for 1C, even though all fans spin fine. The error persists as 1C even if i swap fan module in position 1 with another module

I thus suspect something has happened to the speed sensor circuits on the motherboard.. Whether that was caused by physically pulling on the sata cable that snakes in and out of a bunch of capacitors near fan 1, or whether the fact that a loop of cable under the SAS controller ended up rubbing on one of the fans for a while and possibly causing some electrical issue, I don't know

So it runs with the fans on max.. It's in the garage so I don't really care about the noise, but it's more power and thrashing old components so I'd like to do something about it, but not quite sure what

  • I could buy another motherboard for about $20 (about half what I paid for the server, hah) but I don't know what revision my mobo is, how to identify it, or if it even matters so long as I buy a mobo that the CPUs (SLBBS I think) and memory will fit
  • I could buy a 2950 motherboard for about $10 (why they're less I've no idea), but I don't know if it will fit or be an improvement (but it's more palatable spending only a quarter of the cost of the server on a new mobo, even if the cost of the time of fitting it...)
  • It's apparently possible to boot a linux live CD and jiggle the warning thresholds.. http://dell.ionsystems.uk/ - can they be jiggled to 0 so that the failed sensor no longer triggers an alert?
  • I could perhaps link all 4 fans (as in, the 4 mini fans in one module) together to be powered in series rather than parallel (and do this to all four fan modules) which may be the equivalent of series resistoring them all, slowing them down, and then just leave it in alert state with the server trying to run them all at 15000 rpm, and them actually running a lot slower.. But i don't know if it'll a) work electrically and b) that the server wont then have a fit and shut down anyway. It's a lot of soldering to undo

Can the thresholds be lowered to 0? Will it help? What do I need to do to get the right motherboard?

Caius Jard
  • 165
  • 1
  • 1
  • 8

1 Answers1

0

I might have resolved this by software. Mostly followed the instructions from http://dell.ionsystems.uk the original text of which is:

Guide by Euan Mutch

Link to original post: http://projects.nuschkys.net/2011/11/15/how-to-adjust-the-fan-thresholds-of-a-dell-poweredge/


Boot on a centos 6.5 x86_64 livedvd or cd avaliable here:
http://dell.ionsystems.uk/CentOS-6.5-x86_64-LiveCD.iso
then run the following commands in the terminal.



mkdir temp
cd temp
su
wget -q -O - http://linux.dell.com/repo/hardware/latest/bootstrap.cgi | bash
yum install OpenIPMI freeipmi ipmitool srvadmin-all glibc.i686
wget http://dell.ionsystems.uk/ESM_Firmware_C6H4X_LN32_2.50_A00.BIN
chmod 755 ESM_Firmware_C6H4X_LN32_2.50_A00.BIN
mkdir bmc_firmware
./ESM_Firmware_C6H4X_LN32_2.50_A00.BIN --extract bmc_firmware
cd bmc_firmware
wget http://dell.ionsystems.uk/dell-adjust-fan-thresholds.py --no-check-certificate
chmod 755 dell-adjust-fan-thresholds.py
ipmi-sensors
./dell-adjust-fan-thresholds.py payload/bmcflsh.dat
cd hapi/
tar xvzpf hapi.tgz
cd ..
export PATH=./hapi/opt/dell:$PATH
export LD_LIBRARY_PATH=./hapi/opt/dell/dup/lib:$LD_LIBRARY_PATH
./bmcfl32l -i=payload/bmcflsh.dat -f

With the following differences:

Guide by Euan Mutch

Link to original post: http://projects.nuschkys.net/2011/11/15/how-to-adjust-the-fan-thresholds-of-a-dell-poweredge/

This link is now dead

Centos 6.5 is quite old and is a problem because the repos aren't supported/are gone. I used a Centos 7.1 LiveCD from one of the CentOS mirrors. It was written to a 1GB USB stick using Rufus 3.14 with the following settings:

  • MBR
  • BIOS/UEFI
  • FAT32
  • 4096 cluster
  • ISO write mode, not DD

BIOS settings on the server (F2):

  • USB Stick Mode Auto (choices are Auto/Floppy/Harddisk)
  • BootPriority Front USB First (use - sign to move. Be aware that there are two different priority menus, which one is relevant is dependent on the "stick mode" setting)

The following commands were run:

  • mkdir temp
  • cd temp
  • su
  • wget -q -O - http://linux.dell.com/repo/hardware/latest/bootstrap.cgi | bash - this command didn't work for me
  • yum install OpenIPMI freeipmi ipmitool srvadmin-all glibc.i686

Yum gave an error about its lockfile in use - can't quite remember the process that had it, maybe PackageFix. I killed it with kill <pid>

  • wget http://dell.ionsystems.uk/ESM_Firmware_C6H4X_LN32_2.50_A00.BIN - wget just seemed to jam up. Used firefox instead, saving that file to the relevant folder wget would have
  • chmod 755 ESM_Firmware_C6H4X_LN32_2.50_A00.BIN
  • mkdir bmc_firmware
  • ./ESM_Firmware_C6H4X_LN32_2.50_A00.BIN --extract bmc_firmware
  • cd bmc_firmware
  • wget http://dell.ionsystems.uk/dell-adjust-fan-thresholds.py --no-check-certificate - again, used FF
  • chmod 755 dell-adjust-fan-thresholds.py
  • ipmi-sensors
  • ./dell-adjust-fan-thresholds.py payload/bmcflsh.dat

Altered FAN 1C to have a 0 threshold

  • cd hapi/
  • tar xvzpf hapi.tgz
  • cd ..
  • export PATH=./hapi/opt/dell:$PATH
  • export LD_LIBRARY_PATH=./hapi/opt/dell/dup/lib:$LD_LIBRARY_PATH
  • ./bmcfl32l -i=payload/bmcflsh.dat -f

During the flashing process the fans rev up and down. Post flashing, the speeds gradually dropped. OMSA now looks like:

enter image description here

..which hopefully will have "solved" it

Caius Jard
  • 165
  • 1
  • 1
  • 8