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

[BlueHelix]

GPU[X] problem gone

but Iam missing the miners-TAb after install and added workers

[jondecker76]

gnaget:
  Please do an svn update, then reload the script.  I found the error with the add pool, and it works now as it should. Thanks for the report!

shotgun:
  Also do an svn update and restart.  Miner launch attempts are now logged in ~/smartcoin.log (and I got rid of the annoying "We are running in AUTO..." log message)
Regarding Diablo, I have never used it before, but you just need to add the special tags while defining it in the database:

Code:

DiabloMiner-Linux.sh --user=<#user#> --pass=<#pass#> --host=
<#server#>  -D <#device#>

note that you can also use <#port#> to specify a port, but I'm not familiar with diablo's usage.  What's diablo's -w worker switch for?  Can you give me an example launch string fill full parameters filled out (with fictitious pool/worker information if you want)


EnzoMatrix :
  Failover isn't implemented yet.  Once I can get all the bugs stabilized with what is currently here, then I will work on adding the failover system.  In the future, it will definitely be possible to send all of your power to a single pool, which will fail over to a different pool on disconnection timeout.

[jondecker76]

BlueHelix:

There should only be 2 tabs in the screen session - 0 Control and 1 localhost
The miners are in a separate session, and can be found by disconnecting from smartcoin and running

Code:

screen -r miner

[jondecker76]

Committed a few more updates.

Added a new <#path#> tag that can be used inside the launch string.  I've been finding that in LinuxCoin installs, the miners are stored in /opt, so they either have to be run with sudo, or by launching them in their own interpreter (launching phoenix.py with python for example).  Also, the new default launch string for phoenix is now:

Code:

python <#path#>phoenix.py -v -u http://<#user#>:<#pass#>@<#server#>:<#port#>/ -k $knl device=<#device#> worksize=128 vectors aggression=11 bfi_int fastloop=false

(notice the python <#path#> addition in the beginning)

The old launch string will still work, but i recommend everyone do an svn update, and edit their miner's launch path on the control tab to match the new default.  This fixes another barrier for LinuxCoin users


Other news:
I know the documentation and installation instructions are aweful. I'm going to put some time into cleaning this stuff up and organizing it a bit more today.

Also, while poclbm and diablo do work with smartcoin, they have different output than phoenix and so the status tab will not show correct information unless you are using phoenix.  For now, I disabled poclbm autodetection in the installer until I get a chance to do the routines for the status tab to display correctly with other miners.

[Alan Lupton]

Hi, could you add something like

"0) back to Main Menu"

for all subsettings like "Edit Settings" or "Configure Workers"? Would make life much easier. I guess that requires a routine to first wirte changes made to a tempfile or keep them in RAM and at the end confirm those settings by writing them to file.

[jondecker76]

Alan -
Yes, I badly need to add exit clauses throughout the code. Its on my to-do list, hopefully I'll get a chance to do it soon. Until then,  you can alwasy ctrl+c, then hit o and you'll be back at the main menu.

[Alan Lupton]

Thanks,

I sent you a Bitcoin, I really like your attitude here!

If you have time, please don't forget to answer to post #185 http://forum.bitcoin.org/index.php?topic=16548.msg305278#msg305278

Much appreciated!

Cheers
Alan

[jondecker76]

new update is available in the svn repo
Adds preliminary support for exiting back to the main menu in the add/edit/delete screens with "x".  I still need to add exit clauses to the submenus, but this is a good start.

[jondecker76]

Alan-
I did a recent update that may fix your missing aticonfig stats (its been a little perplexing - on some installs, using DISPLAY=:0 makes it work, on other installs it causes it not to work. I'm still looking into it, but an svn update may get this to work for you (you will have to at minimum reload the status screen with ctrl+c o after the svn update to see if the changes worked)

Regarding linking mining to a maximum GPU temperature, its definitely possible.  In fact, yesterday I was working on a routine that locks all GPU fans in at 100% (future option - I found that on systems with many cards crammed together, running the fan higher on the cooler gpu's had a dramatic effect on helping cool the hottest ones)  That being said, I do have plans in the future for this type of thing. It may be a little while, as I want to finish implementing some of the things that should already be done, and fixing the remaining issues on some installs (I'm still in amazement at all the different configurations people use, getting everything to work with all of them is quite a challenge!)

Thanks for the donation, its greatly appreciated!

[Alan Lupton]

Well, I updated as you said and I get several dozens of the same error message:

Code:

Maximum number of clients reachedMaximum number of clients reachedMaximum number of clients reachedMaximum number of clients reachedMaximum number of clients reachedMaximum number of clients reachedMaximum number of clients reachedMaximum number of clients reachedMaximum number of clients reachedMaximum number of clients reached

Btw, where is the config file so I can save it and spare the process of configuring workers, miners etc. when doing an update?

[jondecker76]

Alan - it looks like you're running either a pretty old version, or the version from the trunk branch.
Please install a fresh copy from the beta1 branch.

The only thing you will need to save, is the smartcoin.db file.  However, I just completed testing of the new update system, which I'm hoping will go live tomorrow.  This will let you choose between stable updates, and experimental updates, and do everything automatically for you. It will be selectable from the config screen

[Alan Lupton]

No, I'm up to date on revision r254.

But maybe there's another problem. I recently used the following script from this http://forum.bitcoin.org/index.php?topic=9239.0;all thread:

Code:

 # ${1} is used as a variable for the username, password and for the gpu device number.  Ex. mineruser0, minerpass0, Device=0 or mineruser1, minerpass1, Device=1
#      HOMEDIR=/home/alan
#      MINERSERV=nl.btcguild.com:8332/
#      MINERUSER=xxxxxx${1}
#      MINERPASS=xxxxxx${1}

      export AMDAPPSDKROOT=home/alan/AMD-APP-SDK-v2.4-lnx64/
      export AMDAPPSDKSAMPLESROOT=home/alan/AMD-APP-SDK-v2.4-lnx64/
      export LD_LIBRARY_PATH=${AMDAPPSDKROOT}lib/x86_64:${LD_LIBRARY_PATH}

      export DISPLAY=:0
      aticonfig --pplib-cmd 'set fanspeed 0 100'
      export DISPLAY=:0.1
      aticonfig --pplib-cmd 'set fanspeed 0 100'
      export DISPLAY=:0.2
      aticonfig --pplib-cmd 'set fanspeed 0 100'
      export DISPLAY=:0.3
      aticonfig --pplib-cmd 'set fanspeed 0 100'

      #Overclock GPU to 875Mhz
#      DISPLAY=:0 aticonfig --od-enable --adapter=all
#      DISPLAY=:0 aticonfig --od-setclocks=875,1000 --adapter=${1}
#      cd ${HOMEDIR}/phoenix-1.50/
#      echo "Startming Miner: ${1}"
#      ${HOMEDIR}/phoenix-1.50/phoenix.py -u http://${MINERUSER}:${MINERPASS}@${MINERSERV} VECTORS BFI_INT AGGRESSION=12 WORKSIZE=128 DEVICE=${1} -k phatk

# jondecker76 section
# ----------------------------------------------------------------------------------------------------------------------------------------
cd /home/alan/smartcoin
smartcoin

together with

Code:

description "Start Bitcoin Mining Screen"
start on runlevel [2345]
stop on runlevel [!2345]
kill timeout 30
script
 LOGINUSER=alan
 #Wait 30 seconds to make sure X is started.
 sleep 30
 #exec /usr/bin/screen -dmS gpu0 su -c '/usr/local/bin/startminer.sh 0' ${LOGINUSER}
 exec /usr/bin/screen -dmS scscr su -c '/usr/local/bin/startminer.sh 0' ${LOGINUSER}
end script

I simply tried to implement smartcoin into it (look 2nd exec command) but did not want to delete the "old" script yet. But it didn't work. What's happening is that smartcoin starts up and the immediately the screen closes. Maybe because the screen is started inside a screen?

Can you help out how to implement smartcoin into bootup or maybe a tutorial. I found out there are many ways and since I'm confused and don't want to mess up the system even more.

Thanks in advance!

Alan

[lordbuggie]

I would say you would want the following as your first script(I removed stuff that wasn't needed and set fanspeed to 90:I don't like running fans at 100 personally):

Code:

#!/bin/bash

export AMDAPPSDKROOT=$HOME/AMD-APP-SDK-v2.4-lnx64/
export LD_LIBRARY_PATH=${AMDAPPSDKROOT}lib/x86_64:${LD_LIBRARY_PATH}

export DISPLAY=:0
aticonfig --pplib-cmd 'set fanspeed 0 90'
export DISPLAY=:0.1
aticonfig --pplib-cmd 'set fanspeed 0 90'
export DISPLAY=:0.2
aticonfig --pplib-cmd 'set fanspeed 0 90'
export DISPLAY=:0.3
aticonfig --pplib-cmd 'set fanspeed 0 90'
#Overclock GPU to 875Mhz
DISPLAY=:0 aticonfig --od-enable --adapter=all
DISPLAY=:0 aticonfig --od-setclocks=875,1000 --adapter=all

#Go into the smartcoin dir and then execute smartcoin
cd $HOME/smartcoin/
/usr/bin/smartcoin

Then your second script looks ok to me.  I have yet to test it, as I don't run 100% headless.  I don't have a monitor installed, but I do use the above, but sudo all my commands to make sure aticonfig doesn't blow up.

[Alan Lupton]

I don't get it. I cleaned the startminer.sh like you told me and now the follwing error occurs:

Code:

smartcoin Management System  Do 30. Jun 21:20:10 CEST 2011
--------------------------------------------------------------------------------
 Host: localhoston found.
GPU[0]: Temp: 44.00 load: 0%
GPU[1]: Temp: 41.00 load: 0%
GPU[2]: Temp: 41.00 load: 0%
GPU[3]: Temp: 45.00 load: 3%
CPU Load : 17,00%

Profile: Automatic
--------BTCGuild--------
GPU[0]: <<<DOWN>>>
GPU[1]: <<<DOWN>>>
GPU[2]: <<<DOWN>>>
GPU[3]: <<<DOWN>>>
Total : [0 MHash/sec] [0 Accepted] [0 Rejected]

Grand Total: [0 hash/sec] [0 Accepted] [0 Rejected] [0.00% Rejected]
grep: /tmp/smartcoin-Miner.1: Unable to find file or directory
grep: /tmp/smartcoin-Miner.1: Unable to find file or directory
grep: /tmp/smartcoin-Miner.1: Unable to find file or directory
grep: /tmp/smartcoin-Miner.1: Unable to find file or directory
grep: /tmp/smartcoin-Miner.1: Unable to find file or directory
Runtime error (func=(main), adr=25): Divide by zero

And every few seconds I get the same error messages. And the CPU Load steadily increases toward 100%.

[jondecker76]

alan, what do you get back when you run this:

Code:

cd ~/smartcoin
svn info | grep "^URL" | awk '{print $2}'

I just want to make sure you're on the beta branch is all.

I think this error is because aticonfig is getting pounded with requests - i think I've seen this error before and it went away when I put a delay in between my aticonfig calls in smartcoin_status.sh (it sleeps for 0.2 seconds between each call).  If you have multiple scripts calling aticonfig, I can see how this would happen.

A quick test would be to comment out the other aticonfig calls in the other script, and see if the problem goes away

[jondecker76]

Alan,
What distro and version are you running?
I would suggest just trying to run smartcoin the standard way for a test (with no other launch scripts) - that way we can try to eliminate any other possibilities.

For a test:
- start out fresh.. Kill all the screen sessions (sudo killall screen)
- start smartcoin

Code:

smartcoin

- Lets make sure the database has the right cards in there:
(disconnect from smartcoin with ctrl+a d, or from the control tab

Code:

cd ~/smartcoin
sqlite3 smartcoin.db "SELECT * FROM device;"

- Did the miner session start?

Code:

screen -ls

You should see a smartcoin and a miner screen session
- If the miner session is there, check it out

Code:

screen -r miner

-RE: grep errors

Code:

cd /tmp
screen -ls

Do you see any Miner files in there?

[Alan Lupton]

alan, what do you get back when you run this:

Code:

cd ~/smartcoin
svn info | grep "^URL" | awk '{print $2}'

//smartcoin.googlecode.com/svn/branches/beta1/smartcoin

Alan,
What distro and version are you running?

Linux version 2.6.38-8-generic (buildd@allspice) (gcc version 4.5.2 (Ubuntu/Linaro 4.5.2-8ubuntu3) ) #42-Ubuntu SMP Mon Apr 11 03:31:24 UTC 2011

For a test:
- start out fresh.. Kill all the screen sessions (sudo killall screen)
- start smartcoin

Code:

smartcoin

- Lets make sure the database has the right cards in there:
(disconnect from smartcoin with ctrl+a d, or from the control tab

Code:

cd ~/smartcoin
sqlite3 smartcoin.db "SELECT * FROM device;"

1|1|GPU[0]|0|gpu|1|0
2|1|GPU[1]|1|gpu|1|0
3|1|GPU[2]|2|gpu|1|0
4|1|GPU[3]|3|gpu|1|0

- Did the miner session start?

Code:

screen -ls

There is a screen on:
        10994.smartcoin (30.06.2011 21:49:24)   (Detached)
1 Socket in /var/run/screen/S-alan.

You should see a smartcoin and a miner screen session
- If the miner session is there, check it out

Code:

screen -r miner

There is no screen to be resumed matching miner.

-RE: grep errors

Code:

cd /tmp
screen -ls

Do you see any Miner files in there?

[jondecker76]

oops, that last one was supposed to be:

Code:

cd /tmp
ls 

and look for any files with smartcoin or miner in the name

re: version. Is it Ubuntu Server? Which version?

re devices: do you have 4 GPU devices? (0,1,2 and 3?) I'm thinking one of them is the CPU? You may have done the install before that bug was fixed, which would explain why the cpu detected as a GPU

re miner screen session:
That is puzzling. It should have started, and even if it failed it should be there, as its launched with zombie mode enabled.
Is there anything at the end of ~/smartcoin.log that may be helpful?

[Alan Lupton]

Within /tmp there is only a pulse file, I suppose for pulse audio.

I have 4 GPU's.

Edit: this is a ubuntu 11.04 full desktop version crippled to only start as a server, basically. I used this tutorial:
http://forum.bitcoin.org/index.php?topic=9239.0;all

[jondecker76]

My own mining rig is an 11.04 32bit desktop install
(I run it headless, but didn't bother to disable any of the startup stuff, just enabled autologin for my user). Its very strange that there could be that big of a difference between our setups, as I used the same guide minus neutering the normal startup.

The grep errors are definitely from the miner screen session not starting - though I have no idea what would stop it from starting.
I recently enabled some new logging, do you see anything in the log referring to the launch paths its attempting to start the miners with?