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

[jaebird]

thank you!

"sudo apt-get install bc" fixed the status update problem!

You are welcome! Btw, I added some of my other LinuxCoin findings to the LinuxCoin thread.

[epg]

I'm having an issue with smartcoin not using one of my GPU's.  This is a 4x5830 box.  Notice that GPU[3] in the top section is idle, but GPU[0] in the bottom sections:

*snip*

Help?

I think I had this issue. Verify that your Devices (option 9 I think from the control menu) are each lined up with an OpenCL device/GPU. I had only 3 out of the 4 5830s working from the start and I had to manually add the fourth card before all 4 started working.

*edit* I also recall making a new profile, but I'm not sure if this step was necessary. Hope this helps!

[jondecker76]

Thanks for all the info guys,  I had a busy day today so I just got a chance to get on here.
Thank you to the LinuxCoin crew for the excellent info that will help getting this running smoother on LinuxCoin!

Just committed the new updates. They include:
- bc is not installed with the install script
- Hopefully, I fixed the "No screen session found." bug. I think what is happening, on some machines, is that after the miner screen session starts, my first call to command the screen session is coming too early (Not giving screen enough time to start up on some systems). There is now a 2 second delay after starting the session before pounding it with commands
- Started adding some of the local feedback to the log file.

Again, how to update:
prior to r273:

Code:

cd ~/smartcoin
svn update

For those after r273, just use the update option from the control screen.




Is anybody having problems after a reboot with errors regarding files in the /tmp folder?  I do have a few reports of this, just wanted to see if it was localized to a few people, or if its common.

[Alan Lupton]

- Hopefully, I fixed the "No screen session found." bug.

No, still have this error after starting up smartcoin.

Is anybody having problems after a reboot with errors regarding files in the /tmp folder?  I do have a few reports of this, just wanted to see if it was localized to a few people, or if its common.

That anybody would also still be me You still have my data? I leave the machine on, will go to bed now.

Just make sure you write down the phoenix startup line within the 4 screens (basically the same, just device), kill them, test around with smartcoin, and bring them back on if smartcoin won't start.

Thx for the continuous hard work!

Cu tomorrow,
Alan

[jondecker76]

I believe i just fixed both bugs (just committed the changes a second ago)
Run an update, and restart smartcoin. Hopefully this will do it

[jondecker76]

One thing to mention:
If for some reason there isn't a Miner screen session running
(you can tell by running:)

Code:

screen -ls

Chances are, your miner launch string is invalid for some reason.  Double check it in the Configure Miners option.

[echo465]

I believe i just fixed both bugs (just committed the changes a second ago)
Run an update, and restart smartcoin. Hopefully this will do it

Well, this looks better, after updating to 279...

Code:

--------------------------------------------------------------------------------
A configuration change has been detected.  Loading the new profile....
Starting miner Miner.1!
Starting miner Miner.2!
Starting miner Miner.3!
Starting miner Miner.4!
Starting miner Miner.5!
Starting miner Miner.6!
Starting miner Miner.7!
Starting miner Miner.8!

But I'm still showing the card to be not working. (and it looks like BTCGuild is having issues tonight (hey! good reason to be using smartcoin!), confirmed in IRC)

Code:

smartcoin Management System  Sat Jul 2 01:09:25 EDT 2011
--------------------------------------------------------------------------------
Host: localhost
GPU[0]: Temp: 80.50 load: 99%
GPU[1]: Temp: 77.50 load: 99%
GPU[2]: Temp: 76.00 load: 99%
GPU[3]: Temp: 49.00 load: 0%
CPU Load : 9.24%

Profile: Automatic
--------DeepBit--------
GPU[0]: <<<DOWN>>>
GPU[1]: [274.92 Mhash/sec] [3 Accepted] [0 Rejected] [RPC (+LP)]
GPU[2]: [274.28 Mhash/sec] [5 Accepted] [0 Rejected] [RPC (+LP)]
GPU[3]: [275.08 Mhash/sec] [2 Accepted] [0 Rejected] [RPC (+LP)]
Total : [824.28 Mhash/sec] [10 Accepted] [0 Rejected]
--------BTCGuild--------
GPU[0]: <<<DOWN>>>
GPU[1]: <<<DOWN>>>
GPU[2]: [0 Khash/sec] [0 Accepted] [0 Rejected] [RPC]
GPU[3]: [0 Khash/sec] [0 Accepted] [0 Rejected] [RPC]
Total : [0 MHash/sec] [0 Accepted] [0 Rejected]

Grand Total: [824.28 hash/sec] [10 Accepted] [0 Rejected] [0% Rejected]

The devices all look correct in option 9.  I'm not sure what the device table in the DB should look like, but:

Code:

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

Not sure if it's relevant or not, but there was a several second pause between miner.1 starting and the rest of them.

[kripz]

i'm getting an error on the 2nd tab, it flashes too quickly so i dont know what it is. looks like something related to aticonfig.

How do i change which display it's using? Mine runs on display=:0

[jondecker76]

New updates are available:
- Added current revision number to the header at the top of the screen
- Hopefully fixed the bug where GPU display gets out of order on some systems (as pointed out by echo465)
- Log files moved to smartcoin folder
- LogRotate implemented. On a new session of smartcoin, smartcoin.log is moved to smartcoin.log.previous, and smartcoin.log starts fresh
- Improved logging - converted a lot of echo'd feedback into Log entries
- I'm trying something new with exporting the DISPLAY variable
- The database is backed up to smartcoin.db.backup upon update. This way, if an update patch modifies the database, you still have the original backed up in case something doesn't work
- Various other cleanup of the code

echo465: Let me know if the newest  update fixes the out of order GPU display. Also, the pause after starting miner1 is to make sure screen successfully loaded before attempting to load the rest (thats what fixed the bug where you would get the error that the screen session wasn't found)

kripz: I'll look into the aticonfig error. I know this is one of the remaining issues, and its a pretty complex one.  I export DISPLAY=:0 before the aticonfig calls, so that shouldn't be the issue.  For some reason, depending on the install, here is what I have found:
A) On some systems, aticonfig will only run correctly if run with superuser permissions
B) On some systems, aticonfig will not run unless the call is directly preceeded by DISPLAY:0
C) On some systems, aticonfig will only run if export DISPLAY=:0 preceeds all aticonfig calls.

Unfortunately, I haven't found a way where it wants to run reliably on all systems yet (though I just commit a change here, so maybe it will be what we need), though I'm working on it still.  Any suggestions on this are welcome!

[Alan Lupton]

I'm using

DISPLAY=0

or

DISPLAY=0.1 sudo aticonfig --od-enable --odsc 900,300 (GPU Clock, RAM Clock)

or --odgt or whatever you might do with aticonfig.

That works for all my three rigs.

[Fletch]

One quick question. How do I get to the monitor screen? I've launched smartcoin and completed all the configuration, but there's no entry in the menu to get to the monitoring screen.

[Alan Lupton]

press "CTRL + A", release, "1" for status monitor

to get back to options:

press "CRTL + A", release, "0"

[sojolly]

Tried this on my CENTOS 5.6 setup.

When i run $./smartcoin.sh and then pick option 7 --> add i get a continuous set of lines that say:

Invalid selection. Please try again.
/home/sysadmin/smartcoin/smartcoin_ops.sh: line 507: read: -i: invalid option
read: usage: read [-ers] [-u fd] [-t timeout] [-p prompt] [-a array] [-n nchars] [-d delim] [name ...]

Thoughts on how to fix this?

Sojolly

[jondecker76]

sojolly:
It looks like CentOS is using a different version of read that doesn't use the -i parameter (the -i parameter lets me set the initial value to what is already in the database).  I wonder if CentOS has an alternate read package.. Beyond that, maybe you'll have to look up GNU read and compile it yourself?  Perhaps somebody in the CentOS community knows?

[jondecker76]

Update available:
r290 - AutoDonate now displays minutes remaining on the status screen.  Nothing important, just thought it would be nice for the user to have this feedback

[Alan Lupton]

Hey, I noticed something. Bitcoinpool appears to be down atm, so I get just rejections. Currently I have 40 accepted (BTCGuild) and 53 rejected. The status tab says that 120% were rejected. That's an error with the calculation, 56% would be correct.

[sojolly]

Switching one of my machines to Ubuntu 11.04 64-bit .. will let you know in a couple of hours....

[Fletch]

Thanks Alan. ctrl + a + 0/1 did the trick.

Question number two. I have two GPUs and have setup one worker for each GPU. However, Smartcoin appears to launch four phoenix instances instead of two. It launches one phoenix instance for each device and each worker. Any way around that?

Example:

Code:

smartcoin Management System r286 Sat Jul 2 19:57:50 CEST 2011
--------------------------------------------------------------------------------
Host: localhost
GPU[0]: Temp: 75.50 load: 99%
GPU[1]: Temp: 74.50 load: 99%
CPU Load : 2.37%

Profile: Automatic
--------Bitcoins.lc--------
GPU[0]: [170.67 Mhash/sec] [1 Accepted] [0 Rejected] [RPC (+LP)]
GPU[1]: [164.99 Mhash/sec] [2 Accepted] [0 Rejected] [RPC (+LP)]
GPU[0]: [170.67 Mhash/sec] [0 Accepted] [0 Rejected] [RPC (+LP)]
GPU[1]: [187.75 Mhash/sec] [2 Accepted] [0 Rejected] [RPC (+LP)]
Total : [694.08 MHash/sec] [5 Accepted] [0 Rejected]

Grand Total: [694.08 Mhash/sec] [5 Accepted] [0 Rejected] [0% Rejected]

[padrino]

I just grabbed smartcoin to give it a shot, love the post on the features. In any case I have 8 GPUs and I configured a worker for each, it appears it's trying to launch 64 miners

[jondecker76]

Kennel:
This is correct behavior.  The auto profile launches 1 instance of each worker that is set to allow auto profiling on each GPU.  Really, what you want, it to only define unique workers.  To properly assign workers to cards, use the profile system.

For example, lets say you have a worker, named "My Worker"
The auto profile would start an instance of phoenix on each GPU, against "My Worker"
If you want to create your own profile (instead of the auto profile), then you have full control over which workers receive their own instance on each specified GPU.