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

[jondecker76]

Update r468e now available!
- There is a new setting under edit settings.. "Miner output format string". As discussed a couple of posts back, it will let you define how the miner output is displayed on the screen. Use the tags <#hashrate#>, <#accepted#>, <#rejected#>, <#rejected_percent#> where you want these displayed.
- Each miner instance now has rejection percentage calculation


I'd still love to hear from anyone that modifies the "Miner output format string" setting and hear how it worked for them (jaebird - try making a narrower format string )

Output formatting works good (I restarted smartcoin for the settings to change output). Now the main header needs a way to be reduced. Perhaps allow some timestamp formatting or shorten experimental to "e" or "exp", or just allow the whole string to be user modified.

Thanks,
jaebird

Yes, I plan on shortening the header to something like:

Code:

Smartcoin r###(e or s) <#date#>

 
I also thing I'll shorten the date string to look like:

Code:

 Jul 19 2011 13:32:29

[1714047569]

1714047569

[TonyHoyle]

I can't get past the install script...

Code:

Would you like smartcoin to attempt to auto-detect installed miners? (y)es or (n)o?
y
Found phoenix miner installed on local system
1) /raid/home/tmh/phoenix/phoenix.py
Select the local phoenix installation from the list above
1

Found poclbm miner installed on local system

Which miner listed above do you want to be the default miner?
poclbm
Invalid selection. Please try again.
poclbm
Invalid selection. Please try again.
1
Invalid selection. Please try again.
phatk
Invalid selection. Please try again.

What do I type here?  It's not obvious....

Breaking out of that, I can't configure it manually... edit settings brings up the following..

Code:

smartcoin Management System r469() Tue Jul 19 18:42:55 BST 2011
--------------------------------------------------------------------------------
EDIT SETTINGS
Select the setting from the list above that you wish to edit

*What* list above???

I'm clearly missing something obvious as others have managed to get it to work, but damned if I can see it..

[jondecker76]

finally after alot of distress with linuxcoin and smartcoin it finally works ,  well not sure if i did it right but it does work

i configure a miner useing 7 as option it mines at 79.32 mhash which is impressive to me .   then i thought well i try to figure out how to do failover which says i need another profile.  well i dont understand this so i go back to useing 7 as option and accidently created another miner mining another pool useing same gpu.    what is strange now is that the hash rate went up doing this ,   i think now this is a mystery and now wonder if i make more miners in same pool and create two miners to one gpu.   

well meanwhile i need to figure out this failover if i am to use it downstairs  ,   


Jen

Glad to hear you're up and running Jen

I will try to give you a quick rundown:

First, lets get you familiar with the terms smartcoin uses.

miner- Software used for mining bitcoins (such as Phoenix, poclbm, cgminer, etc...)
Instance- An individual copy of one of the miners running on the machine (for example, you can have an instance of phoenix hashing to your deepbit worker and another instance of phoenix hashing to your Elegius worker on the same GPU.  You can have many instance running on each GPU and is limited only by memory really)
worker- Your own login information for a specific pool
Profile- A profile is a collection of instances that you can recall at any time. A profile lets you specify how many instance of which miner are running on each GPU, and to which worker account they are mining to.  There are 3 "special" profiles that are built in: 1)An automatic profile which will open an instance on each GPU to each of your worker accounts  2)A manual donation profile which will mine to special donation pools, 3) A failover profile, which will mine to specific profiles in order, depending on their up/down state/
All other profiles must be manually defined by you (and you must manually define at least 2 profiles to use the failover profile - otherwise there will be no profiles to fail over to)



For your initial setup, lets do some things in order:
1)  Under 7) Configure Workers, add all your worker accounts that you have.  Even if you don't plan on mining to them right away, create a new worker for every account that you have with various pools.  This will store them so that you can use them at any time later. You can choose whether they will be used by the automatic profile or not - as long as you have less than 10 workers, I would select 'y' (for yes) when asked this question.

2) You should notice, that as you add more workers, that your automatic profile should continue to grow on the status screen.

3) Now lets create a manual profile so that you understand how to do it.   Go to Configure Profiles -> Add
Create a simple profile and name it "test" when prompted.
You will then need to add a miner instance, when prompted select 1) for phoenix
Then you need to select a worker. Select the first worker in the list
Then tell it which GPU to use, select your first GPU
At this point, the screen will clear, and you will be shown your current progress.  It will show you which miner, which device and which worker it will use on this first instance.
If you have more GPU's, then you can select 'y' to add another instance to the profile, and this time around select the next GPU in the list when prompted.  Keep doing this until you have as many instances as you want in your profile, and finally select 'n' when asked if you want to continue adding instances to the profile.

Here are some examples of a few of my profiles to give you an idea:


Here is my "Deepbit Me!" profile. As you can see, I have 3 GPU's, and I'm opening a phoenix miner instance to my deepbit worker on each GPU

Code:

1) GPU[0] - DeepBit.default - phoenix‌
2) GPU[1] - DeepBit.default - phoenix‌
3) GPU[2] - DeepBit.default - phoenix‌

Here is my "X8S Me DD!" profile (DD stands for "Double Down")  As you can see, I open 2 instances per GPU to my X8S worker.
As you have pointed out,  there is a slight increase in  your hashing for running multiple instances, which is why I doubled up on the instances in this example.

Code:

1) GPU[0] - X8S.default - phoenix‌
2) GPU[0] - X8S.default - phoenix‌
3) GPU[1] - X8S.default - phoenix‌
4) GPU[1] - X8S.default - phoenix‌
5) GPU[2] - X8S.default - phoenix‌
6) GPU[2] - X8S.default - phoenix‌

Here is my "X8S and Triplemine" profile.  As you can see, I'm sending hashes from 2 GPU's to my X8S account, and hashes from 1 GPU to my triplemine worker.

Code:

1) GPU[0] - X8S.default - phoenix‌
2) GPU[1] - X8S.default - phoenix‌
3) GPU[2] - TripleMining.default - phoenix‌

The profiles you can create are limitless!


Once you have created some profiles as tests (I recommend creating a separate profile for each of your workers similar to my "Deepbit Me!" example), you can use option 5) Select Profile to select the profile.  When you do, switch to the status screen and you will see smartcoin reconfigure its self and start mining to the selected profile (may take 5-10 seconds)

Once you get comfortable with the profile system, and practice switching between a few of them, you are ready for failovers.
Use option 12) Set Failover Order, and define an order of your manual profiles that you wish failover to happen in. Then simply go to 5) Choose Profile, and select the special Failover profile.  Smartcoin will take care of the rest!

[jondecker76]

I can't get past the install script...

Code:

Would you like smartcoin to attempt to auto-detect installed miners? (y)es or (n)o?
y
Found phoenix miner installed on local system
1) /raid/home/tmh/phoenix/phoenix.py
Select the local phoenix installation from the list above
1

Found poclbm miner installed on local system

Which miner listed above do you want to be the default miner?
poclbm
Invalid selection. Please try again.
poclbm
Invalid selection. Please try again.
1
Invalid selection. Please try again.
phatk
Invalid selection. Please try again.

What do I type here?  It's not obvious....

Breaking out of that, I can't configure it manually... edit settings brings up the following..

Code:

smartcoin Management System r469() Tue Jul 19 18:42:55 BST 2011
--------------------------------------------------------------------------------
EDIT SETTINGS
Select the setting from the list above that you wish to edit

*What* list above???

I'm clearly missing something obvious as others have managed to get it to work, but damned if I can see it..

It appears that the database failed to be created, which is why all your selection options are blank. The only way I can see this happening is if you checked out the smartcoin source to a path requiring root permissions (/root or /opt, for example), then ran the install script as a user without root permissions.
What is your install location? 
If you are trying to install in a location requiring root access, switch to the root user first:

Code:

sudo su

Aside from that, please list some more details about your setup:
- Distro being used
- 32 or 64 bit
- install location

[Jen4538]

another question i have is

when i start smartcoin ,  it starts mining right away .  is there a way to stop the mining process while i reconfigure options or add more workers?  machine is very sluggish when its mining so i prefer it not to mine while i am busy changing things

i see i have to do more things today to make sure i am comfortable with all this.    my main machine had a headache this morning so i am fixing it before i can go back to smartcoin to try your examples.   and yes i am very curious why it hashs more with 2 workers to one gpu .   well i will take advantage of this either way

Jen

[jaebird]

@jondecker76

I have pretty bad internet access and sometimes goes down for several minutes at a time. It seems that the new GPU lockup mechanism is triggered when my internet goes down. Is this a case you can handle? I really don't want or need to restart the PC as things normally pick back up when the internet connection comes back.

Thanks,
jaebird

[Rob P.]

@jondecker76

I have pretty bad internet access and sometimes goes down for several minutes at a time. It seems that the new GPU lockup mechanism is triggered when my internet goes down. Is this a case you can handle? I really don't want or need to restart the PC as things normally pick back up when the internet connection comes back.

Thanks,
jaebird

Just adding my 0.02 BTC here...

This condition *could* be handled in a custom failure script (lockup.sh).  When run, you shutdown Smartcoin (you've lost your Internet, you're not getting work and your miners are idle anyhow).  Your script runs in a loop performing a ping looking for when the network comes back.  When it comes back, you execute "smartcoin --silent" and fire it back up.  Yes, this would loop indefinitely if you've actually got a lock up...  You could add a condition to see how long it has been since you ran it last, and if too short of a time, take the reboot option...

Just thinking out loud.

[jaebird]

Just adding my 0.02 BTC here...

This condition *could* be handled in a custom failure script (lockup.sh).  When run, you shutdown Smartcoin (you've lost your Internet, you're not getting work and your miners are idle anyhow).  Your script runs in a loop performing a ping looking for when the network comes back.  When it comes back, you execute "smartcoin --silent" and fire it back up.  Yes, this would loop indefinitely if you've actually got a lock up...  You could add a condition to see how long it has been since you ran it last, and if too short of a time, take the reboot option...

Just thinking out loud.

yeah, I was thinking of something like that. Or also smartcoin could attempt to ping known good hosts in the event it "thinks" the cards are locked-up. It is a very very rare occurrence that all 4 cards would be locked up at the same time! In the meantime, how do I disable the lockup detection or action, add an empty "lockup.sh" script?

Thanks.

[jondecker76]

another question i have is

when i start smartcoin ,  it starts mining right away .  is there a way to stop the mining process while i reconfigure options or add more workers?  machine is very sluggish when its mining so i prefer it not to mine while i am busy changing things

i see i have to do more things today to make sure i am comfortable with all this.    my main machine had a headache this morning so i am fixing it before i can go back to smartcoin to try your examples.   and yes i am very curious why it hashs more with 2 workers to one gpu .   well i will take advantage of this either way

Jen

Jen

You should start getting used to accessing your machine via SSH from another computer (this is specifically what smartcoin is designed for). Your SSH session will not be slowed down like a local desktop is.

However, I do realize the need for an "Idle" profile, and I will probably add support for this in the next day or so.

Also, its a pretty well known side effect that running multiple instances per card gains a little hash speed (I get a 7% increase by running 5 instances per card). The reason as I understand it, is there are times the CPU waits idle for the GPU to be ready to accept information (caused by some lag in communication and OpenCL overhead). With multiple instances, the CPU has more of a chance with one of the  "threads" is waiting and ready to receive information from the CPU - hence the slightly higher hash rates.

[jondecker76]

On one of my rig running this, it has weird issues. When I update a worker, the assigned pool # will show up increased by one than it actually is assigned to. Sometimes this happens to miner assignment as well.

Are you talking about the pre-filled in selection while editing?

[jondecker76]

Jaebird and Rob P.

I'll be adding support to customize the threshold for both failover and lockup detection in the next day (setting to 0 will disable).  I'm at work all night, so I'm not sure if I'll get a chance to work on it tonight from  here (depends on how busy things get), or when I get home in the morning.

Here are the changes I am planning to make:
- Make --kill, --reload and --restart options for smartcoin (smartcoin.sh). This will allow you to take actions easily from custom scripts
- On lockup detection, fire custom script, and restart smartcoin automatically (instead of killing smartcoin). You should know if there is a real issue if this happens repeatedly (I'm still debating on whether smartcoin should automatically send emails on these events, or if that should be left up to the user with their custom script..  Thoughts?)
- Regarding the lockup routine when there is a loss of Internet, I think the current scheme (after above changes) will be plenty acceptable. Basically in this case, smartcoin will continually restart itself (unless you do a 'smartcoin --kill' from your custom script) about every 5 minutes until the Internet comes back.  This really won't waste any extra electricity, as the GPUs will be idle themselves with no work available for them. This also has the advantage of dealing with miner software that has locked up (as an example of such has already been shown here).  This also fits the philosophy of smartcoin in that once its running, it should try to automatically deal with things in a sane manner without manual intervention. (If I hadn't had to reboot my mining machine from time to time for testing purposes, I would wager that smartcoin has ensured me 0% downtime with no intervention from me other than initially setting up failover - its literally dealt with every problem automatically on my machine, and I've had to do nothing but check in every now and then). My goal is to have the lockup detection offer this same robustness!

[plantucha]

Jaebird and Rob P.

I'll be adding support to customize the threshold for both failover and lockup detection in the next day (setting to 0 will disable).  I'm at work all night, so I'm not sure if I'll get a chance to work on it tonight from  here (depends on how busy things get), or when I get home in the morning.

Here are the changes I am planning to make:
- Make --kill, --reload and --restart options for smartcoin (smartcoin.sh). This will allow you to take actions easily from custom scripts
- On lockup detection, fire custom script, and restart smartcoin automatically (instead of killing smartcoin). You should know if there is a real issue if this happens repeatedly (I'm still debating on whether smartcoin should automatically send emails on these events, or if that should be left up to the user with their custom script..  Thoughts?)
- Regarding the lockup routine when there is a loss of Internet, I think the current scheme (after above changes) will be plenty acceptable. Basically in this case, smartcoin will continually restart itself (unless you do a 'smartcoin --kill' from your custom script) about every 5 minutes until the Internet comes back.  This really won't waste any extra electricity, as the GPUs will be idle themselves with no work available for them. This also has the advantage of dealing with miner software that has locked up (as an example of such has already been shown here).  This also fits the philosophy of smartcoin in that once its running, it should try to automatically deal with things in a sane manner without manual intervention. (If I hadn't had to reboot my mining machine from time to time for testing purposes, I would wager that smartcoin has ensured me 0% downtime with no intervention from me other than initially setting up failover - its literally dealt with every problem automatically on my machine, and I've had to do nothing but check in every now and then). My goal is to have the lockup detection offer this same robustness!

it will be pretty good.
my miners are still killed about every 3 hours because of wrong lockdown detection.
After restart smartcoin all GPUS adding proof of work so GPU is not locked
I guess many reasons can be in. (internet down, server half way down, or something we don't know yet)
can you do more verbose log?

And why are running 8 minutes 2 profiles together until they got killed?
Don't they have go switch to profile 1 or 2 in seconds?

Code:

 
07/19/11 16:14:15       Update option selected

07/19/11 17:32:16       A change was detected in the failover system

07/19/11 17:32:16       Killing Miners....

07/19/11 17:32:20       Starting miner Miner.1!

07/19/11 17:32:20       Starting miner Miner.2!

07/19/11 17:32:20       Starting miner Miner.3!

07/19/11 17:32:20       Starting miner Miner.4!

07/19/11 17:32:20       Starting miner Miner.5!

07/19/11 17:32:20       Starting miner Miner.6!

07/19/11 17:32:20       Starting miner Miner.7!

07/19/11 17:32:20       Starting miner Miner.8!

07/19/11 17:32:20       Launching miner with launch string: python phoenix.py -v -u 

07/19/11 17:32:20       Launching miner with launch string: python phoenix.py -v -u h
07/19/11 17:32:20       Starting miner Miner.9!

07/19/11 17:32:20       Launching miner with launch string: python phoenix.py -v -u 

07/19/11 17:32:20       Starting miner Miner.10!

07/19/11 17:32:20       Launching miner with launch string: python phoenix.py -v -u 
07/19/11 17:32:20       Launching miner with launch string: ./cgminer -a 4way -t 6 -o 

07/19/11 17:32:20       Launching miner with launch string: python phoenix.py -v -u 
07/19/11 17:32:20       Launching miner with launch string: python phoenix.py -v -u 

07/19/11 17:32:20       Launching miner with launch string: python phoenix.py -

07/19/11 17:32:20       Launching miner with launch string: ./cgminer -a 4way -t 6

07/19/11 17:32:20       Launching miner with launch string: python phoenix.py -v -u htt


07/19/11 17:40:36       ERROR: It appears that one or more of your devices have locked up.  This is most likely the result of extreme overclocking!

07/19/11 17:40:36              It is recommended that you reduce your overclocking until you regain stability of the system

07/19/11 17:40:36       Killing Miners....

so -kill, --reload and --restart will be nice.

[Jen4538]

another question i have is

when i start smartcoin ,  it starts mining right away .  is there a way to stop the mining process while i reconfigure options or add more workers?  machine is very sluggish when its mining so i prefer it not to mine while i am busy changing things

i see i have to do more things today to make sure i am comfortable with all this.    my main machine had a headache this morning so i am fixing it before i can go back to smartcoin to try your examples.   and yes i am very curious why it hashs more with 2 workers to one gpu .   well i will take advantage of this either way

Jen

Jen

You should start getting used to accessing your machine via SSH from another computer (this is specifically what smartcoin is designed for). Your SSH session will not be slowed down like a local desktop is.

However, I do realize the need for an "Idle" profile, and I will probably add support for this in the next day or so.

Also, its a pretty well known side effect that running multiple instances per card gains a little hash speed (I get a 7% increase by running 5 instances per card). The reason as I understand it, is there are times the CPU waits idle for the GPU to be ready to accept information (caused by some lag in communication and OpenCL overhead). With multiple instances, the CPU has more of a chance with one of the  "threads" is waiting and ready to receive information from the CPU - hence the slightly higher hash rates.

i am lucky to even have this working without adding more pain to learning anything else .  if you can add idle command i can probably have this working downstaris .   without it i certainly cant do it due to the lag it creates as you try to make more workers , profiles etc.

Jen

[jondecker76]

Jaebird and Rob P.

I'll be adding support to customize the threshold for both failover and lockup detection in the next day (setting to 0 will disable).  I'm at work all night, so I'm not sure if I'll get a chance to work on it tonight from  here (depends on how busy things get), or when I get home in the morning.

Here are the changes I am planning to make:
- Make --kill, --reload and --restart options for smartcoin (smartcoin.sh). This will allow you to take actions easily from custom scripts
- On lockup detection, fire custom script, and restart smartcoin automatically (instead of killing smartcoin). You should know if there is a real issue if this happens repeatedly (I'm still debating on whether smartcoin should automatically send emails on these events, or if that should be left up to the user with their custom script..  Thoughts?)
- Regarding the lockup routine when there is a loss of Internet, I think the current scheme (after above changes) will be plenty acceptable. Basically in this case, smartcoin will continually restart itself (unless you do a 'smartcoin --kill' from your custom script) about every 5 minutes until the Internet comes back.  This really won't waste any extra electricity, as the GPUs will be idle themselves with no work available for them. This also has the advantage of dealing with miner software that has locked up (as an example of such has already been shown here).  This also fits the philosophy of smartcoin in that once its running, it should try to automatically deal with things in a sane manner without manual intervention. (If I hadn't had to reboot my mining machine from time to time for testing purposes, I would wager that smartcoin has ensured me 0% downtime with no intervention from me other than initially setting up failover - its literally dealt with every problem automatically on my machine, and I've had to do nothing but check in every now and then). My goal is to have the lockup detection offer this same robustness!

it will be pretty good.
my miners are still killed about every 3 hours because of wrong lockdown detection.
After restart smartcoin all GPUS adding proof of work so GPU is not locked
I guess many reasons can be in. (internet down, server half way down, or something we don't know yet)
can you do more verbose log?

And why are running 8 minutes 2 profiles together until they got killed?
Don't they have go switch to profile 1 or 2 in seconds?

Code:

 
07/19/11 16:14:15       Update option selected

07/19/11 17:32:16       A change was detected in the failover system

07/19/11 17:32:16       Killing Miners....

07/19/11 17:32:20       Starting miner Miner.1!

07/19/11 17:32:20       Starting miner Miner.2!

07/19/11 17:32:20       Starting miner Miner.3!

07/19/11 17:32:20       Starting miner Miner.4!

07/19/11 17:32:20       Starting miner Miner.5!

07/19/11 17:32:20       Starting miner Miner.6!

07/19/11 17:32:20       Starting miner Miner.7!

07/19/11 17:32:20       Starting miner Miner.8!

07/19/11 17:32:20       Launching miner with launch string: python phoenix.py -v -u 

07/19/11 17:32:20       Launching miner with launch string: python phoenix.py -v -u h
07/19/11 17:32:20       Starting miner Miner.9!

07/19/11 17:32:20       Launching miner with launch string: python phoenix.py -v -u 

07/19/11 17:32:20       Starting miner Miner.10!

07/19/11 17:32:20       Launching miner with launch string: python phoenix.py -v -u 
07/19/11 17:32:20       Launching miner with launch string: ./cgminer -a 4way -t 6 -o 

07/19/11 17:32:20       Launching miner with launch string: python phoenix.py -v -u 
07/19/11 17:32:20       Launching miner with launch string: python phoenix.py -v -u 

07/19/11 17:32:20       Launching miner with launch string: python phoenix.py -

07/19/11 17:32:20       Launching miner with launch string: ./cgminer -a 4way -t 6

07/19/11 17:32:20       Launching miner with launch string: python phoenix.py -v -u htt


07/19/11 17:40:36       ERROR: It appears that one or more of your devices have locked up.  This is most likely the result of extreme overclocking!

07/19/11 17:40:36              It is recommended that you reduce your overclocking until you regain stability of the system

07/19/11 17:40:36       Killing Miners....

so -kill, --reload and --restart will be nice.

What revision are you running?  I ask because there have been changes to the experimental branch that improves logging by also giving the output of the detected locked miner. Also, there was a couple of bugs fixed in the lockup system that look like they may be related

[jondecker76]

Nevermind, I found the problem.  I forgot to svn update on the development computer this morning before committing changes, so the fixes that I did previously got reverted! (I did the fixes from work last night)  So its the same exact problem, because the fixes that I did are gone!

I'll try to get them pushed back in!

[jondecker76]

Update r471e available!

- Fixes the reverted r457 and r458 changes! (I somehow managed to forget to update my development source before making new changes and committing them this morning, in effect reverting those changes that I made the night before).  This should once again make sure that lockup detection only runs on profiles not marked as "down", and improves the logging (again). Also restores the custom lockup script execution.

- Also went ahead and added the small change that a detected lockup no longer kills smartcoin, but restarts the miners instead.

[jaebird]

@jondecker76

I noticed the DB backup file in my .smartcoin dir is called: smarcoin.db.backup did you miss a "t" in smart?

[jondecker76]

@jondecker76

I noticed the DB backup file in my .smartcoin dir is called: smarcoin.db.backup did you miss a "t" in smart?

It appears I did!    I'll add that to the list of things to fix tomorrow

[jondecker76]

Update r488e now available
- --kill command line argument will kill smartcoin if running
- --delay command line argument will delay=# seconds before continuing with the rest of the program.  For example, if you ran:

Code:

smartcoin --kill --delay=15&

It would wait 15 seconds and then kill smartcoin (the & at the end makes the command do this in the background without haulting script execution).  This may be useful in your own custom scripts if you want to have some time to do things before killing smartcoin.
- New dynamic "Idle" profile.  From 5) Choose Profile, you can now select the special "Idle" profile to stop all hashing on all devices.  This will be useful for people who use the desktop to administer their system instead of SSH.
- Fixes the typo in the smartcoin.db backup file
- Word wrap is now turned off on the status screens (word wrap remains on for the control screen, however)
- The header is now much narrower to help with display on smaller devices. In the future I may add some additional logic to make a fuller display on normal 80-column terminals. Note: the new header display code uses the linux tput command to get the width of the terminal. I don't expect any dependency issues with this as I think it should be standard on pretty much all distros, but please post here if you get any errors related to "tput"

[krzynek1]

hello, in present state every profile that exists is added to failover order

how about situation when user want  to add only 3 of 5 existing profiles to failover order ?