WD MyCloud Mirror Gen2 (BWVZ, GrandTeton)

Hi Carl,

yes the fan stuff is still experimental and I don't know yet what works well. I did put in some old 3.5" inch disks but they are "only" 5400 rpm so still quite cool. I will look for some older disks tomorrow.

I don't think the 80 is hardcoded, if it is, it's a bug, but I used the value since that's about where my fan starts running reliably. Upon startup of the daemon I use (min + max) /2. You should be able to set speedMin in the ini-file to some different value.

If the fan is stopped and needs to be started again I use speedMin + 20% to be sure. It should speed down again quite fast.

I appreciate your input and will probably change the fancontrol as our experience rises.

What I observed with the automatic regulation on my device is, that the fan has "silent" values even at higher speeds and some that make more noise even if the speed is slower. Maybe it would make sense to return to some "step based" control, with the steps available in the ini-file, so everyone can put in the values that are silent for each device.

I don't know, just thinking "loud".

I also changed the usage of smartctl to not wake up sleeping disks. I don't know what to do with the temperatures in this case yet. I also need more error handling and logging when querying the disks.

Also CPU temperature reading is on my todo list. But I will probably not use it for regulation more as an emergency check as the value is changing quite fast.

bye
Martin

Hi Carl,

I implemented two new commands: SetFanMode and SetFanSpeed

With these one can switch off the automatic fan adjustment and try different values under different loads. From my tests, the device needs around 2-3h for the temperatures to stabilize when load and/or fanspeed is changed. So this is a rather slow testing process.

I was thinking some more about a good cooling strategy. Basically the best "slow" mode is the fastest speed that is not noticeable against the noise of the harddisks. There is no benefit in reducing the speed more than below the noticeable level, as the hardware (disks and system) benefit from lower temperatures (longer disk life, less degradation on board components over time). We get an advantage if we can switch off the fan completely for some time (longer fan lifetime, less dirt and dust in the system).

For my "real" use case with the two 3TB 2.5" drives this means, that the fan is mostly off when the system is idle. It switches on for about 5min per hour (speed 100) with min=42 for the system and min=38 for the disks. The system stays that way around 40-42 and the disks cool down to 32.

When the system is under full load (dd if=/dev/zero of=<crypted lvm disk> bs=64k) the temperatures rise to 50 on the system and 39 on the disks also at a fanspeed of 100. So for this easy case I'd just need 3 positions: off, 100 and full.
Full would just be needed for some rare "emergency" in case the temperatures rise too much (maybe in summer I need to test again :-).

Someone promised to give me two old 160GB 7200rpm disks. Let's see what cooling these will require.

bye
Martin

Hi Martin, Hi Carl,

you are far ahead of me in knowledge and tinkering-ability with this stuff, and i hardly understand what you do, but i'm happy that you do it. I just wanted to let you know, that i'm still interested and read all about your ups and downs and ups again :-). You are my heros!

I also wanted to let you know, that i bought another WDMCMV2 in the meantime, because it was again cheaper than the needed disks without this wrapper, and that i donated the box to the russion WDMC-hacker Fox_exe for R&D. Because of the language barriers it`'s tiring to work with him, but i think it's worth to do it anyway because he is a fine guy. That's why i want to suggest, that you get in contact with him and share work and knowledge...

Thank you, best regards
Jan

Hi Jan,

thanks for starting this topic and also for donating a device to Fox_exe. I used some of the information he provided on his drive to implement some things and it saved me time. I also found his posts in the WD forum. I'm not so much a forum guy and post here only, since I don't need to register. Feel free to give him my email or direct him to our github repos. I'd be glad talking to him. I did some valuable work.

As concerning your initial post, I think the is not much missing. I think it will take a couple of weeks to find a good cooling strategy which keeps the device at low temperatures while minimizing the annoyance for users by the fan. With the changes to the mcm-daemon and the kernel patch there is everything in place for your use case. Actually I use the device already to record TV shows via DVB-T to disk (when I'm not testing with older disks).

Carl has done a great work with the initrd and it is working. Also modern kernels can be used on the device which provide real good crypto performance on the armada internal accelerator.

So I guess besides from some minor polishing all would be ready to go for you.

There are some things missing which I'm not very interested in.
- the RTC stuff in the Microcontroller, if it exists at all in this device (waking the device at specific times)
- wakeonlan (which seems to be quite complicated between the armada and the MCU and selecting the correct PHY)

If anyone would figure that out I'd of course integrate it in the mcm-daemon, but I'm too lazy researching this as it is no low hanging fruit and I don't need it.

bye
Martin

@ Martin,
So with my "worse case"-HDD (old-wd-black) your last build works quite well, as you said it takes a while to settle, but it does.
The problem with "fixed"(-even in config) start values is that the fan does not start every time and on every device with the same value: so my fan starts turning with 55-65 and ~2000rpm; 80 value is about 3500 rpm and i "can" hear it.
But this is realy no big deal as it's set in the config.
Maybe two changingspeed would be nice, controlable with the config [slow] [fast] or [silent] [cool]

I changed now to one of my nas disk and with it the fan is most of the time silent :-), so nice work!

For the initrd key-interruped it is working with my pc, vm and notebook from serial, so I can't get bug :-|.

Quote
MM
There are some things missing which I'm not very interested in.
- the RTC stuff in the Microcontroller, if it exists at all in this device (waking the device at specific times)
- wakeonlan (which seems to be quite complicated between the armada and the MCU and selecting the correct PHY)

I agree to you: It's a nice to have but the device is working quite good without. And I dont't think I'll use these features as well.

What still "bothers" me, is the comunication with the daemon:

echo "Command" | nc -q1 localhost 57367

is not very conformable and to enter telnet is nither so here a more strate forward command handly would be great.

Further do you think I would work to write the fan-rpm and temp into files in /sys/class/hwmon1/ to have it readable for programs like "sensors" or other standard monitoring apps?

May I ask what DVB software you are using?

@ Jan
Nice gesture from you and of cause Fox did some good stuff, especialy with the dsm port. But he seemes not so interested in plain Debian or Ubuntu.
If things are tested and all works well I'll do a post at the wd forum and chat again with fox :-)


I'll do a new and clean rootfs, an updated 4.9. kernel and an *.bin firmware to flash from wd-webinterface

cheers carl

-------------------------------------------------------------------------
GitHub https://github.com/cschil

@Carl,

When you mentioned about the bootloader was locked down, it made me curious! could you post a serial bootlog, and the u-boot envs. Just in case I might think of someway to unlock it.This u-boot is probably the same as the one on NAS326.

Update:

I got the serial log in the 1st post. So I only need the envs.

Can you stop it at countdown?

Hit any key to stop autoboot:  0

-bodhi
===========================
Forum Wiki
bodhi's corner (buy bodhi a beer)



Edited 3 time(s). Last edit at 01/26/2017 03:35PM by bodhi.

hey bodhi,
Env's and I guess every promp from the uboot are already posted from JanN (first post as well), and of cause I can stop the boot process (with key "1").

Yes, I think the uboots are very similar, but in the wd uboot version you can't save env's as there is no saveenv command.
So I guess the only way is to flash a free uboot, but with a own initrd, there is a good work around.
If you find a way to flash the 326, it should work with the mcm / ex2u as well

-------------------------------------------------------------------------
GitHub https://github.com/cschil



Edited 1 time(s). Last edit at 01/26/2017 04:08PM by Peacemaker.

@Carl,


> hey bodhi,
> Env's and I guess every promp from the uboot are
> already posted from JanN (first post as well), and
> of cause I can stop the boot process (with key
> "1").
>
> Yes, I think the uboots are very similar, but in
> the wd uboot version you can't save env's as there
> is no saveenv command.

I see. That should be doable if I can get the GPL source for this box u-boot. You might not even need to flash it.

They are the same version, but built at a different date and likely have some mods specific to this box.

-bodhi
===========================
Forum Wiki
bodhi's corner (buy bodhi a beer)



Edited 2 time(s). Last edit at 01/26/2017 05:09PM by bodhi.

Hi Carl,

Peacemaker Wrote:
-------------------------------------------------------
> @ Martin,
> The problem with "fixed"(-even in config) start values is that the fan does not start every time
> and on every device with the same value: so my fan starts turning with 55-65 and ~2000rpm; 80 value
> is about 3500 rpm and i "can" hear it.

Yes, but how would you solve it?
I would just put a too high default value into the config and let the user fix it for himself. What better solution do we have?

> What still "bothers" me, is the comunication with the daemon:

echo "Command" | nc -q1 localhost 57367

> is not very conformable and to enter telnet is nither so here a more strate forward command
> handly would be great.

Yes, I will change that some day to something like:

mcm-daemon -x <some command>

But since you have a powerfull UNIX system just use the following bash function in your .bashrc until I have implemented the -x option: ;-)

function mcm() {
        echo "$*" | nc -q1 127.0.0.1 57367
}

Then you can just do:

root@nas1:~# mcm SetFanSpeed 200
OK

At least that's what I use at the moment to work around the still missing feature (and why will probably invest more time into fancontrol before I code that).

> Further do you think I would work to write the fan-rpm and temp into files in /sys/class/hwmon1/
> to have it readable for programs like "sensors" or other standard monitoring apps?

Not below /sys as that would only work from within the kernel. I could write the data to some file under /var/run/mcm or something like that, preferably some place, that resides on a tmpfs.

> May I ask what DVB software you are using?

Yes I'm using tvheadend from github which I compiled into a debian package on the device itself with a TechnoTrend TVStick CT2-4400 and can happily receive the new H.265 DVB-T2 stream here in germany. to display the stuff and watch live-TV I use kodi as client for tvheadend on my laptop or desktop. You can also stream to mplayer or vlc.

Bye
Martin

Hi bodhi,

bodhi Wrote:
-------------------------------------------------------
> @Carl,
>
> They are the same version, but built at a
> different date and likely have some mods specific
> to this box.

I think fox_exe wrote some text somewhere on his drive on where the environment is located in flash. So one could maybe change it from debian with the u-boot tools.

But until i have a working serial kwboot for the box, I woudn't tinker with u-boot or the environment as I have not seen any jtag headers on the board yet.

Bye
Martin

Hi Martin,

> I think fox_exe wrote some text somewhere on his
> drive on where the environment is located in
> flash. So one could maybe change it from debian
> with the u-boot tools.

True. This is the best way. What I usually do first. So:

- Get the /etc/fw_env.config from stock OS
- Copy it to your rootfs /etc/
- Install mtd-utils
- And reboot back into Linux
- Check with

fw_printenv

The above is quite safe to do.

> But until i have a working serial kwboot for the
> box, I woudn't tinker with u-boot or the
> environment as I have not seen any jtag headers on
> the board yet.
>

Try kwboot it using stock u-boot in mtd0.

-bodhi
===========================
Forum Wiki
bodhi's corner (buy bodhi a beer)



Edited 1 time(s). Last edit at 01/26/2017 07:38PM by bodhi.

@bodhi
I'd try your way some time ago, but as stock OS don't has fw_env, I didn't get access the env's from debian.

@MM
Thanks for the push to bashrc :-)

Quote
MM
Yes, but how would you solve it?
I would just put a too high default value into the config and let the user fix it for himself. What better solution do we have?

Maybe implemet in the config.ini an auto option for min fan value that work like this:

while fanrpm < 1
{
    fanSpeed ++
    fanRpm = readFan();
    sleep 0.25
}
fanSpeed  = fanSpeed + 0.02*fanSpeed  #for to be sure

If I have time this weekend I'll do a commit :-)

Quote
MM
> Further do you think I would work to write the fan-rpm and temp into files in /sys/class/hwmon1/
> to have it readable for programs like "sensors" or other standard monitoring apps?

Not below /sys as that would only work from within the kernel. I could write the data to some file under /var/run/mcm or something like that, preferably some place, that resides on a tmpfs.

Ok, I thougt maybe there is a way from userspace, I don't know if this get any benfit.

-------------------------------------------------------------------------
GitHub https://github.com/cschil



Edited 1 time(s). Last edit at 01/28/2017 09:58AM by Peacemaker.

@Carl,


To get access to u-boot envs in Debian, install mtd-utils.

apt-get install mtd-utils

And then

cat /proc/cmdline
cat /proc/mtd
cat /etc/fw_env.config

and compare the output with ones in stock OS. Make sure they are the same (For cmdline, only mtdparts is relevant for this).

-bodhi
===========================
Forum Wiki
bodhi's corner (buy bodhi a beer)



Edited 3 time(s). Last edit at 01/28/2017 04:18PM by bodhi.

@Bodhi
I know this, but the stock-os has no /etc/fw_env.config.
So I have no idea where the env's are located in the mtd0. and so fw_printenv does not work.

-------------------------------------------------------------------------
GitHub https://github.com/cschil

@Carl,

I see.

[    5.335654] 8 ofpart partitions found on MTD device pxa3xx_nand-0
[    5.341782] Creating 8 MTD partitions on "pxa3xx_nand-0":
[    5.347205] 0x000000000000-0x000000500000 : "U-Boot"
[    5.352740] 0x000000500000-0x000000a00000 : "uImage"

I'll take a look at this u-boot GPL.

Update:

Actually it is right here in the boot log:

U-Boot Environment:      0x00000000:0x00080000 (NAND)

-bodhi
===========================
Forum Wiki
bodhi's corner (buy bodhi a beer)



Edited 1 time(s). Last edit at 01/28/2017 05:59PM by bodhi.

NAND is usually 128K sector size. And the env is usually 128K size:

fw_env.config

# MTD device name	Device offset	Env. size	Flash sector size	Number of sectors
/dev/mtd0 0x00080000 0x20000 0x20000

-bodhi
===========================
Forum Wiki
bodhi's corner (buy bodhi a beer)

And the kernel bootargs should include:

mtdparts=armada-nand:5m(u-boot),5m@5m(kernel),5m@10m(uRamdisk),185m@15m(image.cfs),15m@200m(rescue_fw),20m@215m(config),10m@235m(reserve1),10m@245m(reserve2)

This will hopefully make the envs area write-enable. But be aware that it is not safe to play with them too much yet if you can't do kwboot. Being extra careful not to change anything significant that is related to preboot.

-bodhi
===========================
Forum Wiki
bodhi's corner (buy bodhi a beer)



Edited 1 time(s). Last edit at 01/28/2017 06:18PM by bodhi.

It's not working, so the address might be wrong:

root@wdebian:~# cat /etc/fw_env.config
# MTD device name       Device offset   Env. size       Flash sector size      Number of sectors
/dev/mtd0 0x00080000 0x20000 0x20000
root@wdebian:~# fw_printenv
Warning: Bad CRC, using default environment
bootargs=
bootcmd=bootp; setenv bootargs root=/dev/nfs nfsroot=${serverip}:${rootpath} ip=${ipaddr}:${serverip}:${gatewayip}:${netmask}:${hostname}::off; bootm
bootdelay=3
baudrate=115200
stdin=serial,cros-ec-keyb
stdout=serial,lcd
stderr=serial,lcd

Get permanent changes to u-boot is for me very low priority as I see no big benefit.

-------------------------------------------------------------------------
GitHub https://github.com/cschil

Hey,
I know many of you are not realy a friends of systemd, but may be you can help as well.

I checked some things, but I'm stucked with lvm and systemd:
At boot process systemd did not find partitons on the lvm.
As soon as debian is booted i can just mount them normal,..of cause a walk around is easy but I like to fix this.
On my desktop this config works perfect.

-------------------------------------------------------------------------
GitHub https://github.com/cschil

Carl,

> At boot process systemd did not find partitons on
> the lvm.
> As soon as debian is booted i can just mount them
> normal,..of cause a walk around is easy but I like
> to fix this.

This sounds like an initrd related problem (I assume you have one). Because that's where the volumes are mounted.

Have you tried to to update it:

update-initramfs -u

-bodhi
===========================
Forum Wiki
bodhi's corner (buy bodhi a beer)

You might be right, but as I'm forced to use a custom initrd(due to the locked uboot), update-initramfs -u is not working for me.
Just to make it clear the root partition is mount correct by initrd.
The "systemd-boot-process" does not find partitons on the lvm, partition direct on disk or raid are no problem.

cheers Carl

-------------------------------------------------------------------------
GitHub https://github.com/cschil

Carl,

Sounds like being able to read and write u-boot envs is something to do next. What is the DTS looks like? Eventhough you cannot use saveenv to change envs in u-boot, in Linux you can do that with the right DTS. If the kernel bootargs does not prevent it, then if all NAND mtds are defined in DTS, with the u-boot mtd being writable and the envs defined at correct location, then you can set the u-boot envs permanently. And that will allow you to boot everything from disk, with a normal Debian system.

Sorry I have not followed this thread so the questions sometime sound redundant ! i am just throwing out some feasible ideas.

-bodhi
===========================
Forum Wiki
bodhi's corner (buy bodhi a beer)

And of course, the safest thing to do is to try kwboot this box. If you can do that then everything else is easy.

-bodhi
===========================
Forum Wiki
bodhi's corner (buy bodhi a beer)

See if you can kwboot with this attached u-boot bin.

-bodhi
===========================
Forum Wiki
bodhi's corner (buy bodhi a beer)

Hi Carl,

I'm sorry, I can't help. I was stuck at the same problem, that was the reason I reinstalled with sysvinit-core.

bye
MM

@bodhi
kwboot is not working, either the version from the nas326 nor compiled from 2013.0.1-20014_T3.0p2

kwboot -d -b ./u-boot-a38x-RangerPeak_2014T3_PQ-nand-uart_1G.bin -B 115200 /dev/ttyUSB0 
Sending debug message. Please reboot the target...\
Sending boot image...
M - 1.73
Booting from NAND flashl

General initialization - Version: 1.0.0
Detected Device ID 6820
High speed PHY - Version: 2.0

Load WD GrandTeton Serdes Config:
board SerDes lanes topology details:
 | Lane #  | Speed |  Type       |
 --------------------------------
 |   0    |  06   |  SATA0	|
 |   1    |  05   |  PCIe0	|
 |   2    |  06   |  SATA1	|
 |   3    |  05   |  USB3 HOST1	|
 |   4    |  05   |  USB3 HOST0	|
 |   5    |  00   |  SGMII2	|
 --------------------------------
PCIe, Idx 0: detected no link
High speed PHY - Ended Successfully
DDR3 Training Sequence - Ver TIP-1.26.0
mvSysEnvGetTopologyUpdateInfo: TWSI Read failed
DDR3 Training Sequence - Switching XBAR Window to FastPath Window 
DDR3 Training Sequence - Ended Successfully
BootROM: Image checksum verification PASSED
l
 __   __                      _ _
|  \/  | __ _ _ ____   _____| | |
| |\/| |/ _` | '__\ \ / / _ \ | |
| |  | | (_| | |   \ V /  __/ | |
|_|  |_|\__,_|_|    \_/ \___|_|_|
         _   _     ____              _
        | | | |   | __ )  ___   ___ | |_ 
        | | | |___|  _ \ / _ \ / _ \| __| 
        | |_| |___| |_) | (_) | (_) | |_ 
         \___/    |____/ \___/ \___/ \__| 
 ** LOADER **


U-Boot 2013.01_v1.04 (Aug 20 2015 - 11:11:32) Marvell version: 2014_T3.0p6

mvBoardIdIndexGet(boardId) = 1
Board: GrandTeton DB6820
SoC:   MV88F6820 Rev A0
       running 2 CPUs
CPU:   ARM Cortex A9 MPCore (Rev 1) LE
       CPU 0
       CPU    @ 1332 [MHz]
       L2     @ 666 [MHz]
       TClock @ 200 [MHz]
       DDR    @ 666 [MHz]
       DDR 32 Bit Width, FastPath Memory Access, DLB Enabled, ECC Disabled
DRAM:  1 GiB

Map:   Code:			0x3fece000:0x3ff95bd4
       BSS:			0x3ffef254
       Stack:			0x3f9cdf20
       Heap:			0x3f9ce000:0x3fece000
raise: Signal # 8 caught
raise: Signal # 8 caught
       U-Boot Environment:	0x00000000:0x00080000 (NAND)

NAND:  ID: daad ,256 MiB
MMC:   mv_sdh: 0
PCI-e 0: Detected No Link.
USB2.0 0: Host Mode
USB3.0 0: Host Mode
USB3.0 1: Host Mode
Board configuration detected:
Creating 1 MTD partitions on "nand0":
0x00000f500000-0x00000ff00000 : "mtd=7"
UBI: attaching mtd1 to ubi0
UBI: physical eraseblock size:   131072 bytes (128 KiB)
UBI: logical eraseblock size:    126976 bytes
UBI: smallest flash I/O unit:    2048
UBI: VID header offset:          2048 (aligned 2048)
UBI: data offset:                4096
UBI: attached mtd1 to ubi0
UBI: MTD device name:            "mtd=7"
UBI: MTD device size:            10 MiB
UBI: number of good PEBs:        80
UBI: number of bad PEBs:         0
UBI: max. allowed volumes:       128
UBI: wear-leveling threshold:    4096
UBI: number of internal volumes: 1
UBI: number of user volumes:     1
UBI: available PEBs:             32
UBI: total number of reserved PEBs: 48
UBI: number of PEBs reserved for bad PEB handling: 2
UBI: max/mean erase counter: 5/2
UBIFS: mounted UBI device 0, volume 0, name "reserve2"
UBIFS: mounted read-only
UBIFS: file system size:   4063232 bytes (3968 KiB, 3 MiB, 32 LEBs)
UBIFS: journal size:       1015809 bytes (992 KiB, 0 MiB, 6 LEBs)
UBIFS: media format:       w4/r0 (latest is w4/r0)
UBIFS: default compressor: LZO
UBIFS: reserved for root:  200807 bytes (196 KiB)
Loading file '/mac_addr' to addr 0x02000000 with size 18 (0x00000012)...
Done
Set lan 2 WakeOnLan ok
  0 % [+xmodem: Protocol error

./a.out -p ./u-boot-a38x-RangerPeak_2014T3_PQ-nand-uart_1G.bin -B 115200 /dev/ttyUSB0
Usage: a.out -b <image> [ -p ] [ -t ] [-B <baud> ] <TTY>

  -b <image>: boot <image>
  -p: patch <image> to type 0x69 (uart boot)

  -t: mini terminal

  -B <baud>: set baud rate


./a.out -p -b ./u-boot-a38x-GrandTeton_2014T3_PQ-nand-uart.bin -B 115200 /dev/ttyUSB0
./u-boot-a38x-GrandTeton_2014T3_PQ-nand-uart.bin: Invalid image.

./a.out -p -b ./u-boot-a38x-RangerPeak_2014T3_PQ-nand-uart_1G.bin -B 115200 /dev/ttyUSB0
./u-boot-a38x-RangerPeak_2014T3_PQ-nand-uart_1G.bin: Invalid image.

./a.out  -b ./u-boot-a38x-RangerPeak_2014T3_PQ-nand-uart_1G.bin -B 115200 /dev/ttyUSB0
Sending boot message. Please reboot the target...|
Sending boot image...
  0 % [+++++++++++++++++xmodem: Bad message

@mm

I thought so :-|, I just don't get what is wrong.

-------------------------------------------------------------------------
GitHub https://github.com/cschil

@Carl

> ./a.out -b ./u-boot-a38x-RangerPeak_2014T3_PQ-nan
> d-uart_1G.bin -B 115200 /dev/ttyUSB0
> Sending boot message. Please reboot the target...|
> Sending boot image...
> 0 % [+++++++++++++++++xmodem: Bad message

This is actually a lot better than no response. When you see this xmodem bad message, up arow to recall the command and execute it again. Try repeating this several times until it handshakes.

If this works, please point me to the source code where you compile the image from!

-bodhi
===========================
Forum Wiki
bodhi's corner (buy bodhi a beer)

@ bodhi

Not working,... if you are nevertheless interested in the sources, its form the mcm-gpl package form wd.

-------------------------------------------------------------------------
GitHub https://github.com/cschil

Peacemaker,

> Not working,... if you are nevertheless interested
> in the sources, its form the mcm-gpl package form
> wd.

Strange! in my experience with booting kwboot, the xmodem bad message is a really good indicator that UART handshake is working, but out of sync. The source code most likely will not be useful in this case.

-bodhi
===========================
Forum Wiki
bodhi's corner (buy bodhi a beer)

Hi Carl,

would you please update your initrd in your repo? In the ready made image there is still the init execution bug with sysvinit. You fixed it already in the source but the image seems to be old.

I have created a how-to page that walks people through installing debian on the system, since some other people I know also want to use it.

Howto install debian jessie on the WDMC EX2 Ultra

Thanks in advance

bye
MM