Pages:
Author

Topic: Smartcoin Linux mining administration. [MULTI-MACHINE SUPPORT NOW IN!] - page 13. (Read 105029 times)

full member
Activity: 238
Merit: 100
Here is some info on how to create your own custom "lockup.sh" script.


First, lets talk about rebooting..  Under Linux, normal users can't issue a reboot command without sudo and entering a password. While this won't be an issue for some as they already installed as root, for those that are running without root privileges it presents a problem as the  user might not be there when the script wants a password entered..  The way to allow a user to use the reboot command without a password is to do the following:
Code:
sudo visudo
This will open up a file in a special editor.  You need to add a line to the bottom of this file that looks like:
Code:
username ALL=NOPASSWD: /sbin/reboot
For example, my line looks like this:
Code:
jondecker76 ALL=NOPASSWD: /sbin/reboot

(press ctrl + x when finished editing the file)

This will allow a normal user to run the reboot command WITHOUT having to use a password (though the command must still be prepended by 'sudo')


Ok, now that you can reboot from the cli without a password, lets make a simple lockup script:
1) Create a file, lockup.sh in the smartcoin directory

2) Lets give it some contents:
Code:
#!/bin/bash

# Reboot the computer because of the lockup condition
sudo reboot

3) We need to make sure the script is executable
Code:
chmod +x ~/smartcoin/lockup.sh
(change the path accordingly, but most people install smartcoin to ~/smartcoin)

Thats it!  When a lockup is found, the computer will now reboot its self. Assuming that you have smartcoin setup to start automatically on startup, then the computer will reboot, and smartcoin will continue on its merry way.


Perhaps you only want the computer to reboot after 2 lockups. We just need to change lockup.sh to something like:
Code:
#!/bin/bash

if [[ -f "/tmp/failed_once" ]]; then
     # We created /tmp/failed_once so we must have already failed once. Lets reboot!
     sudo reboot
else
     # /tmp/failed_once has not been created yet. Lets create it now so that if another lockup happens, we will know its the 2nd time!
     touch /tmp/failed_once
fi
Since the /tmp directory is cleared each reboot, once the reboot happens this will "reset" its self to needing 2 lockups to force a reboot again.



Perhaps you want to send yourself an email or text message when this happens so you can investigate. There are many ways to send an email, but I'll use sendemail for the example. (Most cellular carries have special email addresses for your account that will forward them to your phone as a text message, this allows you to send text messages via commandline email programs!)  For this example, I'm also using the gmail smtp relay (you need a gmail account to use their smtp relay, though you can use your ISP's smtp relay if you wish)

first, install the sendemail program and some dependencies:
Code:
sudo apt-get install sendemail libnet-ssleay-perl libio-socket-ssl-perl

Now, we just need to add some email stuff to the script:
Code:
#!/bin/bash

if [[ -f "/tmp/failed_once" ]]; then
     # Lets email ourselves!
     sendemail -f "[email protected]" -t "[email protected]" -u "Lockup Detected!" -m "A lockup event was detected!  Machine is automatically rebooting!" -s "smtp.gmail.com" -o tls=yes -xu your_gmail_username -xp your_gmail_password
     # We created /tmp/failed_once so we must have already failed once. Lets reboot!
     sudo reboot
else
     # Lets email ourselves
     sendemail -f "[email protected]" -t "[email protected]" -u "Lockup Detected!" -m "A lockup event was detected!  Smartcoin is going to restart the miners now, but if another lockup happens we will reboot!" -s "smtp.gmail.com" -o tls=yes -xu your_gmail_username -xp your_gmail_password
     # /tmp/failed_once has not been created yet. Lets create it now so that if another lockup happens, we will know its the 2nd time!
     touch /tmp/failed_once
fi
Notes:
- Replace '[email protected]' with the email address you wish to receive the message. 
- the from (-f) parameter can contain any email address, even your own.  Of course, [email protected] is just an example - its doesn't exist, but it will work
- If you use a different smtp server than google, you  may or may not need the "-o tls=yes" part, depending on what smtp server you use.


One last thing.. I mentioned text messages.  Here is a brief list of carriers and what email address you need to send to for you to receive a text on your cell phone:
Quote
sr. member
Activity: 383
Merit: 250
plantucha:
regarding your stress test - it looks like everything is happening exactly as it should.  Your GPUs are locked up, and when the restart of the miners happens, they just hang (resulting in the IDLE message)  This is where in your lockup.sh script, you should do a reboot of the system and send yourself an email (perhaps by creating a file the first time, and if it happens again, then do the reboot even)  Even though it was GPU3 that tripped the lockup condition, during those 7 minutes the other GPU's probably had locked up as well.


Can you post sample of lockup.sh script please?

I will probably manage to reboot it.
But create file and sending email is out of my possibilities.  Cry

+1 to that.

Maybe you could include a "sample.lockup.sh" script, that we can remove the "#" char's from the lines we want to run. Then all we have to do is either use it as an example to create our own, or just rename to "lockup.sh".

Maybe include it in the next stable update?
newbie
Activity: 56
Merit: 0
plantucha:
regarding your stress test - it looks like everything is happening exactly as it should.  Your GPUs are locked up, and when the restart of the miners happens, they just hang (resulting in the IDLE message)  This is where in your lockup.sh script, you should do a reboot of the system and send yourself an email (perhaps by creating a file the first time, and if it happens again, then do the reboot even)  Even though it was GPU3 that tripped the lockup condition, during those 7 minutes the other GPU's probably had locked up as well.


Can you post sample of lockup.sh script please?

I will probably manage to reboot it.
But create file and sending email is out of my possibilities.  Cry
full member
Activity: 238
Merit: 100
plantucha:
regarding your stress test - it looks like everything is happening exactly as it should.  Your GPUs are locked up, and when the restart of the miners happens, they just hang (resulting in the IDLE message)  This is where in your lockup.sh script, you should do a reboot of the system and send yourself an email (perhaps by creating a file the first time, and if it happens again, then do the reboot even)  Even though it was GPU3 that tripped the lockup condition, during those 7 minutes the other GPU's probably had locked up as well.



Regarding the hanging update, this can happen if googleCode's server is down temporarily or reacting slowly.  I've had it hang there for several minutes before being able to contact google's svn servers sometimes. (it does more than just an svn update, it gets other information such as specific revision number of the version file, the head and stable revision numbers, etc. so sometimes it does take a while) I should probably ping the server before I attempt to do the svn update, and tell the user to try again later.  When this happens to me, i usually ctrl+c to kill the control script, then hit 'o' to open it again, then try the update again.
newbie
Activity: 56
Merit: 0
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"


I can't upgrade  r488
Miner is mining but upgrade just hangs.
If I kill it + restart, smartcoin is working fine until I'll try upgrade to R488
then it hangs again.

Code:
Smartcoin r479(experimental) Wed Jul 20 04:43:30 EDT 2011
---------------------------------------------------------
Getting svn information. Please be patient...
Your current version is r479(experimental).
The current experimental version is r488
The current stable version is r452
Are you sure that you wish to perform an update?
y

Bring update script up to current...

Bring helper functions up to current...

manual svn update works fine

Code:
/smartcoin$ svn update http://smartcoin.googlecode.com/svn/branches/beta1/smartcoin ./  Skipped 'http://smartcoin.googlecode.com/svn/branches/beta1/smartcoin'
U    smartcoin_control.sh
U    smartcoin.sh
Conflict discovered in 'smartcoin_ops.sh'.

Select: (p) postpone, (df) diff-full, (e) edit,
        (mc) mine-conflict, (tc) theirs-conflict,
        (s) show all options: df
--- .svn/text-base/smartcoin_ops.sh.svn-base    Tue Jul 19 22:48:02 2011
+++ .svn/tmp/smartcoin_ops.sh.5.tmp     Wed Jul 20 04:54:11 2011
@@ -15,11 +15,25 @@
 fi
 . $CUR_LOCATION/sql_ops.sh

+# GLOBALS
+Q="SELECT value FROM settings WHERE data='dev_branch';"
+R=$(RunSQL "$Q")
+branch=$(Field 1 "$R")
+if [[ "$branch" == "stable" ]]; then
+       G_BRANCH="$branch"
+       G_BRANCH_ABBV="s"
+elif [[ "$branch" == "experimental" ]]; then
+       G_BRANCH="$branch"
+       G_BRANCH_ABBV="e"
+fi
+
+
+
 # SVN STUFF
 GetRevision() {
   Q="SELECT value FROM settings WHERE data='dev_branch';"
   local branch=$(RunSQL "$Q")
-  echo $(svn info $CUR_LOCATION/ | grep "^Revision" | awk '{print $2}')"($branch)"
+  echo $(svn info $CUR_LOCATION/ | grep "^Revision" | awk '{print $2}')
 }

 GetRepo() {
@@ -88,8 +102,16 @@
 }

 ShowHeader() {
-       echo "smartcoin Management System r$REVISION"    $(date)
-       echo "--------------------------------------------------------------------------------"
+<<<<<<< .mine
+       echo "Smartcoin r$REVISION"    $(date)
+       echo "---------------------------------------------------------"
+=======
+       echo "Smartcoin r$REVISION$G_BRANCH_ABBV" $(date "+%T")
+       local cols=$(tput cols)
+       for i in $(seq $cols); do echo -n "-"; done
+
+       #echo "----------------------"
+>>>>>>> .r488
 }


@@ -227,6 +249,9 @@
                # Generate the FieldArray via DonateProfile
                Log "Generating Donation Profile"
                FieldArray=$(GenDonationProfile "$thisMachine")
+  elif [[ "$thisProfile" == "-4" ]]; then
+    # Generate a blank Field array for the "idle" profile
+    FieldArray=""
        elif [[ "$thisProfile" == "-3" ]]; then
                # Generate the FieldArray via Failover
                FieldArray=$(GenFailoverProfile "$thisMachine")
@@ -569,6 +594,8 @@

        if [[ "$Donate" ]]; then
                echo "Donation (via AutoDonate)  - $Donate minutes remaining."
+       elif [[ "$thisProfile" == "-4" ]]; then
+               echo "Idle"
        elif [[ "$thisProfile" == "-3" ]]; then
                echo "Failover"
        elif [[ "$thisProfile" == "-2" ]]; then
Select: (p) postpone, (df) diff-full, (e) edit, (r) resolved,
        (mc) mine-conflict, (tc) theirs-conflict,
        (s) show all options: mc
G    smartcoin_ops.sh
Updated to revision 488.
Summary of conflicts:
  Skipped paths: 1
newbie
Activity: 56
Merit: 0
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"


Idle problem
Stress test:
1. I overclocked too high 950 instead usual 900
2. in 2 minutes  G3 lockup
3. in 7 minutes lockup detection killed miners
4. and restart them
5. but to <> profile except CPU miner
6. hanging in IDLE


Code:
 
~$ aticonfig --od-setclocks=950,1050 --adapter=all

Adapter 0 - AMD Radeon HD 6700 Series
            New Core Peak   : 950
            New Memory Peak : 1050

Adapter 1 - AMD Radeon HD 6700 Series
            New Core Peak   : 950
            New Memory Peak : 1050

Adapter 2 - AMD Radeon HD 6700 Series
            New Core Peak   : 950
            New Memory Peak : 1050

Adapter 3 - AMD Radeon HD 6700 Series
            New Core Peak   : 950
            New Memory Peak : 1050
michal@rig:~$ smartcoin
Starting SmartCoin at location: /home/michal/smartcoin...
Running check
Re-attaching to smartcoin...
Smartcoin r479(experimental) Wed Jul 20 04:21:29 EDT 2011
---------------------------------------------------------
Host: localhost
G0: Temp °C: 71.00 Load: 99%
G1: Temp °C: 72.00 Load: 99%
G2: Temp °C: 72.00 Load: 99%
G3: Temp °C: 55.00 Load: 99%
CPU Load: 0.78%

Profile: Failover
--------BTCGuild--------
G0:     [219.11 MHash/s] [19 OK] [0 BAD] [0% BAD]
G1:     [219.00 MHash/s] [18 OK] [0 BAD] [0% BAD]
G2:     [219.10 MHash/s] [23 OK] [0 BAD] [0% BAD]
G3:     [207.52 MHash/s] [2 OK] [0 BAD] [0% BAD]
CP:     [20.8 MHash/s] [1 OK] [0 BAD] [0% BAD]
Total : [885.53 MHash/s] [63 OK] [0 BAD] [0% BAD]

Grand Total : [885.53 MHash/s] [63 OK] [0 BAD] [0% BAD]

Code:
07/20/11 04:30:50       ERROR: It appears that one or more of your devices have locked up.  This is most likely th$

07/20/11 04:30:50              It is recommended that you reduce your overclocking until you regain stability of t$

07/20/11 04:30:50              Below is a capture of the miner output which caused the error:

07/20/11 04:30:50       plantucha_gpuasdfghbtcguild.com8332BTCGuild
LAUNCH: python phoenix.py -v -u http://plantucha_gpu:[email protected]:8332/ -
k phatk device=0 worksize=64 vectors aggression=13 bfi_int fastloop=false
Launching miner with launch string: python phoenix.py -v -u http://plantucha_gpu
:[email protected]:8332/ -k phatk device=0 worksize=64 vectors aggression=13 b
fi_int fastloop=false

07/20/11 04:30:50       User lockup script found. Running lockup script.

07/20/11 04:30:50       Killing Miners....

07/20/11 04:30:54       Starting miner Miner.1!

07/20/11 04:30:54       Starting miner Miner.2!

07/20/11 04:30:54       Starting miner Miner.3!

07/20/11 04:30:54       Starting miner Miner.4!

07/20/11 04:30:54       Starting miner Miner.5!

07/20/11 04:30:55       Launching miner with launch string: python phoenix.py -v -u http://plantucha_gpu:asdfgh@bt$

07/20/11 04:30:55       Launching miner with launch string: python phoenix.py -v -u http://plantucha_gpu:asdfgh@bt$

07/20/11 04:30:55       Launching miner with launch string: python phoenix.py -v -u http://plantucha_gpu:asdfgh@bt$

07/20/11 04:30:55       Launching miner with launch string: python phoenix.py -v -u http://plantucha_gpu:asdfgh@bt$

07/20/11 04:30:55       Launching miner with launch string: ./cgminer -a 4way -t 6 -o http://btcguild.com:8332/ -u$





Code:
Smartcoin r479(experimental) Wed Jul 20 04:29:59 EDT 2011
---------------------------------------------------------
Host: localhost
G0: Temp °C: 38.00 Load: 0%
G1: Temp °C: 40.00 Load: 0%
G2: Temp °C: 38.00 Load: 0%
G3: Temp °C: 55.00 Load: 99%
CPU Load: 0.78%

Profile: Failover
--------BTCGuild--------
G0:     <<>>
G1:     <<>>
G2:     <<>>
G3:     <<>>
CP:     [17.8 MHash/s] [1 OK] [0 BAD] [0% BAD]
Total : [17.8 MHash/s] [1 OK] [0 BAD] [0% BAD]

Grand Total : [17.80 MHash/s] [1 OK] [0 BAD] [0% BAD]

full member
Activity: 238
Merit: 100
I'll finish the multi-machine support probably in about a week or just a bit longer (at least a rough initial implementation).  Most of what is needed is already in place, just hasn't been turned on yet.  I'll be doing this work in a special development branch so that I won't make the experimental branch users pull their hair out

One thing I am missing is an extra machine to test with, so if anyone has a spare miner laying around for me to test with and wouldn't mind giving me some ssh access to it for the next week or so and putting up with some down time on the machine, shoot me a PM!
newbie
Activity: 22
Merit: 0
After the latest updates things have been looking great for me (with perhaps only recent high level or stales, but that most probably is Phoenix r112's fault). Anyways I was wondering when will you be implementing multi-machine support. I plan on deploying a couple new rigs in the near future and would ABSOLUTELY LOVE to be able to monitor and control everything from one place Tongue
newbie
Activity: 41
Merit: 0
in my case maybe it was issue with no swap file on usb

now on hdd, there is no problem to start 20+ workers at one time

thanks for reply
full member
Activity: 238
Merit: 100
Newer revisions of smartcoin no longer have this problem, as the failover data in the database is cleared each restart now, so it will always start now with only one profile loaded.

Also,  I have tested 30 instances running with no problems on an old P4 miner with 1GB of memory.  If it ever becomes an issue, I can probably make a setting where you can specify the maximum levels deep the failover system will go, but I don't forsee any problems for now at least
newbie
Activity: 41
Merit: 0
reason behind that is when launching so many instances per card may crash system i think

previously my system was installed on usb flash and when smartcoin was launching many profiles my system wasnt responding any more

now im on hdd, maybe that situation will never happen again
full member
Activity: 238
Merit: 100
When specifying failover order (option 12 on the control screen), you don't have to specify every profile on the list (for example, if it shows the current failover order of 15 profiles, you can still just enter 3 comma-separated values to specify the failover order of those 3 profiles.  Any profiles not specified in the comma-separated list will automatically be added to the end of the failover order, in the order that the profile was created.

I don't see a reason in excluding a profile from the failover system.  I have seen my own system go 4 levels deep into the failover system while multiple pools were being ddos'd.  Surely having somewhere to put your hashes is better than having nowhere to put them?  I guess its possible to add code to not include specially marked profiles to the failover, I just don't see a real reason why anyone would need to do this - though please share your thoughts, perhaps there are reasons that I'm not thinking of
newbie
Activity: 41
Merit: 0
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 ?

full member
Activity: 238
Merit: 100
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"
full member
Activity: 238
Merit: 100
@jondecker76

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


It appears I did!    I'll add that to the list of things to fix tomorrow Smiley
member
Activity: 79
Merit: 10
@jondecker76

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

full member
Activity: 238
Merit: 100
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.
full member
Activity: 238
Merit: 100
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!
full member
Activity: 238
Merit: 100
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

full member
Activity: 167
Merit: 100
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
Pages:
Jump to: