Topic: [ mining os ] nvoc - page 348. (Read 418558 times)

It would prevent them from being able to mount the Windows partition and edit directly. I'd suggest you consider making the onebash file available by itself for download. People can edit on Windows/Mac/etc, and then use scp to copy it to the mining rig.

Setup would be - burn image to whatever. Once the machine starts - scp onebash, and go.

BTW - I only used SSD's because I have a bunch of 30G ones laying around, and I'm impatient and like faster boot times.

hi fullzero - i am repurposing my older AMD ZEC boards for nvoc.

Which image should I use compatible for the folloiwng board:

BIOSTAR RACING Z170GT7 LGA 1151 Intel Z170 HDMI USB 3.1 USB 3.0 ATX Intel Motherboard
https://www.newegg.com/Product/Product.aspx?Item=N82E16813138421


It has a 4 x 16x PCI slot motherboard meant for riserless implementation for 4 GPUs.

I intend to replace existing 4 x R9 Nanos with 4 x 1080ti

Please advise.

Thanks

fullzero.... any "dashboard" feature ?

if I have say... 10 nvidia zec rigs powered by nvoc.... it will be nice to see a simple dashboard to ensure all nvoc rigs are up and running.

maybe this is a monitoring issue rather than nvoc.... comments?

There's no real reason SSD and USB support can't be the same image. The issue that makes the SSD special is the Windows partition. I'd recommend removing that and allow have the users edit the file once the mining machines starts. Then, the image supports both USB and small SSDs.

You might also want to through this up on github. People can submit issues and track progress. Others, such as myself, can contribute. For instance, I'd be happy to build an ansible playbook so you can automate the build process.

I know that expect is not meant to be used within bash, and there are much cleaner ways to handle privilege escalation.  I want to avoid having oneBash call other scripts, keeping as much as possible within one bash script (oneBash).  It would probably be easier to make oneBash suid; by this do you mean:

sudo chmod 4750 '/media/m1/1263-A96E/oneBash'

or

sudo chmod u+s '/media/m1/1263-A96E/oneBash'

?

I am always open to suggestions for improvements.  I would also never claim to know everything about linux; which is good: because its fun to learn.  


[/quote]

No criticism directed - you've done something VERY useful to the community. The fact you did so with little Unix experience is that more impressive! I'm just trying to help.

Both of those commands do the same thing. Here's what you'd have to do:

1. sudo chown root /media/m1/1263-A96E/oneBash
2. sudo chmod u+s /media/m1/1263-A96E/oneBash
3. sudo chmod o+x /media/m1/1263-A96E/oneBash

Now onebash is owned by root and executes with all of it's privs, and anyone on the system can execute the script. After doing that, you should be able to just delete the expect code that provides the password to elevate privs. The down side of this is that the miners now all run as root.  To take care of that, chmod u+s to each mining program and make sure it is owned by the user m1 with chown m1 . Now, each of the miner programs should drop privs when they execute.

This approach allows you to keep one big script -

You should definitely confirm all of this. I'm in management now, and I'm not on the keyboard as much

For the benefit of others using a SSD, the windows partition is not automatically mounted. It is /dev/sda1 and it's type is MSDOS. You can mount it with the followng command: mount -t msdos /dev/sda1 /media/m1/1263-A96E

To have it automount at boot time, you'll have to edit the /etc/fstab file.  The easier option is just use dos2unix to remove the Windows' linefeeds from the file, and move it to a proper Unix partition. That's what I've done.

@fullzero DM me - there's no reason to be using expect here. It looks like expect is used just to handle privilege escalation for commands. You can easily just make the script suid, or call out smaller scripts that need the extra privs. I'll try and update the script when I can - the coming week is a bit crazy though.

Thanks for your work.
Would the ASUS Z270-F GAMING image work for Asus - PRIME Z270-A?
Or is it possible to have a walkthrough on how to build an image?
I haven't received them yet, can test next week.

EXCELLENT! Hopefully Gigabyte will release an updated bios that fixes that issue. I'll shoot then an email to see what they have to say about it. Glad to hear it's not just me though.

Mine is the gaming 3 and I'm using a G4400.

Thanks again for all the work, loving it. Now to see what I can push these cards to.

Also quick question since you have obviously used this MB before. GA-B250M

Have you ever had trouble with it booting? When I do an initial power up I have to hit ctrl-alt-del to reboot it to get the gigabyte splash screen to show up. After that it boots fine. And it reboots fine if I restart from the OS. When i have to hit ctrl-alt-del i can see a cursor underline flashing on the screen in the upper right quadrant??

I have never experienced anything like this before.

Figured out my issue. I had to set GEN 1 in the bios for my setup to work right. Now TDP is around 115-118w for all 4 cards in Nvidia-SMI and usuage is above 95% for all cards. I appreciate your help.
Getting just under 1200 H/s on my pool with 505 watts at the wall! Pretty happy with that.


Two issues I am having, one I solved just wanted to let you know what I ran into.

#1 - I tried running ETH and kept getting it error. It was rejecting my shares. It took me a minute but I figured it out. I was trying to use ethermine.org and it was rejecting me because of the address/worker name.

On line 483

ETHADDR="$ETH_ADDRESS/$ETH_WORKER"

ethermine.org wants this input as ETHADDR="$ETH_ADDRESS.$ETH_WORKER"
The . separating them rather than the /

I am not sure if this is the case for all ETH as thats the first time I have mined for it. Just a heads up, it worked great after I made that change.


#2 - I am using the 15.5 GB ADATA 3.0 drive that you said would be ok. Sometimes it wont boot, it goes to the boot loader and shows a purple screen and never loads. Sometimes I get a black screen with
/dev/sda2:recovering journal
/dev/sda2: clean, 314012/901120 files, 2478157/3584000 blocks.
If i let it sit there it takes several minutes to finally boot up but seems to work ok after that. Sometimes it boots fine. Should I try a different Drive? Or is this another issue.

So I was finally able to get nvOC going. A few hints for those that use Macs.

1. unzip will error out on the zip file.
2. tar will extract the file but it will be corrupt.

I just unzipped on a PC, and burned there.

As a side note, the images do work on 30GB SSD's as well.

A couple of asks: please add openssh-server, and screen. The addition of both of those allows us old command line guys to still use it.

Finally - where is onebash on the file system? A find from '/' with wild cards couldn't find it.

Not sure if this has been asked but how does this os support updating mining software?

I did that command and it says assigned 0.  But nvidia-smi shows P2 still.

These are EVGA 1060 sc 3gb.
I am using 1x-16x adapters v006 blue, USB from deepinthemines.
I am powering then directly with molex from my PSU (no adapters) and I only have two risers per modular molrx cable from my PSU.

Ok I was able to test some on my 1060 system today.

I seem to get different results each time I restart leading to away from a potential hardware issue??

This restart has been my best and it was a restart from the OS using the restart under the gear icon. The others have been shutdowns and startup from the off position.

All cards are choosing solver 0.

It seems to be one card with lower power draw and and gpu usage with the nvidia-smi command. Also all cards are getting fans set properly and it shows that the overclocks are applied (any way to check from guake?) All cards are running in P2? Is there a way to force p0 on this? I know my other 1060 in win 8 won't apply the memory properly in P2 and I can't force p0 like I did on my Maxwell cards.

(First row is tdp, 2nd is going usage)

Start1 (shut down and swapped riser on 3)
GPU 0 110
GPU 1 112
GPU 2 60
GPU 3 115

Start2
Tdp
119,120,118
112,109 ,108
107,109,98
114,113,114
Usuage
97,98,96
93,95,93
85,84,87
100,100,100


Start 3
Tdp
73,65,64
109,113,110
96,98,106
117,119,118

GPU
46,43
87,92
73,75
98,99


Reboot (start4)
Tdp
110,118,110
105,114,114
100,95,104
117,118,117

GPU
88,90,82
94,96,97
90,93,92
99,99,100


Performance is ok. Not as good as my stand alone 1060 I have running (same card). It gets 305 sol/s at 135/400 mem

I'm getting 271,286,288,305 now on these with 135/300


I appreciate any info you can give me on adjustments to make or things to try.

Hello, I'm quite new to mining, but I wanted to delve into it, so after some research and hardware shopping I ended up with a Z170A GAMING PRO CARBON https://us.msi.com/Motherboard/Z170A-GAMING-PRO-CARBON.html#hero-overview

Does your OS work with it? I mean is it like the normal Z170A or is there something else I'd need to watch out for? I still didn't get the cards, but got my eyes on 7x1070 Gigabyte Windforce. Would you recommend something else perhaps, or do you believe it's possible to make them work?

I'm aware that these Z170A series won't let me OC the last GPU, but I don't really mind as long as I can make them all work.

Any info is greatly appreciated since I'm quite new to this.

I am running all of them stock at the moment and getting them to stabilize on this mobo and nvoc.
I will be playing around with the OC settings later (currently default 100/100)

thanks fullzero for the explanation.

After testing the builds for Gaming M5 and this Z270-A Pro, the Z270 is best for my NVIDIA rigs.

I will keep the M5 for AMD builds since using all 7 slots on smOS is not an issue.

So far running 3 x 1070s, 1 x 1080ti and 1 x 980ti on your os renting out my hash at NH marketplace without any issues.

hi fullzero,

I am happy using the Z270-A Pro image build and my copy of the nvoc is on a USB3 Sandisk 32GB thumbdrive.

All is well, hashing ZEC at Nicehash rental market on my dev rig -- with 3 x 1070, 1 x 1080ti and 1 x 980ti

I tried changing from USB thumbdrive to a normal SSD SATA drive (Kingston 120GB SSDNOW).

Using HDDRAW utility, cloning completed successfully and I changed the necessary parameters on the onebash file.

Now booted from SSD and seems to work nicely until these error messages keeps on looping in terminal.

===
.
.
.
spawn sudo dos2unix /media/m1/1263-A96E/oneBash
[sudo] password for m1: spawn sudo ldconfig /usr/local/cuda/lib64
[sudo] password for m1: bash: /media/m1/1263-A96E/oneBash: No such file or directory
spawn sudo dos2unix /media/m1/1263-A96E/oneBash
[sudo] password for m1: spawn sudo ldconfig /usr/local/cuda/lib64
[sudo] password for m1: bash: /media/m1/1263-A96E/oneBash: No such file or directory
spawn sudo dos2unix /media/m1/1263-A96E/oneBash
[sudo] password for m1: spawn sudo ldconfig /usr/local/cuda/lib64
[sudo] password for m1: spawn sudo dos2unix /media/m1/1263-A96E/oneBash
[sudo] password for m1:
.
.
.

===

Looks like some directories maybe pointed differently when using SSD vs USB stick -- any ideas?

Thanks

It is possible to use teamviewer in this os ?
did he auto recognize USB WIFI adaptator ?

i got 4x1060 6gb + 2x1060 3gb and have some gpuclock problem in windows, nvOC can fix this  ? (https://bitcointalksearch.org/topic/6x1060-gtx-problem-with-gpu-clock-1933393)

thanks!

edit: i didnt have dummyplug, can i use nvOC anyway?

These are Gigabyte G1 1070's.  When I manually run the command it tells me that the acceptable number is 90w to 200w.  So I manually entered 90 and that took effect and brought the wattage down, now tweaking it upwards to find a happy medium.  Will add that to oneBash and boot and and see if it is happier.