Topic: [ mining os ] nvoc - page 54. (Read 418529 times)

@papampi:

the /dev/watchdog* device is created by the module (one device per loaded module is created). /etc/default/watchdog is where you specify the watchdog module to load at watchdog start.

both /etc/init.d/watchdog and the systemctl service look there and load it.

What happens at boot when you use rc.local is that the "start" command gets stuck and never returns. After some time you can do stop and start again and it works then, but not immediately after boot, which is "puzzling" me.

I tried to add a delay (120sec) in rc.local and it won't start watchdog regardless. So I'm trying variations here atm.

I was able to install/update watchdog with removing /etc/inint.d/gpu and then added it back, but your solution is definitely better to fix that script.
Then it got stuck at start at boot time too. Fixing the watchdog.service did not fix it either, may be we add it to rc.local so it start at boot?

Also I read there should be a watchdog in /dev, but after manual start it does not show up in there too.


Edit:
remove/purged then installed watchdog again, now /dev/watchdog show up with manual start

Edited /etc/init.d/watchdog:

And still wont start at boot

watchdog install fails because /etc/init.d/gpu by th00bear does not complain to lsb requirements. You need to add the following to the header (from line 3 onwards):

Also there's an unbalanced quote in /lib/systemd/system/watchdog.service. The execstartpre line should have a single quote at the end ('):
after that apt-get install watchdog (or apt-get -f install to fix it).

then

#systemctl start watchdog

then check that it's running with

#systemctl status watchdog

and finally enable at boot with

#systemctl enable watchdog

and don't forget to fine-tune with /etc/watchdog.conf
also a hardware driver needs to be specified in /etc/default/watchdog.
The drivers are in /lib/modules/$(uname -r)/kernel/drivers/watchdog
Usually (proper) mainboard manuals have the type of watchdog installed. On intel hardware, usually iTCO_wdt works.
You can use the softdog.ko module, but it's not as effective

After all this, I tried rebooting my nvoc rig and watchdog start is stuck in starting state. If I stop and start it by hand on the console, it will start but it won't start by itself. Perhaps there's an issue related to nvoc or not (my other ubuntu boxes start normally).

I tried looking for a fix for about an hour now, and still nogo. :/

EDIT: RE: Blacklisted modules:
There's no need to remove blacklisted modules. /etc/default/watchdog contains the module name to load and watchdog start will load it regardless of blacklist. This is something that cannot be automated because there's not reliable way to know if a hardware watchdog module is present. Some modules load and assume they work, creating the /dev/watchdog* devices. Other will check and fail. It's a mess in general.

Updated claymore to 11.0
Will add more options in 1bash and 0miner for 11.0 dual new blake2s and keccak coins

Thanks mate
I know how it works, and I can load watchdog manually but couldn't get it to auto start on system boot.
Also no idea which modules to remove from blacklist .

@papampi re: hardware watchdog

Sorry for being brief here, but I'll try to give you some pointers.

you should look at ubuntu package watchdog and the watchdogd daemon. When implemented correctly (I hope it is) with the help of the kernel is should work as hardware watchdog timer (on mobos where the hardware watchdog can be enabled in bios or where it's enabled by default).

It works by starting watchdogd which is a daemon that will reset a watchdog (hardware bit in the cpu/mobo) once a minute. Once started, if watchdogd does not reset the timer, a hardware reset occurs.

Sometimes it's possible that there's a hardware issue with the gpus where a simple reset will not be good and you might need a physical power on-off.

hth, bytiges.

I dont have any amd rigs and dont use rxOC
But I think the set of compile miners has same path as rxOC, but others like ethminer, dstm, bminer, ... may have different path in rxOC that should be changed in 3main/1bash or move them to the appropriate location.

Yes, and thats what I do
Just set your coins with normal 17xxx port and add them to wtm_coins

Edit:
Normal port 20xxx should be used, 17xxx is auto algo switch.

nvOC v0019-2.0 Community Release - Miner Update Script

@papampi: can this work also with latest rxOC?

Is there a way to use MPH pool auto exchange feature using papampi WTM auto switching?  I have not tried this but i assume if i point all the different coins to the correct 72XXX ports that it would work?  Has anyone tried this or have any ideas on how to implement this in the code?

Be carefull in Mining VIVO coin, i mine it but it seems that his blockchain is pericliting or desynchronize on severals pools.

If you are minning VIVO, Please let US a Message to tell everyone where is the good blockchain.

Because of this désynchonized VIVO blockchain, i prbably loose 0.5 Vivo ...even if bulwarkpool.com confims all blocks it mine on the wrong blockchain. But i think it's not a good solution to resolve.

I just saw thant hashraffinery seems to have the same problem but i'm not sure because they don't precise tha coins they are desynchronized

Please tell everyone what pool your are using.

whattomine has just added Innova Coin to the possible coin selection in its comprarative list.
If you want to add it to your WTM_SWITCHER list, it's still easy:

To add Innova (INN) (Neoscrypt) to the selection of cryptocurrency in 1bash script file and therefore also in the WTM switch:

1 / Find TZC Variables in 1bash file (just below the VIVO variables, you will copy / paste as a template Under TZC lines):



Add this below to the 0miner script file (in the same way below the command to launch the TZC ccminer and you can copy TZC but be careful to replace TZC_ by INN_ in the whole command):

At the end of 1 bash file, you can also add INN coin in ALGO_SPECIFIC_OC section, Like This :


Don't mine directly to Cryptopia Wallet Address because if Maintenance on INN Blockchain, cryptopia will be slow to reput online a wallet, and you risk to mine in the void for all  the time cryptopia INN wallet could be offline or désynchronised of official Blockchain.
Download their smaller wallet. There is a portable wallet for windows users.
https://innovacoin.info/#wallets

To consult your pool balance :
https://www.unimining.net/?address=INN_Wallet Address

You can mine INN on other pools :
http://www.zpool.ca/site/mining => 2% pool fee
http://pool.hashrefinery.com/ => 1.25% pool fee
https://altminer.net/ => 0.9% fee
https://www.lycheebit.com/ => 1% fee

More pools on their webpage, I do not have all consulted :
https://innovacoin.info/pools/

So I have now tried the standard MPH_EQUIHASH switcher, MPH_ETHASH switcher and the SALFTER_MPH_PROFIT switchers from nvOC.
I have not yet tried the WTM switcher. That's next. 
I auto_exchange to ZEC on the MPH pool and that also works very well.
I have noticed from my test runs that I gain very little additional ZEC, if any additional, using these switchers than just pointing all my GPUs directly at ZEC.

It seems that the switching occurs often and is based on something like instantaneous profit margin on an exchange.
If that is what is happening, there is very little chance to:
have the miners switch
get the coin mined
into the wallet
confirmed
onto the exchange
and the exchange made before the switching equation is no longer valid
 
And frequently the switch formula will select a coin that only has the highest margin for a very short time, so if you make the time between profit checks too long your wasting time mining a suddenly low-profit coin.
It would seem that auto switching would be more advantageous if one had many rigs that could start the profit checking timeout in a staggered manner.
That way it is more likely that only one rig would be stuck mining a short-time-most-profitable coin.

Can someone help me with what I am missing in my thinking?
My farm is not that large so the benefits may be so small that I can't measure them.
The work done on the switchers is fantastic but I don't yet see the benefit.
Thanks.

Glad so many liked it.
Good idea for "All" option, Added.

I'll look into the Pi, would be great to set it up to ping and then restart if needed. Of course, finding the issue of the freeze is my priority and I think I have an idea now.

I have two rigs both mining ZCL on Supernova and this has happened twice that both rigs will freeze within a minute of eachother. I saw the first time while being connected via teamviewer that the server "disconnects" briefly within the miner and that watchdog will then re-start the 3main, get hung up and freeze. I assume this is what happened again. I would point to a hardware issue, but both separately going down at the same time with no telegram errors? Too much of a coincidence. I don't have any of the cards overclocked at this point, both systems were up for almost two days before the last freeze.

Anything you can think of in this regard?

I am in the process of updating the miners and re-compiling. Thanks!

This is really amazing pap, it will help users who are not good at compiling or having issues with compiled miners.Will share it with the discord community too.

Please don't mind about people writing in bad tone, they don't understand the effort and time you are putting into this and there are really nice people up in here writing good things, they are the ones we really care about. Appreciate the work so much.

Like you say : " keep calm and carry on mining"

Yes, disabling groestl in WTM URL that you put in 1bash was also my solution that i proposed in my post here :
https://bitcointalksearch.org/user/cryptatometrader44-1157121

But i think it would be possible to filter multi-algo XVG coins like nicehash multi-algo added.
simply add one or two lines of code in forked python script papampi.

Here is what I propose :
Search this section in the python script :


with adding XVG in WTM_SWITCHER python script

Of course, you will also have to modify 0miner minning command script file accordingly and also 1bash config file to update  parameters of coin XVG and OC_SPECIFIC_ALGO section too.

I think if you're familiar with code you'll be able to do it all.
You just have to take model on added nicehash-algo coins for other scripts files.

This is a propsition that I have not tested myself because I also think, as damnmad, that groestl is not profitable and therefore it is easier to disable it.

I have currently disabled the minning of XVG because of the sharp fall of recent days. Everything that was below 50,000 satochis was removed from my list of coin in WTM. So, suddenly I found myself mining ETH several times in recent days with using of WTM profit switch.

I just ran the script and had it re-compile all 8 miners as a test.
I did not sit and watch all 8 compile but it completed "successfully" in less than 20 minutes on my lowly G3220.
I would say it works great.
Thanks again for the effort.

Yep, that's the best way going forward, we never expected this sort of deadlock, like a coin using multiple algos.

There should be a way of dealing with it but what you suggested is the easiest way for now.