Smartcoin Linux mining administration. [MULTI-MACHINE SUPPORT NOW IN!] - page 4.

jondecker76

full member

Activity: 238

Merit: 100

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 <<>> 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.

plantucha

newbie

Activity: 56

Merit: 0

Quote from: Jessi on August 06, 2011, 03:04:45 PM

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

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.

Jessi

sr. member

Activity: 333

Merit: 250

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

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

newbie

Activity: 56

Merit: 0

Quote from: Jessi on August 06, 2011, 02:31:05 PM

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://bitcointalksearch.org/topic/modified-kernel-for-phoenix-15-7964

Jessi

sr. member

Activity: 333

Merit: 250

I've got 3xATI 6990 in my RIG.

plantucha

newbie

Activity: 56

Merit: 0

Quote from: Jessi on August 06, 2011, 11:00:20 AM

Quote from: jondecker76 on August 05, 2011, 09:34:42 PM

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: <<>>
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 <<>>. 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 <<>>.

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?

Jessi

sr. member

Activity: 333

Merit: 250

Quote from: jondecker76 on August 05, 2011, 09:34:42 PM

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: <<>>
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 <<>>. 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 <<>>.

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.

jondecker76

full member

Activity: 238

Merit: 100

Quote from: Jessi on August 05, 2011, 09:29:37 PM

I have copied from the linuxcoin-thread, to keep this where it shoud be. ;-)

Quote from: jondecker76 on August 05, 2011, 07:38:51 PM

Quote from: Jessi on August 05, 2011, 05:18:01 PM

have everything up and running since some days and I am very happy with LinuxCoin.

Really good work. Thank you for that!!

Anyway I still have some little questions.

One is regarding a little error which seem not to affect the mining process, but anyway I would like to know if this is normal behavior or if I can do anything to fix it.
During mining with smartcoin failover sometimes appearing this:

Code:

/opt/smartcoin/smartcoin_status.sh: line 151: [[: 10 10 10: syntax error in exp)

What does that mean?

Its caused from multiple entries being present in the database (this was from a bug, there should have only been one entry in the database). It is fixed in latest experimental releases, and I'd recommend that you update by:
1) Go to the control screen -> Edit Settings then change the development branch from 'stable' to 'experimental'
2) Use the Update function from the control screen - this should take give you a partial update to r607. Restart smartcoin and run the update again to go from r607 to the latest release. There are many many new features and improvements (including a completely revamped settings engine, multi-machine support, poclbm support and a lot more). Not only will you have the latest code, but your multiple database entries will be fixed in the process!

Feel free to post any smartcoin related questions in the smartcoin thread as I monitor it closely for support issues!

Thank you for your answer. I have done the update and the version is now 607s but the error is still there.

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

Jessi

sr. member

Activity: 333

Merit: 250

I have copied from the linuxcoin-thread, to keep this where it shoud be. ;-)

Quote from: jondecker76 on August 05, 2011, 07:38:51 PM

Quote from: Jessi on August 05, 2011, 05:18:01 PM

have everything up and running since some days and I am very happy with LinuxCoin.

Really good work. Thank you for that!!

Anyway I still have some little questions.

One is regarding a little error which seem not to affect the mining process, but anyway I would like to know if this is normal behavior or if I can do anything to fix it.
During mining with smartcoin failover sometimes appearing this:

Code:

/opt/smartcoin/smartcoin_status.sh: line 151: [[: 10 10 10: syntax error in exp)

What does that mean?

Its caused from multiple entries being present in the database (this was from a bug, there should have only been one entry in the database). It is fixed in latest experimental releases, and I'd recommend that you update by:
1) Go to the control screen -> Edit Settings then change the development branch from 'stable' to 'experimental'
2) Use the Update function from the control screen - this should take give you a partial update to r607. Restart smartcoin and run the update again to go from r607 to the latest release. There are many many new features and improvements (including a completely revamped settings engine, multi-machine support, poclbm support and a lot more). Not only will you have the latest code, but your multiple database entries will be fixed in the process!

Feel free to post any smartcoin related questions in the smartcoin thread as I monitor it closely for support issues!

Thank you for your answer. I have done the update and the version is now 607s but the error is still there.

jondecker76

full member

Activity: 238

Merit: 100

Updates r643s / r643e available
- Bringing stable branch current with experimental. While there are still a few known issues with the new multi-machine support, it is running very stable and I've had very few reports on new issues.
A note to people using the stable branch - You will have to run the update to get to r607 (it was a partial update with a breakpoint). Then restart smartcoin, and run the update again to get to r643 - restart smartcoin one last time and you're live with the newest additions!

jondecker76

full member

Activity: 238

Merit: 100

Quote from: plantucha on August 05, 2011, 05:49:10 PM

Quote from: jondecker76 on August 05, 2011, 05:24:09 PM

i ssh'd in, and it was running fine, so I must have missed it (or you may have reset it).
Just to test, I manually cycled through a few of your manual profiles and was unable to reproduce Sad

Could have just been a fluke? I'll do some failover tests my manually blocking out a pool with iptables on my machine just to make sure there isn't some new problem, but for now its hard to say what caused the problem you witnessed.

now i can see
I'm looking at screen with logs
not monitor screen

0-$ control 1$ localhost (2*$Log)

so i'm reporting feature not bug

My bad

before ctrl+a switched between control an monitor, now control and log

Aah, yep - thats the new log tab

I normally use ctrl+a 0, ctrl+a 1, ctrl+a 2, etc. to switch directly to a window. You can also use ctrl+a p for previous window, and ctrl+a n for next window.

A tip for the log window - you can use ctrl+a [ to switch to a scrolling mode where you can scroll through the log, then press escape when you are done with scrolling mode

plantucha

newbie

Activity: 56

Merit: 0

Quote from: jondecker76 on August 05, 2011, 05:24:09 PM

i ssh'd in, and it was running fine, so I must have missed it (or you may have reset it).
Just to test, I manually cycled through a few of your manual profiles and was unable to reproduce Sad

Could have just been a fluke? I'll do some failover tests my manually blocking out a pool with iptables on my machine just to make sure there isn't some new problem, but for now its hard to say what caused the problem you witnessed.

now i can see
I'm looking at screen with logs
not monitor screen

0-$ control 1$ localhost (2*$Log)

so i'm reporting feature not bug

My bad

before ctrl+a switched between control an monitor, now control and log

ctrl a 1
fixed it

jondecker76

full member

Activity: 238

Merit: 100

i ssh'd in, and it was running fine, so I must have missed it (or you may have reset it).
Just to test, I manually cycled through a few of your manual profiles and was unable to reproduce Sad

Could have just been a fluke? I'll do some failover tests my manually blocking out a pool with iptables on my machine just to make sure there isn't some new problem, but for now its hard to say what caused the problem you witnessed.

plantucha

newbie

Activity: 56

Merit: 0

lately , since about 2 versions back
monitor screen doesn't work well.

sometimes in case of failover it take to 1 minute to display 2 profiles,
it just showing starting miners.

and now miners are running ok.
server receiving everything but monitor screen stuck for 3 hours now
and showing only starting miners.

I didn't touch it so you can see yourself @ssh.

Clipse

hero member

Activity: 504

Merit: 502

great poclbm working

jondecker76

full member

Activity: 238

Merit: 100

Update r642e now available

- Small change in the smartcoin.xml file that is generated each refresh and placed in /tmp... There will now be an xml file for each machine, named "smartcoin..xml; i.e. smartcoin.1.xml, smartcoin.2.xml, and so on...

- Poclbm now working (poclbm underwent command-line paramater changes, and display changes since I had last seen it - therefore, only the "newer" poclbm versions which have these changes are supported.)

Note: I haven't yet updated the default launch string for new autodetections. Does anyone that uses poclbm have a suggestion for a generic sane default? I used the following launch string to test it out (anyone updating and trying to run poclbm will have to change their launch string to something similar via Configure Miners->Edit->Machine->poclbm)

Code:

python <#path#>poclbm.py -d <#device#> -v -w 128 -f0 http://<#user#>:<#pass#>@<#server#>:<#port#>

Clipse

hero member

Activity: 504

Merit: 502

Quote from: jondecker76 on August 04, 2011, 06:08:29 PM

After updating to the latest experimental release, you may want to try poclbm again, as I have disabled the <<>> message for now

poclbm doesnt display info in dashboard, just shows down Wink

jondecker76

full member

Activity: 238

Merit: 100

After updating to the latest experimental release, you may want to try poclbm again, as I have disabled the <<>> message for now

Clipse

hero member

Activity: 504

Merit: 502

Quote from: jondecker76 on August 04, 2011, 05:15:02 PM

Quote from: Clipse on August 04, 2011, 12:00:25 PM

Hey john

Have you been able to iron out the poclbm issue yet ? Cheesy

How do you update to experimental versions ?

I'm working midnights tonight (12 hour shifts), so I'll probably work on the poclbm issue in the morning when I get home (about 0630 EST.

To upgrade to the experimental version, go to the control screen, Edit Settings. There is a setting there for setting the development branch. Simply change it from 'stable' to 'experimental'. You can then go to the update option and run the update. (You'll likely have to run the update a 2nd time after restarting smartcoin, as there was a breakpoint update at r607)

Damn thats hectic hours, get rest rather

jondecker76

full member

Activity: 238

Merit: 100

Quote from: Clipse on August 04, 2011, 12:00:25 PM

Hey john

Have you been able to iron out the poclbm issue yet ? Cheesy

How do you update to experimental versions ?

I'm working midnights tonight (12 hour shifts), so I'll probably work on the poclbm issue in the morning when I get home (about 0630 EST.

To upgrade to the experimental version, go to the control screen, Edit Settings. There is a setting there for setting the development branch. Simply change it from 'stable' to 'experimental'. You can then go to the update option and run the update. (You'll likely have to run the update a 2nd time after restarting smartcoin, as there was a breakpoint update at r607)

Topic: Smartcoin Linux mining administration. [MULTI-MACHINE SUPPORT NOW IN!] - page 4. (Read 105029 times)