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

[plantucha]

Run one more update - this should bring you to r643 (r607 is termed a "breakpoint update", which means its an update which requires a smartcoin restart before updating beyond this point. I will be making this more clear in the future, and possibly even automating the process of restarting smartcoin and continuing the update).
After this update, a restart of smartcoin should fix the problem. There is new database validation and sanity checking code in changes made after r607 which will automatically fix this issue.

Please report back and let me know how it goes

Yes, thank you. Now the message seem not appearing any more.


Sometimes I have another message in the line right after a GPU-name: <<<idle>>>
And in some rare cases, allmost if I use as GPU-MHz something above 900 MHz (using AMDOverdriveCtrl or aticonfig instead) the CPU goes up to 100% and all GPUs are changing to <<<idle>>>. In this case I found as only solution a complete reboot via putty/ssh.

If I leave it on GPU 900MHz and Mem 840Mhz this does not happen, only sometimes a single GPU is short time on <<<idle>>>.

If I do mining via script and not via smartcoin this behavior do not appear.

What does this mean and how can I prevent this?

As some people are telling that 910:840 and agression=7 is the best (refer this: https://bitcointalk.org/index.php?topic=25798.0;all - the guy who told this has exactly the same config as I have), I would like to do that also, because he makes about 450 MH/s each GPU and I make only 350.

what model GPU do you have?

[1713567353]

1713567353

[1713567353]

1713567353

[1713567353]

1713567353

[1713567353]

1713567353

[Jessi]

I've got 3xATI 6990 in my RIG.

[plantucha]

I've got 3xATI 6990 in my RIG.

as somebody already said:
His picture in the signature shows now: 2314 Mhash/s
Some calculation reveals: 2314 Mhash/6 cores = 385,6 Mhash per core.
Don't believe some blury pictures.


but
if you wanna go faster try kernel phatk 2.1 or later 2.2
https://bitcointalk.org/index.php?topic=7964.0

[Jessi]

Ok... I was believing his initial post, because he described very exactly what he is doing. 

Anyway, the new kernel sound intresting for me, but can I change this in linuxcoin also and how can I determine which is used inside linuxcoin allready?

[plantucha]

Ok... I was believing his initial post, because he described very exactly what he is doing. 

Anyway, the new kernel sound intresting for me, but can I change this in linuxcoin also and how can I determine which is used inside linuxcoin allready?

1. you can ask linuxcoin creator
or
2. Look in folder with phatk kernel, compare size , compare date

but you don't need to,
you can simply just copy new kernel folder and put it in original folder
if you have already new kernel => no change
if you have old kernel => you will compute faster
no regression was reported so it's relatively safe.

[jondecker76]

Jessi

Just getting on here now, had to catch up on some sleep today.  I wanted to point out a few things and answer some of your questions that came up

If a GPU locks-up, its pretty much always caused by an overly aggressive overclock. Smartcoin its self has no role in this at all, it simply launches and shuts down miner profile instances by a set of rules and defined profiles.  Also, another thing to realize is that just because "User X" can push their card to a certain limit doesn't always mean that everyone can - small differences such as power supply used, motherboard, cooling method and even slight changes in different revisions of the same exact card model can make possible output vary quite a bit.

Regarding the <<<IDLE>>> message, its normal if it pops up from time to time - it normally just means that the miner's output didn't contain a certain search string used to yank the hash information from its output (remember, smartcoin its self isn't a miner, it just reads the output of miner instances for monitoring and control purposes). You can always directly look at the miner instances by detaching from smartcoin, and typing:

Code:

screen -r miner

Then use ctrl+a p and ctrl+a n (previous tab, and next tab) to see the instances for yourself.

If the idle message is happening a lot, or appears to be happening in cycles, look at the GPU temperature. Catalyst has its own temperature limiting algorithms, and on most cards they will start throttling themselves and even shutting down for short periods if temperatures go over 90 degrees.

[Jessi]

Thank you for your very detailed explaination of this situation.

Is there a way to check this idle-state via a script (maybe with a cronjob every 10 Minutes) and make a restart of the rig and let smartcoin start again?

[jondecker76]

Thank you for your very detailed explaination of this situation.

Is there a way to check this idle-state via a script (maybe with a cronjob every 10 Minutes) and make a restart of the rig and let smartcoin start again?

Smartcoin already has everything in place for this.

1) Failover system- A miner showing as <<<IDLE>>> for a determined amount of "refreshes" (See the Failover Threshold setting in Edit Settings) will failover following your set profile order IF 1) You have manual profiles defined and an order defined AND 2) You are running the special failover profile

2) Lockup System- (See the Lockup Threshold setting in Edit Settings)
If a miner instance locks up (does not change for a determined number of refreshes), it will trip the lockup system which will automatically reload the profile and miner instances (a lockup condition can happen from crashed miner software and even a GPU hard lockup).  Also, if  "lockup.sh" exists in the smartcoin directory, it will execute that script. This allows you to do advanced things like send yourself an email and even reboot the computer if you want.  Its a ways back in the post, but I did post an example of a custom lockup.sh script. You can read the post here https://bitcointalk.org/index.php?topic=16548.msg382427#msg382427, as well as some of the discussion around it.


While on the subject of custom scripts, smartcoin also will launch your own custom "init.sh" script if it is located in the smartcoin directory when you launch smartcoin.  Some people use this to crank up their fan speeds when smartcoin starts for example.

I know the documentation is a little crummy right now (I simply haven't had much of any time to document a lot of the features, i barely have time to code a lot of the time).  However, all features are covered somewhere in this thread, so although its long and probably a pain, you can learn a lot if you want to start from the beginning and read through this post history.


The beauty in using smartcoin, is that once you get it set up, you barely ever have to do anything but sit back and check in from time to time.  Pools getting DDOS's? No problem, failover takes over.  Miner software locks up? No problem, lockup system takes over. GPU locked up?  No problem if you have your machine set to reboot automatically on a lockup via lockup.sh.  I absolutely never touch my miners anymore, they just keep doing their job - I usually check in once a day and thats about it!

[jondecker76]

Update r647e now available
- Started writing the macro feature routines. All stubs now in place
- Huge optimizations on the multi-machine code.  12+ SQL queries per iteration per machine were eliminated (except for the localhost machine).

[jondecker76]

Update r652e now available!
- New macro system implementation complete (except for 'Edit Macro')


Here is some information about the macro system:
Macros are only useful if you are running multiple machines. The macro system allows you to change the profile on multiple machines at the same time all with one command (Think of a macro as a "macro-profile"). To use, go to Configure Macros->Add.
You can then build a macro and define which profile to switch to on which machines when the macro is executed. If a profile isn't assigned for a machine, then the profile will stay the same for the machine when the macro is executed.

An example of a useful macro would be to create a macro named "Fallover All", then assign the failover profile for each machine.  When you execute the "Failover All" macro, then the profile on each machine will change to failover.  Pretty simple!

Simply use the "Execute Macro" option to choose a defined macro to execute!


Sample Usage Case
Imagine that you are running a mining farm with 10 machines. You get news that your favorite pool has been cheating its users and you want to switch all of your machines to a different pool as soon as possible. Simply Execute one of your pre-built macros and all of your machines will redirect to the indicated profiles - all within seconds!


Known Issues: Edit Macro not yet implemented, though you can always delete a macro and add a new one to start over for now.  The edit macro functionality should be finished in the next day or two.

[norulezapply]

Hey getting some bizarre problems with my smartcoin
Only seems to have started recently.
It's been mining fine for a while now.
But now on the status screen it shows <<IDLE>> for both my miners.
In the log it says they have locked up, but to be honest I doubt that both of my cards would lock up at the exact same time would they?
When trying to reload the miners, the miner screens just sit at the "Launching phoenix" screen and have no more output.
Then eventually I lose the ability to monitor my temperature (this occurs 10 minutes or so afterward, when i attempted to manually restart smartcoin).
And i get the message "No protocol specified" in the status window (a log from aticonfig maybe?)
The only solution is to restart
Any ideas?

[jondecker76]

This sounds exactly like a hardlock condition - and would make sense that they happen at the same time because most people overclock all their cards together (--adapter=all)

If it happens again, disconnect from smartcoin then type:

Code:

screen -r miner

Then you can look at all of the miner instances manually and see what is going on.

Another easy test would be to reduce clocks a little and see if the hardlock no longer happens.

Please post back what you find out!

[sunny day]

I can't get smartcoin to run the miners. on the localhost page all the miners say idle and occasionally a bunch of "aticonfig: no supported adapters are found" messages scroll by. on the log page i try to manually run the phoenix command but it gives me a kernel not found error.

when i run aticonfig manually it works but only when i sudo it.
i normally use cgminer and that works with my cards just fine.
AMDOverdrive also detects the cards fine.
any ideas jondecker?

[norulezapply]

@sunny day: try running smartcoin as root (sudo smartcoin)

@jondecker: Well I overclock my cards separately using atitweak because it's a 5770 and 5830 so they have different clocks.
The status screen shows <<DOWN>> for both cards.
It's just happened again now and I can still see the temperatures so aticonfig must still be working.
When I look at the miner screen instances all I see is
"Exporting: /opt/AMD-APP-SDK-v2.4-lnx64/lib/x86_64
LAUNCH: python /opt/miners/phoenix//phoenix.py -v -u http://btcrig:btcrig@localhost:8337/ device=0 worksize=256 vectors aggression=10 bfi_int fastloop=false -k phatk"

Also another thing I noticed is if phoenix fails to connect it will just stop trying after a couple of attempts:
e.g. "Exporting: /opt/AMD-APP-SDK-v2.4-lnx64/lib/x86_64
LAUNCH: python /opt/miners/phoenix//phoenix.py -v -u http://btcrig:btcrig@localhost:8337/ device=0 worksize=256 vectors aggression=10 bfi_int fastloop=false -k phatk
[10/08/2011 22:45:12] Phoenix r111 starting...
[10/08/2011 22:45:22] Failed to connect, retrying...
[10/08/2011 22:45:32] Failed to connect, retrying..."

Even though the server is up and running as I'm mining it on my other PC. It's been sat like that for 25 minutes without attempting to reconnect again now and smartcoin doesn't attempt to restart them either.

If I restart them manually they connect fine.

[jondecker76]

I can't get smartcoin to run the miners. on the localhost page all the miners say idle and occasionally a bunch of "aticonfig: no supported adapters are found" messages scroll by. on the log page i try to manually run the phoenix command but it gives me a kernel not found error.

when i run aticonfig manually it works but only when i sudo it.
i normally use cgminer and that works with my cards just fine.
AMDOverdrive also detects the cards fine.
any ideas jondecker?

What linux distro are you running under? What install guide did you use?
You can run smartcoin as root, but that means you will also need to run the installer as root to set things up

[jondecker76]

@sunny day: try running smartcoin as root (sudo smartcoin)

@jondecker: Well I overclock my cards separately using atitweak because it's a 5770 and 5830 so they have different clocks.
The status screen shows <<DOWN>> for both cards.
It's just happened again now and I can still see the temperatures so aticonfig must still be working.
When I look at the miner screen instances all I see is
"Exporting: /opt/AMD-APP-SDK-v2.4-lnx64/lib/x86_64
LAUNCH: python /opt/miners/phoenix//phoenix.py -v -u http://btcrig:btcrig@localhost:8337/ device=0 worksize=256 vectors aggression=10 bfi_int fastloop=false -k phatk"

Also another thing I noticed is if phoenix fails to connect it will just stop trying after a couple of attempts:
e.g. "Exporting: /opt/AMD-APP-SDK-v2.4-lnx64/lib/x86_64
LAUNCH: python /opt/miners/phoenix//phoenix.py -v -u http://btcrig:btcrig@localhost:8337/ device=0 worksize=256 vectors aggression=10 bfi_int fastloop=false -k phatk
[10/08/2011 22:45:12] Phoenix r111 starting...
[10/08/2011 22:45:22] Failed to connect, retrying...
[10/08/2011 22:45:32] Failed to connect, retrying..."

Even though the server is up and running as I'm mining it on my other PC. It's been sat like that for 25 minutes without attempting to reconnect again now and smartcoin doesn't attempt to restart them either.

If I restart them manually they connect fine.

What distro are you running smartcoin on?  When the miner instance screen shows blank, is it like that when it starts, or after its already been running and then hangs? (Does smartcoin ever work properly, if even just for a little bit?  If/when it is working, do the miner instance screens look normal?)

Regarding phoenix connection issues, there isn't much I can do about phoenix issues, though it appears to me that lockup detection should happen in this instance.  Also, i have gotten reports of r 111 and r112  of phoenix having these types of problems (I run phoenix r107 and never have issues like you are describing, perhaps you could verify?)

[Math Man]

I am trying to get Smartcoin working on one of my rigs, which is running a persistent install of LinuxCoin 0.2b.  It has 3 HD 5970s in it, but I'm just trying to get the first GPU working.  I do have read/write permissions to the phoenix folder.  The Smartcoin host is running Debian Squeeze.

This below flashes across the screen a bunch of times when I switch to screen 2*$btc1.  Note: rylan is my user name on the Linuxcoin client and the Debian host.

Code:

Starting miner Miner.1!
Creating persistent ssh connection to machine 2
Bad port 'rylan'
Bad port 'rylan'

What does Bad port 'rylan' mean?  I have the ssh port set to 22.


Then the below screen shows up, where GPU[0] is idle.  There is a connection to btc1 though, because it's showing the correct temperature of GPU[0]

Code:

Smartcoin r652s 01:09:57
----------------------------------------
No screen session found. Host: btc1
GPU[0]: Temp: 56.00 load: 0%
CPU Load Avgs: 0.10 0.10 0.16

Profile: Automatic
--------Mainframe--------
GPU[0]: <<<IDLE>>>
Total : [0 MHash/sec] [0 Accepted] [0 Rejected] [0.00% Rejected]

Grand Total : [0 MHash/sec] [0 Accepted] [0 Rejected] [0.00% Rejected]


[ debian ][   0-$ control  1$ localhost  (2*$btc1)  3$ Log   ][ 08/11/11  1:10 ]

Below is my log

Code:

Starting SmartCoin at location: /home/rylan/smartcoin...
Running check
Starting main smartcoin screen session...


Creating persistent ssh connection to machine 2

Verifying integrity of general settings...
Verifying integrity of settings for machine 1...
Verifying integrity of settings for machine 2...
Creating tab for each machine...
Creating a logtail tab...
08/11/11 01:11:56       DEBUG: "screen -r miner -X hardstatus string '%{= kG}[ %{G}%H %{g}][%= %{= kw}%?%-Lw%?%{r}(%{W}%n*%f%t%?(%u)%?%{r})%{w}%?%+Lw%?%?%= %{g}][%{B} %m/%d/%y %{W}%c %{g}]'"

08/11/11 01:11:56       Starting miner Miner.1!

08/11/11 01:11:56       Launching miner with launch string: python /opt/miners/phoenix/phoenix.py -v -u http://MYUSERNAME:MYPASSWORD@mining.mainframe.nl:8343/ -k phatk -q 6 DEVICE=0 VECTORS BFI_INT FASTLOOP=false WORKSIZE=128 AGGRESSION=9

08/11/11 01:11:56       Creating persistent ssh connection to machine 2

08/11/11 01:11:56       DEBUG: "screen  -d -r miner -X screen -t Miner.1 /tmp/smartcoin_launcher.sh /opt/miners/phoenix/ 'python /opt/miners/phoenix/phoenix.py -v -u http://MYUSERNAME:MYPASSWORD@mining.mainframe.nl:8343/ -k phatk -q 6 DEVICE=0 VECTORS BFI_INT FASTLOOP=false WORKSIZE=128 AGGRESSION=9' "

08/11/11 01:11:56       Creating persistent ssh connection to machine 2

08/11/11 01:11:56       DEBUG: screen -r miner -p miner-dummy -X kill

Halp!   

[jondecker76]

I am trying to get Smartcoin working on one of my rigs, which is running a persistent install of LinuxCoin 0.2b.  It has 3 HD 5970s in it, but I'm just trying to get the first GPU working.  I do have read/write permissions to the phoenix folder.  The Smartcoin host is running Debian Squeeze.

This below flashes across the screen a bunch of times when I switch to screen 2*$btc1.  Note: rylan is my user name on the Linuxcoin client and the Debian host.

Code:

Starting miner Miner.1!
Creating persistent ssh connection to machine 2
Bad port 'rylan'
Bad port 'rylan'

What does Bad port 'rylan' mean?  I have the ssh port set to 22.


Then the below screen shows up, where GPU[0] is idle.  There is a connection to btc1 though, because it's showing the correct temperature of GPU[0]

Code:

Smartcoin r652s 01:09:57
----------------------------------------
No screen session found. Host: btc1
GPU[0]: Temp: 56.00 load: 0%
CPU Load Avgs: 0.10 0.10 0.16

Profile: Automatic
--------Mainframe--------
GPU[0]: <<<IDLE>>>
Total : [0 MHash/sec] [0 Accepted] [0 Rejected] [0.00% Rejected]

Grand Total : [0 MHash/sec] [0 Accepted] [0 Rejected] [0.00% Rejected]


[ debian ][   0-$ control  1$ localhost  (2*$btc1)  3$ Log   ][ 08/11/11  1:10 ]

Below is my log

Code:

Starting SmartCoin at location: /home/rylan/smartcoin...
Running check
Starting main smartcoin screen session...


Creating persistent ssh connection to machine 2

Verifying integrity of general settings...
Verifying integrity of settings for machine 1...
Verifying integrity of settings for machine 2...
Creating tab for each machine...
Creating a logtail tab...
08/11/11 01:11:56       DEBUG: "screen -r miner -X hardstatus string '%{= kG}[ %{G}%H %{g}][%= %{= kw}%?%-Lw%?%{r}(%{W}%n*%f%t%?(%u)%?%{r})%{w}%?%+Lw%?%?%= %{g}][%{B} %m/%d/%y %{W}%c %{g}]'"

08/11/11 01:11:56       Starting miner Miner.1!

08/11/11 01:11:56       Launching miner with launch string: python /opt/miners/phoenix/phoenix.py -v -u http://MYUSERNAME:MYPASSWORD@mining.mainframe.nl:8343/ -k phatk -q 6 DEVICE=0 VECTORS BFI_INT FASTLOOP=false WORKSIZE=128 AGGRESSION=9

08/11/11 01:11:56       Creating persistent ssh connection to machine 2

08/11/11 01:11:56       DEBUG: "screen  -d -r miner -X screen -t Miner.1 /tmp/smartcoin_launcher.sh /opt/miners/phoenix/ 'python /opt/miners/phoenix/phoenix.py -v -u http://MYUSERNAME:MYPASSWORD@mining.mainframe.nl:8343/ -k phatk -q 6 DEVICE=0 VECTORS BFI_INT FASTLOOP=false WORKSIZE=128 AGGRESSION=9' "

08/11/11 01:11:56       Creating persistent ssh connection to machine 2

08/11/11 01:11:56       DEBUG: screen -r miner -p miner-dummy -X kill

Halp!   

I am uploading an update right now that will fix your problem - there was a bug in a couple places where the user and the port were switched. It is now fixed.

Also,  I noticed that you are trying to remotely manage a Linuxcoin machine. See this from the multi-machine release note:

- If a remote machine normally runs smartcoin as root (with sudo - Linuxcoin would be an example of this), then when setting up a remote machine from Configure Machines, you must use the username 'root'.  This also means that the remote machine must have a root password for this to work.  To set up a root password, on the remote machine (before setting up the machine entry on the control screen):
Code:

Code:

sudo su

passwd

then enter a password twice.  You can now set up the remote machine in Smartcoin using the 'root' user

Here is what I would do in your situation to get things right again:
1) Delete the remote linuxcoin machine from Configure Machines->Delete
2) Update smartcoin to latest
3) Exit smartcoin
4) On your linuxcoin machine, make sure you have a root password set (per above instructions)
5) Fire smartcoin back up and add the linuxcoin machine again, and make sure you use the 'root' user.

This should get things working correctly again!  Thanks for the report!

[jondecker76]

Update 655e/655s now available:
- Fixed a remote machine bug where the port and username were switched around.
- Small change to donation system. Now only 1 worker per device will start up - this will be easier on resources.  I also switched to Eligius for the donation pool

[jondecker76]

Update r657e/657s now available
- There is a new tag supported in the launch string. The new tag is <#device+1#>.  This is useful if a miner always treats the CPU as device 0, which ends up offsetting the GPU devices.  This would cause the GPU load and temperatures to be out of sync on a few installs, as well as cause the CPU to be fired up as GPU[0] on some installs.  By using the <#device+1#> tag in the launch string, you can counteract this problem directly on a per-miner basis.  If your devices ever seemed out of order, or you ever had to manually change your devices to start with 1 - you may want to give this new tag a try.