Theterabyte: Great, much appreciated. I will implement this tomorrow.
Cheers
Topic: Starting a new FPGA mining farm/contract! Cognitive Resurrected on[Havelock] - page 40. (Read 300616 times)
March 07, 2014, 04:43:44 AM
March 07, 2014, 04:43:36 AM
It's been almost 2 days and we are well under 1.5TH. What's going on?
The low hash rate was once again because of power cycling and faulty firmware. As you can see, Garrett was able to get the units running again, so hopefully they will stay that way.
Theterabyte, thank you for your help! I believe this software will help us react more quickly to malfunctioning hardware, and stay as profitable as possible. This should be a very good month for Cognitive.
More updates to come,
Samuel
March 07, 2014, 03:59:45 AM
Garr - thanks for the update!
It's ugly as sin, but as we discussed, I whipped up something that I hope will help.
https://github.com/terabyte/cog-mon
This script should make it easy to set up monitoring to send an email/text when any miner drops below the desired threshold. It is written in ruby, and it scrapes the eligius website directly because their APIs don't show individual devices separately. You will want to modified the hardcoded values in the DEVICE_HASH variable and set the threshold/expected values to what you want. The README file explains how to run the script and gives an example crontab that will email/SMS you if it is broken, every 5 minutes. You might want the poll time longer, if you don't have free texts though =)
The next step will be to talk more about how we might have this script automatically trigger a restart of a specific miner that is failing, but to do that I need to find more time and chat a bit more about the current setup. Even as it is, you could imagine writing a script yourself (if you are fluent in those sorts of things) and just using the return code of the script itself to decide to restart something or not.
We'll chat more tomorrow I'm sure.
It's ugly as sin, but as we discussed, I whipped up something that I hope will help.
https://github.com/terabyte/cog-mon
This script should make it easy to set up monitoring to send an email/text when any miner drops below the desired threshold. It is written in ruby, and it scrapes the eligius website directly because their APIs don't show individual devices separately. You will want to modified the hardcoded values in the DEVICE_HASH variable and set the threshold/expected values to what you want. The README file explains how to run the script and gives an example crontab that will email/SMS you if it is broken, every 5 minutes. You might want the poll time longer, if you don't have free texts though =)
The next step will be to talk more about how we might have this script automatically trigger a restart of a specific miner that is failing, but to do that I need to find more time and chat a bit more about the current setup. Even as it is, you could imagine writing a script yourself (if you are fluent in those sorts of things) and just using the return code of the script itself to decide to restart something or not.
We'll chat more tomorrow I'm sure.
March 07, 2014, 03:23:38 AM
We're up and above 9th/s now from 8 terraminers. The warehouse we are currently in is maxed out power wise (which is why we took the scrypt miners offline to make room for another terraminer, which is much more profitable) and we are finalizing the process of moving into a proper tier-3 datacenter, albeit delayed. It is my promise that we will be hashing above 30th/s this month, so long as CoinTerra follows through on our deal (28th/s + 1% per day delivered past Jan 31).
Best,
Garrett
Best,
Garrett
March 07, 2014, 12:34:06 AM
Where has our scrypt hardware been mining? It seems we have been averaging under 2ths with our asics, what is being done about this? The first 2 weeks of owning hardware are the most profitable and our hardware has been offline more than on. Is there going to be further explanation on the dividend? For example why was it so much less than the week before? Yes the company grew by around 40% in terms of shares issued because of the conversion. But the div from the week before was for around 2 days of running the equipment was it not? This week we had even more equipment with Garrs claim that he had manage to fix some of the units and we had it all week long. Yet the dividend was less that half of the previous div. Thank you
FIMB is going to be repaid as promised to alincoin within an hour. I am very sorry for the delays here, they will not happen any longer.
Also, I just got two sCrypt rigs set up for Cognitive on Tompool.org. The mining output is viewable by looking for the address starting in "1cog".
Will respond to the rest of the questions in this thread later tonight, busy setting up more miners right now.
Cheers,
Garrett
Also, I just got two sCrypt rigs set up for Cognitive on Tompool.org. The mining output is viewable by looking for the address starting in "1cog".
Will respond to the rest of the questions in this thread later tonight, busy setting up more miners right now.
Cheers,
Garrett
We are mining at Tompool.org with address 1cogHCDW6ScMxuXQ5DNxS6LvLRVP18vBt or where but it seems where no longer hashing. When it was fully running we where getting around 1484 Kh/sec
March 06, 2014, 08:18:03 PM
It's been almost 2 days and we are well under 1.5TH. What's going on?
March 06, 2014, 07:40:06 PM
I sent theterabyte a PM right after seeing this post.
We are sorting out the final datacenter details and will be moving in ASAP. The remainder of our hardware will be arriving this week or next!
We are sorting out the final datacenter details and will be moving in ASAP. The remainder of our hardware will be arriving this week or next!
Hello Garr, I am looking forward receiving this month's installment of FIMB repayment, as per our agreement back in February. I believe I speak for other bondholders too which, unlike me, reportedly never received anything back.
It doesn't matter whether you can send 10BTC or 1BTC, but keeping up with your word and sending something at the beginning of every month goes a long way in demonstrating your compromise to our agreed terms.
So far I have received 38.56 BTC (50%). My BTC address is 1Bqg9YKE25U1BRCRm3pJiGsiFE5vvgirtB
March 06, 2014, 06:42:02 PM
PM received. I don't check bitcointalk.org PMs very frequently, so I'll move the convo to another medium, but just so you guys know, they did reach out to me right away, and the delay in response is from my end. I'll see if there is something I can do to help out and get our hashrate up. Thanks all - I'll keep you posted.
March 06, 2014, 05:19:53 AM
I sent theterabyte a PM right after seeing this post.
We are sorting out the final datacenter details and will be moving in ASAP. The remainder of our hardware will be arriving this week or next!
We are sorting out the final datacenter details and will be moving in ASAP. The remainder of our hardware will be arriving this week or next!
March 06, 2014, 02:38:52 AM
12 hour hashrate is currently at 2TH/s still =( Can you give us an idea of what it takes to make a malfunctioning unit function (i.e. 1. powercycle, 30 seconds 2. re-flash, 5 minutes, 3. run mining software 30 seconds - this is just an example, but what I would like to see)?
Is there really only a single unit functioning reliably right now? are none of the units functioning properly but one is up "on average"? Is cointerra doing something about this or are we expected to add automatic redundancy ourselves?
I am officially offering my technical services as a programmer, free of charge, to help out however I can. I think I already PM'd my phone number, but in case I haven't, PM me to arrange communications if there is anything I can do to help.
Is there really only a single unit functioning reliably right now? are none of the units functioning properly but one is up "on average"? Is cointerra doing something about this or are we expected to add automatic redundancy ourselves?
I am officially offering my technical services as a programmer, free of charge, to help out however I can. I think I already PM'd my phone number, but in case I haven't, PM me to arrange communications if there is anything I can do to help.
Samuel please react IMMEDIATELY to this more than a generous offer .
March 05, 2014, 08:29:43 PM
it seems silly to let 6-8Th/s of gear just lie around doing nothing most of the time. Right now we're a mining farm that doesn't mine.
Seriously. This.
Also thanks to theterabyte for the help.
March 05, 2014, 05:19:40 PM
12 hour hashrate is currently at 2TH/s still =( Can you give us an idea of what it takes to make a malfunctioning unit function (i.e. 1. powercycle, 30 seconds 2. re-flash, 5 minutes, 3. run mining software 30 seconds - this is just an example, but what I would like to see)?
Is there really only a single unit functioning reliably right now? are none of the units functioning properly but one is up "on average"? Is cointerra doing something about this or are we expected to add automatic redundancy ourselves?
I am officially offering my technical services as a programmer, free of charge, to help out however I can. I think I already PM'd my phone number, but in case I haven't, PM me to arrange communications if there is anything I can do to help.
Is there really only a single unit functioning reliably right now? are none of the units functioning properly but one is up "on average"? Is cointerra doing something about this or are we expected to add automatic redundancy ourselves?
I am officially offering my technical services as a programmer, free of charge, to help out however I can. I think I already PM'd my phone number, but in case I haven't, PM me to arrange communications if there is anything I can do to help.
March 05, 2014, 12:32:50 PM
According to the spreadsheet there is 10THs or so is there a problem with some of the equipment?
Yes, some of the units have faulty firmware, and have been shutting off when power cycles. Garrett is currently in negotiations with CoinTerra regarding this problem, so hopefully they will compensate Cognitive for this loss in hash power.
- Samuel
This is one thing Garr has done right in his dealings with CT so far. Please keep holding them responsible as far as our contract can (and ensure future contracts more clearly protect us against late/defective units). If cointerra isn't planning on dumping a pile of shovels on us then disapearing - if they want to be in the long-term shovel selling business - then they actually care about their reputation and want us to be happy. Let's hope anyways...
You are absolutely correct here. I have spoken with Ravi many times and it is very clear to me he is in this for the long-haul, and is all about creating and maintaining good relationships.
March 05, 2014, 12:29:41 PM
How about deploying remotely managed power distribution units? If it is a problem where the hardware hangs but works after a reboot, maybe being able to remotely powercycle the unut would be helpful.
Great idea!The only problem here is they can be quite pricey. It will be more practical to remotely soft-reboot, then if that doesn't work remotely upload firmware again.
March 05, 2014, 12:19:35 PM
How about deploying remotely managed power distribution units? If it is a problem where the hardware hangs but works after a reboot, maybe being able to remotely powercycle the unut would be helpful.
Great idea! March 05, 2014, 12:15:55 PM
How about deploying remotely managed power distribution units? If it is a problem where the hardware hangs but works after a reboot, maybe being able to remotely powercycle the unut would be helpful.
March 05, 2014, 12:06:44 PM
I get the sense that the hardware is not monitored, in the sense there is no one physically near it so that if it goes down it then stays down for a long time, until someone can get back to fiddle with it. At least that's what it looks like from the heartbreaking Eligius chart.
a) Is that true?
b) If so, can that be changed? Ideally things would run reliably on their own, but since they rarely do, it seems silly to let 6-8Th/s of gear just lie around doing nothing most of the time. Right now we're a mining farm that doesn't mine.
Peta-Mine is a larger operation, but they're also running mostly Cointerra machines and their graph is very different than ours.
a) Is that true?
b) If so, can that be changed? Ideally things would run reliably on their own, but since they rarely do, it seems silly to let 6-8Th/s of gear just lie around doing nothing most of the time. Right now we're a mining farm that doesn't mine.
Peta-Mine is a larger operation, but they're also running mostly Cointerra machines and their graph is very different than ours.
March 05, 2014, 11:41:30 AM
Where has our scrypt hardware been mining? It seems we have been averaging under 2ths with our asics, what is being done about this? The first 2 weeks of owning hardware are the most profitable and our hardware has been offline more than on. Is there going to be further explanation on the dividend? For example why was it so much less than the week before? Yes the company grew by around 40% in terms of shares issued because of the conversion. But the div from the week before was for around 2 days of running the equipment was it not? This week we had even more equipment with Garrs claim that he had manage to fix some of the units and we had it all week long. Yet the dividend was less that half of the previous div. Thank you
March 04, 2014, 04:18:42 PM
According to the spreadsheet there is 10THs or so is there a problem with some of the equipment?
Yes, some of the units have faulty firmware, and have been shutting off when power cycles. Garrett is currently in negotiations with CoinTerra regarding this problem, so hopefully they will compensate Cognitive for this loss in hash power.
- Samuel
This is one thing Garr has done right in his dealings with CT so far. Please keep holding them responsible as far as our contract can (and ensure future contracts more clearly protect us against late/defective units). If cointerra isn't planning on dumping a pile of shovels on us then disapearing - if they want to be in the long-term shovel selling business - then they actually care about their reputation and want us to be happy. Let's hope anyways...
March 04, 2014, 04:05:59 PM
Cognitive-PR, you may want to remove the two links in the footer of www.cognitivemining.com. One is MtGox, which is rumored to be having problems of some sort, and the other is to bitbank.me, which appears to have expired and is now owned by a domain squatter.
Guybrush01, thanks for the suggestion! The links have been removed.
Suggestion:
It would be helpful to tag the miners in the Eligius Pool to be able to track the status of each miner worker. By appending "_name_of_system" to the end of the 1cogxXyYU2EXEbdcV9j18PWenRU7gAjt1 address for the username, each miner can be tracked separately. Could help troubleshooting which miner has an issue by checking Eligius stats as well.
Here is an example - http://eligius.st/~wizkid057/newstats/userstats.php/1LxYXZpBaGTV11qFAy8DP2rHneBNRmQZ4s
By tagging the username, it shows the "Toggle Display of Worker Details"
It would be helpful to tag the miners in the Eligius Pool to be able to track the status of each miner worker. By appending "_name_of_system" to the end of the 1cogxXyYU2EXEbdcV9j18PWenRU7gAjt1 address for the username, each miner can be tracked separately. Could help troubleshooting which miner has an issue by checking Eligius stats as well.
Here is an example - http://eligius.st/~wizkid057/newstats/userstats.php/1LxYXZpBaGTV11qFAy8DP2rHneBNRmQZ4s
By tagging the username, it shows the "Toggle Display of Worker Details"
djjacket, this is excellent input. Garrett is going to set up this sort of system as soon as he can.
According to the spreadsheet there is 10THs or so is there a problem with some of the equipment?
Yes, some of the units have faulty firmware, and have been shutting off when power cycles. Garrett is currently in negotiations with CoinTerra regarding this problem, so hopefully they will compensate Cognitive for this loss in hash power.
- Samuel
Jump to: