BAMT version 0.5 - Easy USB based mining Linux with farm wide management tools

[Splirow]

lodcrappo,

why is the /my-ip/mgpumon/ css not working properly as i stated on my previous post?

[lodcrappo]

lodcrappo,

why is the /my-ip/mgpumon/ css not working properly as i stated on my previous post?

Probably the  "httpcss":  setting in /etc/mgpumon.conf doesn't point to the right place.  

Edit - yep, looks like by default it points to /mgpumon.css which doesn't work

change to

"httpcss": "/bamt/mgpumon.css"

[Splirow]

lodcrappo,

why is the /my-ip/mgpumon/ css not working properly as i stated on my previous post?

Probably the  "httpcss":  setting in /etc/mgpumon.conf doesn't point to the right place.  

Edit - yep, looks like by default it points to /mgpumon.css which doesn't work

change to

"httpcss": "/bamt/mgpumon.css"

Yes!! that fixed it!!

When will we be able to access the bamt.conf/settings through the web interface?

thanks lodcrappo.
I will send 5 BTC to you by Friday

[lodcrappo]

I got 'boot disk failure' while POSTing message upon restart, does that mean my pendrive is already dying? I reinserted it and pressed enter and it's fine but I'm not sure if this is not gonna repeat when I am away and a GPU locks up.
So it's not actually related to BAMT but I'm really interested how long can a decent pendrive last (I'm using an 8GB QPI)

How long a usb key can last..  it really, really depends.

I am still using the very first key that ever held the very first BAMT, nearly a year ago.  It runs my test rig.  It's been rewritten with dozens of new images, its run buggy versions that did waaay to many disk i/os, its basically been tortured equal to probably 10 years of a normal BAMT system.  it works 100% fine.

OTOH,  I bought several 4gb keys at a local sprawlmart back when I was throwing new bamt rigs into my farm left and right.  Every one of them died in under 2 weeks time.

There is nothing special about my key that has lasted forever, its some totally generic 2gb jey.  The 4gb keys that died after a couple weeks were not any crappier looking or cheaper.

Some folks report that "quality" keys last longer.  Sandisk, name brand etc.  

[nathanghart]

Yep, you are correct... I just found it surprising that some people thought some large enterprises used reservations for all their end users... something I was interested in learning more about/discussing best practices, but completely off topic, so I apologize for that and will shut up now

A little off-topic, but why not share my experience in corporate IT. I would advise DHCP for everything except infrastructure (routers, firewalls, etc....) and most servers (only for servers in that have a static mapping for public access through the firewall to the DMZ are really needed). Then setup internal DNS servers/ and or WINS at each physical location that has an AD or LDAP server, depending on environment type. The DHCP servers will auto-update the DNS servers when a reservation is granted.  I have almost never advised the use of reservations except on my home network where I don't have an internal DNS server or when when implementing HA devices, such as load balancers, that use MAC-Address masquerading. It is limiting on migration, updating, and clustering, etc....

[Cobra]

Your overclocking was too high.

Your GPU locked up.

In order to make the best of the situation, BAMT has disabled O/C on that GPU and rebooted.

This allowed your rig to continue mining in a situation where otherwise it would have sat there locked up.  You're welcome.

Now you must reduce overclocking settings, remove the file in /live/image/BAMT/CONTROL/ACTIVE that has suspended o/c, and restart mining.

PS - Graphs and all recorded data are generated in ram only.  They are lost at every reboot.  This is to prevent your USB key from being destroyed.  You can run Munin on a regular linux box and gather stats/graphs from an unlimted # of bamt rigs and store them for as long as you want, but a USB based system cannot do this.

Worked perfectly, Thanks.  I'm glad to know that location to save me the time in case an overclock has issues again. Very happy with Phoenix2 as well, same hash rate but significantly reduced the invalids.

Edit: Actually it looks like I am wrong on the invalids.  on one rig 15900 shares, 0 invalid reported by gpumon. Strange thing is invalid number may show 1 for a second then go back to 0. This did not happen with Phoenix1. 2 rigs doing the same thing with Phoenix2.

[nathanghart]

Is there a way to check VRM temps without X?

[Inaba]

Ok, so I've been poking and prodding... prodding and poking, and then poking some more at BAMT.

I'm just having a devil of a time getting it stable under any circumstance.  I've installed it on 4 or 5 machines, and on all of them, if I use the default tools to mine, the systems constantly reboot themselves with "stuck" GPUs or for some other reason... seem to vary.

Anyway, if I then disable auto mining and disable all GPUs in BAMT, then open a cgminer session manually, the systems are rock stable and happily mine away. 

Why is Phoenix, Phoenix 2 and cgminer, when invoked via the bamt scripts so unstable/picky?  The system is stable when doing it's own thing, so it's not the hardware, and the systems were stable prior to bamt with a Ubuntu install... what is bamt doing that it keeps thinking the system has become unstable and forces a reboot when none is actually required?

[Splirow]

Ok, so I've been poking and prodding... prodding and poking, and then poking some more at BAMT.

I'm just having a devil of a time getting it stable under any circumstance.  I've installed it on 4 or 5 machines, and on all of them, if I use the default tools to mine, the systems constantly reboot themselves with "stuck" GPUs or for some other reason... seem to vary.

Anyway, if I then disable auto mining and disable all GPUs in BAMT, then open a cgminer session manually, the systems are rock stable and happily mine away. 

Why is Phoenix, Phoenix 2 and cgminer, when invoked via the bamt scripts so unstable/picky?  The system is stable when doing it's own thing, so it's not the hardware, and the systems were stable prior to bamt with a Ubuntu install... what is bamt doing that it keeps thinking the system has become unstable and forces a reboot when none is actually required?

Well, I have been mining on .5b since Monday night. And it is running strong without one problem. I have 1 rig with 3 5970 on a msi motherboard. and rig 2 has a ASUS motherboard with 1 5970.

I have them clocked at 825/300 and getting 382mh/s constant. No variance.

My temps range from 69c to 78c

Probably, you are doing something wrong with the bamt.conf file or over clocking too much.

[DeathAndTaxes]

I don't know but I wonder if some of us should finance a bounty for a cgminer only distro.  BAMT does a lot (overclocking, pool management, restarting sick GPUs, etc) which is already duplicated by cgminer.

Since BAMT is 32bit I had to install LinuxCoin (which hasn't been updated in what 9 months? ) for my 8GPU watercooled rig.  Just to get a test environment up and running quickly I just dropped LinuxCoin on a USB drive, installed cgminer, wrote a couple scripts to launch cgminer in screen on boot and edited a cgminer.conf.

Trust me when i say my Linux skills are "special" and not in a good way.  That "temporary" rig has been rock solid.  Not a peep just hashing away @ 3.2GH/s.  It has done so well I am reluctant to touch anything.

So maybe we just need a simpler "static" distro (I would pay extra for one which has simple steps to setup and host via PXE) which grabs config files and pushes API stats to a "server" for centralized management and monitoring. 

In essence dumb simple rigs which become little more than hashing appliances.

On edit: to clarify my jumbled thoughts.

[jamesg]

I don't know but I wonder if some of us should finance a bounty for a cgminer only distro.  BAMT does a lot (overclocking, pool management, restarting sick GPUs, etc) which is already duplicated by cgminer.

Since BAMT is 32bit I had to install LinuxCoin (which hasn't been updated in what 9 months? ) for my 8GPU watercooled rig.  Just to get a test environment up and running quickly I just dropped LinuxCoin on a USB drive, installed cgminer, wrote a couple scripts to launch cgminer in screen on boot and edited a cgminer.conf.

Trust me when i say my Linux skills are "special" and not in a good way).  That "temporary" rig has been rock solid.  Not a beep just hashing away @ 3.2GH/s.  It has done so well I am reluctant to touch anything.

So maybe we just need a simpler "static" distro (I would pay extra for one which has simple steps to setup and host via PXE) which grabs config files and sends statuses to a separate monitoring/management "server"?

+1

I will help fund this project.

[gnar1ta$]

If you don't like or think the restart script is causing stability problems, just turn it off.  Add detect_defunct: 0 to your bamt.conf.  I use cgminer exclusively now on 3 rigs and am still tweaking clocks, I've always had the script and all BAMT GPU control off and handle everything with cgminer.  I set 30 sec delay to start X and 0 seconds between GPU's - no issues so far with 16 GPU's.

[max in montreal]

have you tried the older version of bamt? maybe for you it might be more stable.

[Inaba]

Ok, I should have clarified that I have loaded systems... I'm sure bamt works great with 1 or 2 GPUs... but when it's trying to manage 6 - 7 GPUs it seems to be having trouble. 

I realize I can disable the stuff in bamt that's causing the problems (which is what I've basically done by just invoking cgminer directly), but I would like those options ... there's clearly something wrong with the scripts that are causing the (perceived) instability.  It's very seriously doubtful that I'm "doing something wrong," since I seem to be able to get bamt "stable" by disabling most of bamt.  I won't discount the possibility, but I'm open to suggestions. 

gnar1ta$- how are you starting the GPUs in your system if you're not using the bamt scripts?

[Splirow]

Ok, I should have clarified that I have loaded systems... I'm sure bamt works great with 1 or 2 GPUs... but when it's trying to manage 6 - 7 GPUs it seems to be having trouble. 

I realize I can disable the stuff in bamt that's causing the problems (which is what I've basically done by just invoking cgminer directly), but I would like those options ... there's clearly something wrong with the scripts that are causing the (perceived) instability.  It's very seriously doubtful that I'm "doing something wrong," since I seem to be able to get bamt "stable" by disabling most of bamt.  I won't discount the possibility, but I'm open to suggestions. 

gnar1ta$- how are you starting the GPUs in your system if you're not using the bamt scripts?

Like I told you before I have 3 5970's which is 6 GPU total on 1 rig. I have no issues.

To help you better, please answer the following questions:

What GPU's you have?

Are you using extenders?

please post your bamt.conf here so I can see them.

thanks

[gnar1ta$]

Ok, I should have clarified that I have loaded systems... I'm sure bamt works great with 1 or 2 GPUs... but when it's trying to manage 6 - 7 GPUs it seems to be having trouble. 

I realize I can disable the stuff in bamt that's causing the problems (which is what I've basically done by just invoking cgminer directly), but I would like those options ... there's clearly something wrong with the scripts that are causing the (perceived) instability.  It's very seriously doubtful that I'm "doing something wrong," since I seem to be able to get bamt "stable" by disabling most of bamt.  I won't discount the possibility, but I'm open to suggestions. 

gnar1ta$- how are you starting the GPUs in your system if you're not using the bamt scripts?

Code:

GPU 0:  75.0C 3403RPM | 297.7/297.0Mh/s | A:19040 R:454 HW:0 U: 2.09/m I: 8
 GPU 1:  72.0C 2464RPM | 297.7/297.9Mh/s | A:19219 R:432 HW:0 U: 2.11/m I: 8
 GPU 2:  73.0C 3524RPM | 297.1/297.1Mh/s | A:19162 R:472 HW:0 U: 2.11/m I: 8
 GPU 3:  74.0C 3730RPM | 295.3/295.3Mh/s | A:18542 R:432 HW:0 U: 2.04/m I: 8
 GPU 4:  74.0C 3182RPM | 295.4/295.3Mh/s | A:18668 R:448 HW:0 U: 2.05/m I: 8
 GPU 5:  73.0C 2783RPM | 297.7/297.8Mh/s | A:19197 R:440 HW:0 U: 2.11/m I: 8

I have 2 rigs with 6 single GPU's.  BAMT still launches them, I just use cgminer for everything.  Here is bamt.conf (GPU 0-5 are the same)

Code:

---
settings:

  miner_id: 6870
  miner_loc: miner3
  do_bcast_status: 1
  do_manage_config: 0
  start_mining_init_delay: 30
  start_mining_miner_delay: 0
  do_monitor: 0
  do_mgpumon: 0
  do_cgsnoop: 0
  detect_defunct: 0
 
  cgminer: 1
  cgminer_opts: --api-listen --submit-stale --failover-only -o http://xx -u xx -p xx -o http://xx -u xx -p xx -I 8 -g 1 --auto-gpu --gpu-engine 800-950 --gpu-memdiff -100 --auto-fan
  
# GPU Configurtation
gpu0:
  disabled: 0
  cgminer: 1
  monitor_temp_lo: 45
  monitor_temp_hi: 80
  monitor_load_lo: 80
  monitor_hash_lo: 125
  monitor_shares_lo: 1

[Splirow]

Ok, I should have clarified that I have loaded systems... I'm sure bamt works great with 1 or 2 GPUs... but when it's trying to manage 6 - 7 GPUs it seems to be having trouble. 

I realize I can disable the stuff in bamt that's causing the problems (which is what I've basically done by just invoking cgminer directly), but I would like those options ... there's clearly something wrong with the scripts that are causing the (perceived) instability.  It's very seriously doubtful that I'm "doing something wrong," since I seem to be able to get bamt "stable" by disabling most of bamt.  I won't discount the possibility, but I'm open to suggestions. 

gnar1ta$- how are you starting the GPUs in your system if you're not using the bamt scripts?

Code:

GPU 0:  75.0C 3403RPM | 297.7/297.0Mh/s | A:19040 R:454 HW:0 U: 2.09/m I: 8
 GPU 1:  72.0C 2464RPM | 297.7/297.9Mh/s | A:19219 R:432 HW:0 U: 2.11/m I: 8
 GPU 2:  73.0C 3524RPM | 297.1/297.1Mh/s | A:19162 R:472 HW:0 U: 2.11/m I: 8
 GPU 3:  74.0C 3730RPM | 295.3/295.3Mh/s | A:18542 R:432 HW:0 U: 2.04/m I: 8
 GPU 4:  74.0C 3182RPM | 295.4/295.3Mh/s | A:18668 R:448 HW:0 U: 2.05/m I: 8
 GPU 5:  73.0C 2783RPM | 297.7/297.8Mh/s | A:19197 R:440 HW:0 U: 2.11/m I: 8

I have 2 rigs with 6 single GPU's.  BAMT still launches them, I just use cgminer for everything.  Here is bamt.conf (GPU 0-5 are the same)

Code:

---
settings:

  miner_id: 6870
  miner_loc: miner3
  do_bcast_status: 1
  do_manage_config: 0
  start_mining_init_delay: 30
  start_mining_miner_delay: 0
  do_monitor: 0
  do_mgpumon: 0
  do_cgsnoop: 0
  detect_defunct: 0
 
  cgminer: 1
  cgminer_opts: --api-listen --submit-stale --failover-only -o http://xx -u xx -p xx -o http://xx -u xx -p xx -I 8 -g 1 --auto-gpu --gpu-engine 800-950 --gpu-memdiff -100 --auto-fan
  
# GPU Configurtation
gpu0:
  disabled: 0
  cgminer: 1
  monitor_temp_lo: 45
  monitor_temp_hi: 80
  monitor_load_lo: 80
  monitor_hash_lo: 125
  monitor_shares_lo: 1

I don't have
do_mgpumon: 0

what does it do? I use the web mgpumon without that on my settings

[gnar1ta$]

I don't have
do_mgpumon: 0

what does it do? I use the web mgpumon without that on my settings

Code:

# start mgpumon in background using /etc/bamt/mgpumon.conf? 
# (this will put the mgpumon web interface at http://this/mgpumon/
  do_mgpumon: 0

This one doesn't run mgpumon, it's set on another rig.

[lodcrappo]

Ok, I should have clarified that I have loaded systems... I'm sure bamt works great with 1 or 2 GPUs... but when it's trying to manage 6 - 7 GPUs it seems to be having trouble.  

I realize I can disable the stuff in bamt that's causing the problems (which is what I've basically done by just invoking cgminer directly), but I would like those options ... there's clearly something wrong with the scripts that are causing the (perceived) instability.  It's very seriously doubtful that I'm "doing something wrong," since I seem to be able to get bamt "stable" by disabling most of bamt.  I won't discount the possibility, but I'm open to suggestions.  

gnar1ta$- how are you starting the GPUs in your system if you're not using the bamt scripts?

There are a great many people using bamt with 6 or 7 GPUs.  Most of them use phoenix however.

The script simply starts cgminer im a screen session with the parameters you have supplied in the bamt.conf.  

It does nothing else.

It is not running at all once cgminer has been started.  It's only function is to start cgminer with your given parameters.

BAMT's dead gpu detection does not run if BAMT is in cgminer mode.

If you are observing different behavior when bamt starts cgminer compared to when you start it manually, I think you need to examine your bamt.conf and make sure the cgminer is getting the same parameters in both cases.  otherwise, there isn't any difference (other than being in screen? unless you do that manually too).

[lodcrappo]

I don't know but I wonder if some of us should finance a bounty for a cgminer only distro.  BAMT does a lot (overclocking, pool management, restarting sick GPUs, etc) which is already duplicated by cgminer.

And all of which is completely optional, using a simple config file.

Since BAMT is 32bit I had to install LinuxCoin (which hasn't been updated in what 9 months? ) for my 8GPU watercooled rig.  Just to get a test environment up and running quickly I just dropped LinuxCoin on a USB drive, installed cgminer, wrote a couple scripts to launch cgminer in screen on boot and edited a cgminer.conf.

The 32 vs 64 bit thing has been done to death.  For people that want 8 gpus per rig, bamt will not be an option.
However, given how little economic sense an 8 gpu rig makes, and therefore how rare they are, and with the impending switch to FPGA based mining, I just don't see this as a very big problem.

Trust me when i say my Linux skills are "special" and not in a good way).  That "temporary" rig has been rock solid.  Not a beep just hashing away @ 3.2GH/s.  It has done so well I am reluctant to touch anything.

So maybe we just need a simpler "static" distro (I would pay extra for one which has simple steps to setup and host via PXE) which grabs config files and sends statuses to a separate monitoring/management "server"?

How much will you pay?  It sounds like all you want is BAMT with the options disabled.   If you pay enough to make it worth my time, I would gladly generate any image type you want, PXE, 64bit, whatever.