Smartcoin Linux mining administration. [MULTI-MACHINE SUPPORT NOW IN!]

[jondecker76]

Anybody test this on LinuxCoin yet?
I do have one report so far from a LinuxCoin user that they get aticonfig errors if they try to start it remotely in a terminal, but if they start it from a terminal running in X its fine, after which point they can ssh in and control it like they should.

Just looking to see if this is a common thing for LinuxCoin users, I'd like to get this fixed ASAP so any information will be very helpful

[sharky112065]

Sharky,

I understand that a lot of people will want a web front end, but again, I created this specifically for those that wanted/needed something easily controllable over ssh.
I understand that its not going to be for everyone, and that's ok.
Another thing to consider here is security.  Do you really want your miners controllable over a web page?  What if I did a web page, a hacker got by security and changed all your pool workers, or peeked at your passwords.  For these reasons (and many others such as encrypted communication), there are some of us that will use only use shh to manage our miners.  Exposing your machine to http/https where there is the possibility that a user has a wallet.dat on that machine (or even the same network) is a bad idea, IMO.  I also don't want to make software that a user might download with the false assumption that I'm a security expert (likewise I can't expect all users to be security experts themselves). By entrusting this security to the ssh client/server, I eliminate problems on both sides of this equation


Don't worry, a lot of effort will go into making it reasonably easy to control multiple machines from one session, that is one of the main goals of this project. 

I think you misunderstood me. I only want to view status information on the web page, not make changes. Viewing that one page from either here at my home, or remotely, I could see if there is a problem and then ssh in to fix it. Sure if all you have is one computer running then it is really easy to ssh in to get a quick status, but if you have like 10 computers running, then it is a real mess.

I think people with more than one computer would be really interested and thus more likely to donate to you. For example. I think it is useful, so I would probably donate a few coins. If I could monitor by a web page on one of my servers, I would be be donating a lot more because you would have just reduced my workload.

I'm not saying you have to write the actual web page, just give us the ability to pull the data.

And as far as wallet.dat being exposed, I do not have the bitcoin client and or wallet on any computer that I do web browsing on or any mining computers.

[jondecker76]

sharky,

thanks for clearing that up.
The ability to pull data through a simple API is planned, so that part will definitely be realized.
I'm also experimenting with netcat for a super simple way to check in on things over a web browser if you want to (though this will be turned off by default for security reasons)

A few other things that may also help neutralize your concerns:
-Email support is coming soon!  You'll be able to have hourly/daily automatic reports emailed, as well as options to automatically send you emails when there are problems with your miners.

-Multi-machine support is also coming. This means, you decide on one of your machines to be the "master" smartcoin node.  You run smartcoin on the master node, and all child nodes will all be controllable from the master node (each machine will have its own tab in the screen session, and options in the control screen such as "Change Profile" will start asking you which machine to change profiles on) - so you will only need to ssh into the master node to be able to control all 10 machines in your example.  Since your miners get started on the child nodes in their own screen sessions, they will keep running even if the master node goes down for some reason.

[BlueHelix]

I reinstalled and this is automatic detected:

./smartcoin_devices.py
0   GPU[0]   0   gpu
1   GPU[1]   0   gpu
2   GPU[2]   0   gpu
3   GPU[3]   0   gpu
4   GPU[4]   0   gpu

The launch string I use is the same only difference was UPPER case in arguments.

Is it possible to exclude one Device because I have a stability issue .

[Clipse]

I reinstalled and this is automatic detected:

./smartcoin_devices.py
0   GPU[0]   0   gpu
1   GPU[1]   0   gpu
2   GPU[2]   0   gpu
3   GPU[3]   0   gpu
4   GPU[4]   0   gpu

The launch string I use is the same only difference was UPPER case in arguments.

Is it possible to exclude one Device because I have a stability issue .

The extra device is your cpu being detected as openCL device, just manually remove it.

For safety I removed all devices and added the proper devices manually, its better cause on my one machine it seemed to list my cpu under device=3 where most of the times it would automaticly list as the last device.

[shotgun]

Why not use sqlite? It seems much more appropriate for this project.

I totally agree. MySQL is awesome for certain apps (I'm actually a MySQL DBA for a large video game company) but SQLite would be much better for this.

Additionally, I have the following issue when trying to install rev-100

Code:

./smartcoin_install.sh: line 18: RunSQL: command not found
./smartcoin_install.sh: line 19: Field: command not found
SmartCoin requires root permissions to install dependencies, create SymLinks and set up the database.

I checked the installer bash file and the RunSQL function doesn't exist anywhere. How should I go about installing the app if it can't write to the DB?

[shotgun]

A web interface is crucial if you want people using/donating that have more than one Linux box up and running.
Say I have 10 Linux boxes up and running but no ability to monitor them on one screen, it would be a real pain in the A**.
I would like the ability to go to one web page on my web server and see if every thing is going OK.

Agreed - I'm writing a webapp that will connect to configured pools and report on your workers. You can see some basic stuff here, this is only a day of coding though: http://minerwatch.strangl3r.com/

I'm planning on having drill-down graph support so you can look at historical information about all of your workers. I think the best way to manage multi-box workers is to name them according to their hostnames and GPU. ie: strangler.ultra40m1-01 is my Sun Ultra40-M1 workstation running one miner, if there were additional cards in the box they would be strangler.ultra40m1-02, strangler.ultra40m1-03, etc.

Anyway, I'm looking forward to trying out SmartCoin once the RunSQL issue is addressed. Seems a useful app for running on individual boxes to manage multiple GPUs and load balancing/process monitoring. Then you can use my app (or miningmonitor.com or similar) to see your array of rigs all at once.

[jondecker76]

BlueHex:
Yes, it is possible to exclude devices from the Automatic profile.  Configure Devices -> Edit.  There will be a choice to allow it to be available to the automatic profile, type n for "no" (plus, there is also a disable option, which would work as well). I also think that you should review which devices are present, and delete any not matching what you know you have. (FYI, your devices are as detected by PyOpenCL.  I will look into this problem eventually, hopefully I'll come up with a better way to auto-detect in the near future...)

shotgun:
Smartcoin does use sqlite now, I switched database backends on the same day the suggestion was made
Also, you mentioned installing r100.  As of right now, the current revision is r200 - things are evolving rapidly so please make sure you have the latest version of the installer!
The RunSQL error is due to smartcoin_ops.sh apparently not being included for some reason.  What linux distro and version are you running?
Currently, smartcoin relies on the $HOME environmental variable for determining paths.  Are you running smartcoin_install.sh as root (you shouldn't, as it will make the $HOME variable point to /root - when the superuser permissions are needed, you will be prompted for a password.
What does it say if you type

Code:

echo $HOME

in the terminal?

[gnaget]

Great work, by far this looks like the best linux based miner out there.  I tried to get it working on one machine in my farm, and I ran into an issue.  I skimmed the thread here, and I don't see anyone else with the problem so maybe I am doing something wrong. 

I am trying to add a solo pool, using my server's ip address as the host.  When I use 192.168.1.2 I get a sql error "invalid syntax near .1".  When I try using my hostname (server.pedanticstoner.com), I get another syntax error.

[shotgun]

BlueHex:
Yes, it is possible to exclude devices from the Automatic profile.  Configure Devices -> Edit.  There will be a choice to allow it to be available to the automatic profile, type n for "no" (plus, there is also a disable option, which would work as well). I also think that you should review which devices are present, and delete any not matching what you know you have. (FYI, your devices are as detected by PyOpenCL.  I will look into this problem eventually, hopefully I'll come up with a better way to auto-detect in the near future...)

shotgun:
Smartcoin does use sqlite now, I switched database backends on the same day the suggestion was made
Also, you mentioned installing r100.  As of right now, the current revision is r200 - things are evolving rapidly so please make sure you have the latest version of the installer!
The RunSQL error is due to smartcoin_ops.sh apparently not being included for some reason.  What linux distro and version are you running?
Currently, smartcoin relies on the $HOME environmental variable for determining paths.  Are you running smartcoin_install.sh as root (you shouldn't, as it will make the $HOME variable point to /root - when the superuser permissions are needed, you will be prompted for a password.
What does it say if you type

Code:

echo $HOME

in the terminal?

I'm running r200 - I guess that was a slip of the keyboard. I tried installing when I was logged in under my usual user account "user", then did a "sudo su", then ran the installer. I guess I'm confused about the SQLite vs MySQL because the instructions on the 4th page say to install mysql server and not sqlite, and there is no README with the checkout or any instructions on the google code wiki. Is there a more recent set of instructions you can point me to, as I was going off of the ones on page 4 from the beta release post.

I'm running LinuxCoin btw.

[shotgun]

I added "source smartcoin_ops.sh" to the installer on the first empty line and that fixed the RunSQL error. Then I did apt-get install sqlite3 and it's working now (the installer anyway).

[jondecker76]

shotgun:
Oh geez, i just noticed what I did..

There was already an include of smartcoin_ops.sh, but I had made a few new functions recently and it got bumped down (to line 38) - bash can be picky about defining functions before using them, so I think thats what caused the error.

Just comitted the change, moving the include back to the top and comitted it to svn

[jondecker76]

revision r202 is up!
Fixed the problem where CPU was auto-detected as a GPU on some installs.  New version uses the pyOpenCL device_type to ensure only GPU devices are added!

[jondecker76]

Ok, made another discovery....

A generous LinuxCoin user gave me SSH access to one of his boxes so I could poke around.
LinuxCoin installs the ATI SDK to /opt (which isn't a problem),

But its $LD_LIBRARY_PATH is never set. This is why LinuxCoin users can't run aticonfig commands remotely.  I'm going to contact the LinuxCoin author and point this out.  

For now, I advise LinuxCoin users to perform this step BEFORE running/installing smartcoin:

Temporary version (must be done after each reboot)

Code:

export LD_LIBRARY_PATH=/opt/AMD-APP-SDK-v2.4-lnx64/lib/x86_64:$LD_LIBRARY_PATH

(64 bit)

or

Code:

export LD_LIBRARY_PATH=/opt/AMD-APP-SDK-v2.4-lnx32/lib/x86:$LD_LIBRARY_PATH

32 bit



Permanent version: (you won't have to mess with it again!)

Code:

echo 'export LD_LIBRARY_PATH=/opt/AMD-APP-SDK-v2.4-lnx64/lib/x86_64:${LD_LIBRARY_PATH}' >> ~/.bashrc
source ~/.bashrc

64 bit

Code:

echo 'export LD_LIBRARY_PATH=/opt/AMD-APP-SDK-v2.4-lnx32/lib/x86:${LD_LIBRARY_PATH}' >> ~/.bashrc
source ~/.bashrc

32 bit


Between the recently committed code, and this...  Many more installs should be working now

[shotgun]

I've setup my miner (Diablo) and my workers, and my pools. But it just sits there doing nothing saying that Slush is down. Something is wrong but I can't tell what. The smartcoin.log is basically useless, and doesn't tell me what it's doing other than "We are running in AUTO" all the time. Can you add a debug option or at least have the log file tell me when a miner is running and with what options? I have no way to troubleshoot what is happening. Of course Diablo runs fine when I run manually, and LD is set to the following in my bashrc file: LD_LIBRARY_PATH=/opt/AMD-APP-SDK-v2.4-lnx64/lib/x86_64

[jondecker76]

Shotgun:
For troubleshooting, I recommend disconnecting from smartcoin (either via the control screen, or ctrl+a d), then from the terminal:

Code:

screen -r miner

This will connect you to the screen session where all the miners are running.
You can use ctrl+a p and ctrl+a n  for previous and next, so you can look at each miner and see what the problem is.

I've never used or tried Diablo yet - Do you care to share your launch string?
(or better yet, post the result of the following command here)

Code:

cd ~/smartcoin
sqlite3 smartcoin.db "SELECT * FROM miner"

[jaebird]

Anybody test this on LinuxCoin yet?
I do have one report so far from a LinuxCoin user that they get aticonfig errors if they try to start it remotely in a terminal, but if they start it from a terminal running in X its fine, after which point they can ssh in and control it like they should.

Just looking to see if this is a common thing for LinuxCoin users, I'd like to get this fixed ASAP so any information will be very helpful

Yes it works great. LinuxCoin is setup to automatically log in to an X server on card 0. Unfortunately it is still a requirement to have X running to use OpenCL, so this is a good default. If you want to do any aticonfig commands you must have first set the environment var: "export DISPLAY=:0.0"

I have an overdrive.sh script that looks like this:

export DISPLAY=:0.0
aticonfig --od-enable
AMDOverdriveCtrl -b -i 0 overdrive_mining.ovdr
AMDOverdriveCtrl -b -i 3 overdrive_mining.ovdr
AMDOverdriveCtrl -b -i 6 overdrive_mining.ovdr
AMDOverdriveCtrl -b -i 9 overdrive_mining.ovdr

This sets up the fan speed and overclock of the cards. Then I simply run smartcoin and it is off to the races.

My plan is to have this done at startup in the future...

Cheers,
jaebird

[EnzoMatrix]

What is the best way to setup SmartCoin to allow me to focus my efforts on one pool server then failover to a different pool such as

Main Server : uscentral.btcguild.com
Failover Server : uswest.btcguild.com
Failover Server : useast.btcguild.com

as opposed to dividing my efforts across the pool servers

[shotgun]

Shotgun:
For troubleshooting, I recommend disconnecting from smartcoin (either via the control screen, or ctrl+a d), then from the terminal:

Code:

screen -r miner

This will connect you to the screen session where all the miners are running.
You can use ctrl+a p and ctrl+a n  for previous and next, so you can look at each miner and see what the problem is.

I've never used or tried Diablo yet - Do you care to share your launch string?
(or better yet, post the result of the following command here)

Code:

cd ~/smartcoin
sqlite3 smartcoin.db "SELECT * FROM miner"

I just switched to Phoenix but I can give you a sample Diablo launch string that I've been using:
./DiabloMiner-Linux.sh --user=$USER --pass=$PASS --host=$URL -v $VECTOR -w $WORKER -D $DEVICES

[Alan Lupton]

Hi, my monitor looks like this:

Code:

smartcoin Management System  Do 30. Jun 07:02:51 CEST 2011
--------------------------------------------------------------------------------
Host: localhost
GPU[0]: Temp: load:
GPU[1]: Temp: load:
GPU[2]: Temp: load:
GPU[3]: Temp: load:
CPU Load : 6,15%

Profile: Automatic
--------BTCGuild--------
GPU[0]: [217.01 Mhash/sec] [114 Accepted] [2 Rejected] [RPC (+LP)]
GPU[1]: [242.75 Mhash/sec] [140 Accepted] [1 Rejected] [RPC (+LP)]
GPU[2]: [283.62 Mhash/sec] [173 Accepted] [0 Rejected] [RPC (+LP)]
GPU[3]: [283.92 Mhash/sec] [193 Accepted] [2 Rejected] [RPC (+LP)]
Total : [1027.30 MHash/sec] [620 Accepted] [5 Rejected]

Grand Total: [1027.30 Mhash/sec] [620 Accepted] [5 Rejected] [.806% Rejected]


[ jenny ][            0-$ control  (1*$localhost)            ][ 06/30/11  7:02 ]

It doesn't show load and temp and you can easily see, though I can find out temperature via commandline using "DISPLAY=:0 sudo aticonfig --odgt -adapter=all". How to fix this?


EDIT: Could you also add a feature to specify a "max temp", so when max temp is reached the worker will shutdown until max temp - 10% is reached?