Re: ReadyData 5200 running OS 6.10.3 with Sys fan control on the motherboard.

Bullfrog2 · ‎2020-09-15

Hi

I know this is an unsupported configuration, thought I'd let others know my experiences of attempting to run OS6 on a ReadyData5200 unit.

I followed some instructions online which bizarrely I can't find now which showed how to extract the OS6 firmware manually and put it on the internal stick.

Unit seems to work but the fans were incredibly noisy and I would occasionally get errors on the unit relating to 3VSB and Vbat.

I stumbled across this post relating to fan controls on the 4200 -

https://community.netgear.com/t5/Using-your-ReadyNAS-in-Business/ReadyNAS-4200V2-OS6-6-1-fan-control...

And thought to myself I see if I can modify the provided RD5200.conf file located in /etc/frontview/sensors and attached the 3 sys fans on the back of the backplane to the motherboard, I also noticed that although the original conf file has reference to "CPU Temperature" this wasn't appearing in the GUI front end.

The top part of the original RD5200.conf file looks like this

chip "coretemp-isa-0000"
	label temp1 "CPU Temperature"
	compute temp1 @%35, @%35
	ignore temp2
	ignore temp3
	ignore temp4
	ignore temp5
	ignore temp6
	ignore temp7
	ignore temp8
	ignore temp9

I attached the cabling for the fans as such.

MB Header Fan1 = CPU (This wasn't altered)

MB Header Fan2 = SysFan3

MB Header Fan3 = SysFan2

MB Header Fan4 = SysFan1

Through various attempts of trial & error, disconnecting various fans while the system was running to determine what label in the conf file related to which fan, I came up with this conf file.

chip "coretemp-isa-0000"
	ignore temp1
	label temp2 "CPU"
	compute temp2 @%35, @%35
	label temp3 "CPU-2"
	compute temp3 @%35, @%35
	label temp4 "CPU-3"
	compute temp4 @%35, @%35
	label temp5 "CPU-4"
	compute temp5 @%35, @%35
	ignore temp6
	ignore temp7
	ignore temp8
	ignore temp9
	ignore temp10

chip "w83627dhg-isa-0a10"
	label in0 "CPUVCore"
	label in1 "in1 (VIN0)"
	label in2 "AVcc"
	label in3 "3.3Vcc"
	label in4 "VDimm (VIN1)"
	set in4_max 1.75
	label in5 "in5 (VIN2)"
	label in6 "in6 (VIN3)"
	label in7 "3.3Vsb"
	label in8 "Vbat"
	set in8_max 3.63
	label in9 "in9"
	label in10 "in10"
	label in11 "in11"
	label in12 "in12"
	label in13 "in13"
	label in14 "in14"
	ignore fan3
	label fan2 "CPU"
	set fan2_min 500
	compute fan2 @*0.965,@/0.965
	label fan1 "Sys2"
	set fan1_min 800
	compute fan1 @*0.965,@/0.965
	label fan4 "Sys1"
	set fan4_min 800
	compute fan4 @*0.965,@/0.965
	label fan5 "Sys3"
	set fan5_min 800
	compute fan5 @*0.965,@/0.965
	label temp1 "Sys1"
	set temp1_offset 0
	set temp1_max 80
	label temp2 "Sys2"
	set temp2_offset 0
	set temp2_max 80
	label temp3 "Sys3"
	set temp3_offset 0
	set temp3_max 80
	ignore temp4
	ignore temp5
	ignore temp6
	ignore temp7
	ignore temp8
	ignore temp9
	ignore temp10
	ignore intrusion0
	ignore intrusion1
	ignore beep_enable
	ignore cpu0_vid

For the most part the unit works fine, but after several hours for no apparent reason at all the fans will suddenly ramp up to 7500rpm or more causing a total racket. I've come across various online documents suggesting to change the fan settings in the Bios, the best setting so far being "Balanced" , but the problem continued.

Anyway after some persistance and some private messages between myself and @Sandshark I now have control of the 3 System fans.

Without any alterations, the system seems to be running the super io chip in smart fan III mode. I can find no info on how that actually works.

You can change this so you have manual control or under temperature control.

I'll probably turn this into a startup script however in the meantime if you can get to a ssh command prompt

cd into /sys/devices/platform/w83627ehf.2576

for Manual mode

echo "1" > pwm2_enable

echo "value" > pwm2 where value is 0 for slowest 1500rpm 255 is for fastest around 7000rpm

for temperature controlled mode which is what i'm going to use

echo "2" > pwm2_enable

echo "34000" > pwm2_target which is 34 degrees in millidegrees or your target temperature.

I have the setting in BIOS set to balanced at the moment, I don't think that makes any difference.

Data Integrity seems to be fine, copied several terabytes on/off the unit and used various utilities to compare what's copied with the original data, and it's all been good so far. Just wish I could sort the fan problem.

I'd post a picture of my Gui here but can't see how to post pictures. They are now in my profile though

Hope this helps others. Remember this is all unsupported so shouldn't be used in a business critical scenario.

Thanks

Bullfrog2

Bullfrog2 · ‎2020-09-16

Hi

I've since settled for 40 degrees as my target temperature otherwise as soon as the machine ceases to idle the fans start ramping up. Essentially with the mode I've selected , the sensor connected to temp2 controls the three fans. The readynasd process still controls the CPU fan.

These datasheets/documents were my source of info on the fan/temp chip.

https://www.kernel.org/doc/Documentation/hwmon/w83627ehf

http://read.pudn.com/downloads649/doc/2643726/W83627DHG-Programming-Guide.pdf

https://www.nuvoton.com/resource-files/DA00-WW83627DHG-P.pdf

Bullfrog2

Sandshark · ‎2020-09-16

I've had problems with OS6 controlling the fans on my converted RD5200 and had turned to using the BIOS for it (which just sets a set percentage of full speed). I'm now testing the pwm2_enable = 1 change (mine was set to 4) and will post more. I'm using a copy of the RD4200V2.conf as my RN5200conf.

Note that both of us moved the fans from the SAS backplane, where they always run full blast, to the motherboard, with the connections as @Bullfrog2 described. That's the same as in an RN4200V2, which has the same motherboard (sans IPMI hardware).

Bullfrog2 · ‎2020-09-17

Hi

With pwm_enable set to 1 you have to set the speed yourself. I suspect the BIOS is taking over once the temperature gets to a critical value, it's hard to tell as I believe I'm fighting a hardware fault too.

My pwm_enable was also originally set to 4.

Anyway here is some more info i've discovered while researching this.

Chip is capable of controlling 4 seperate fan outputs using either DC or PWM.
Controlled by pwm1 to pwm4, not sure what order these are in but this is the names of the 4 pins on the chip for fans.

SYSFANOUT
CPUFANOUT0
CPUFANOUT1
AUXFANOUT

On the supermicro board all fan headers including CPU are wired to the same pwm2 output, so they all go up and down together. I don't have anything plugged into header 5 at the back of the motherboard so this connector isn't verified.

Chip is capable of monitoring 5 Fans ,showing their speed, or setting triggers for alerts if they fail.

Some fans generate 4 pulses per revolution others 2 per revolution, so settings need to be tweeked to accomodate this. Nothing to do with divisor which is more to do with showing and having better control between min and max speed.

The Netgear conf file monitors temp1 which doesn't budge much to control the cpu fan, the other fans we've added to the motherboard just tag along for fun.

The chip is also capable of controlling fan speed via the southbridge, so our fan pcb headers may not be connected directly to the w83627 chip at all, also it appears the IPMI can override stuff too.

I've also discovered that temp2 rather than temp1 on the w83627 is the cpu temp diode, so perhaps all that is needed on a properly functioning machine is for the RD5200.conf file to check temp2 rather than temp1.

Bullfrog2

Bullfrog2 · ‎2020-09-18

Another update - at one point I'd updated readynasd's debug level to 5 (

/etc/systemd/system/multi-user.target.wants/readynasd.service - change this line

ExecStart=/usr/sbin/readynasd -v 3 -t --- change 3 to a 5)

which generated more logging information available for view by running journalctl or viewing systemd-journal.log. journalctl -b will show entries from last boot.

This now lists lots of information amongst other things about the device Id's it uses to sense temperature/fan speeds etc.

It then seems to make a calculation based on drive temperatures, cpu temp, system temp to make an assesment as to whether to increase/decrease fan speeds.

It appears that conceptually readynasd expects to write to fan1 to control fan1 , write to fan2 to control fan2....., write to fan5 to control 5 and this appears to be where the source of our permission denied error message is coming from, this chip can only control 4 fan outputs although it can monitor 5

If you change the .conf file to ignore fan5, the "Permission Denied" error stops.

And if you type echo "64" > pwm5 you indeed get a "Permission Denied" error message.

So I'm now running with yet another conf file.

I've also just ordered some fan extension cables, so I can plug a fan into the header at the back of the motherboard to see if that is controlled seperately.

I'm also now running from one PSU only to see if my random notifications of voltages out of spec disappears, this being the original reason for me adding in entries for the voltages in the conf file.

Bullfrog2

Sandshark · ‎2020-09-18

Fortunately, the power supplies are standard ones for 2U SuperMicro chassis and readily available. The newer "gold" versions are also a lot more quiet, as I understand it. You can try each one separately and see if one is your problem, but the problem could be the monitoring system rather than the actual voltage. You should probably also try both slots The NAS only runs on one, even if two are plugged in. The other is a hot spare. I just run mine with one and purchased a blank to fill the other spot. I don't have a place to plug both into my UPS (too much other stuff in the rack) and wasn't sure the same one always comes up as primary, and I want the primary to be on the UPS so it sees the current draw for the hold-up time estimate and so the UPS doesn't see a load spike at the same time it shifts to battery. Plus, it's more quiet with just one of those little fans running and my spare is now in the closet and won't go bad before even called on to power the NAS. You can just leave it not fully engaged or totally missing instead of getting the blank, but I figured the air flow would be better with the blank and they are pretty cheap on eBay. If both are inserted and only one plugged in, you'll get an incessant alarm.

My .conf, which is just a copy of RN4200.conf has ignore fan3, and fan3 seemed to be the rear one when I did my experiments. So something you've done with lmsensors may have changed the designations from the what Netgear uses. My fan5 is labeled "Fan 3", and that does show up in the GUI.

chip "coretemp-isa-0000"
        ignore temp1
        ignore temp2
        ignore temp3
        ignore temp4
        ignore temp5
        ignore temp6
        ignore temp7
        ignore temp8
        ignore temp9

chip "w83627dhg-isa-*"
        ignore in1
        ignore in4
        ignore in5
        ignore in6
        label fan1 "Fan 2"
        set fan1_min 400
        label fan2 "CPU"
        set fan2_min 0
        ignore fan3
        label fan4 "Fan 1"
        set fan4_min 400
        label fan5 "Fan 3"
        set fan5_min 400
        label temp1 "System"
        set temp1_max 60
        label temp2 "CPU"
        set temp2_max 74
        ignore temp3
        ignore temp4
        ignore temp5
        ignore temp6
        ignore temp7
        ignore temp8
        ignore temp9
        ignore temp10
        ignore intrusion0
        ignore intrusion1
        ignore beep_enable

I've seen posts concerning this family of Supermicro boards that makes me think I may also be getting the access denied errors because the BIOS and/or IPMI are trying to access the fan control at the same time as the OS. There is no "external control" BIOS fan setting like there is on newer boards. I don't know if your problems with the other sensors could be associated with colliding reads of the data by the BIOS and/or IPMI, as I don't get them. I don't see anything in the .conf file to set them up, either, but I see the voltages in the log files.

Bullfrog2 · ‎2020-09-18

Hi

Thanks for the updates via various route.

I've been running machine today now for around 8 hours with no voltage alerts. I removed the top PSU last night. However I've been caught out before thinking something has been solved with this unit. So I'll wait before declaring this a success.

I've essentially removed the original references to fan5 and just have ignore fan5 there now in the conf file, no permission errors are being logged anymore.

With the BIOS fan setting set to Balanced, and the GUI setting set to quiet, the fans at boot were sitting around 2200 rpm (CPU) and 4000 rpm (System ones) , I'm not giving the unit a great deal of work but after 8 hours CPU is now 1054rpm and System ones are 1687rpm.

So hopefully this time I'm on the correct track.

I don't think this has made any difference to the above but mention it anyway , I've also changed a setting on the IPMI board from failover to Dedicated in the network settings, the idea of this setting is that if there's no cable in the management port when the chassis is powered up then the IPMI sits on the same network as port1 using an internal network switch. My thoughts which I haven't actioned yet were to pull the memory from IPMI if that's actually possible to disable the IPMI part of the board and remove the cable from the management port, thus disabling it. The idea being if it's disabled it can't interfere. Yes i too wondered if it was a read clash but as your not getting the bizarre voltage errors or random fan failure announcements that I am I suspect this problem is specific to my machine and a real alert problem and not due to an error in the .conf file.

Bullfrog2

Sandshark · ‎2020-09-19

Before you go to the extreme of removing components, you may want to try the BMC Disable jumper.

Bullfrog2 · ‎2020-09-20

Now that would be far too sensible - lol - didn't even think of that.

Thanks for the suggestion.