Cointerra Terraminer IV Unboxing and Setup

[-ck]

Direct USB connections to the hashing boards isn't supported.  :-(

CoinTerra Support (CoinTerra)

Feb 17 10:47


Unfortunately using an external controller is not possible without opening the box, and that would void the warranty.

Darn.

[1713482694]

1713482694

[1713482694]

1713482694

[1713482694]

1713482694

[1713482694]

1713482694

[Alvi]

Edit: After some fiddling around (took cgminer down temporarily, changed the power setting around, restarted everything via SSH, etc.) I'm seeing 1.7Thash locally now instead of 1.5Thash and I've switched to Eligius...the hashrate is still climbing up at the moment so we'll see where it settles-maybe Eligius works better than BTCGuild with Terraminers?  I know Cointerra was testing on Eligius so maybe this is the only pool that really deals well with these.  We'll see soon enough I guess.

Can you provide some instruction on how to adjust power settings and restart via ssh. I'm on Eligus and getting about 1.4 on pool but loally shows about 1.6

[dbell]

After running over 24 hours on Eligius with its own payout address, the Eligius reported performance matches the local performance, about 1.62 TH.

I am still running into the problem where one of the "cores" will sometimes stop hashing and total hash rate will drops by about 100 GH both locally and on Eligus.  Manifests like this in the stats.

https://i.imgur.com/npJNrBn.png

ASIC1Core1 obviously the one not hashing.

https://i.imgur.com/N3jbVyC.png

A reset of the machine gets that downed core running again and the hash rate up above 1.6 TH again.

You can see in the Eligius plot the rate drop down about 100 GH about 4 hours before Feb 19th.  Not sure what causes this intermittent outage of a core for a long duration.  And it has always been this same core and similar symptoms.

https://i.imgur.com/5M4ac1f.png

... sent inquiry on this to Cointerra

[lucita777]

We are also seeing some issues with one of the CT miners:

The unhealthy miner:
http://imagizer.imageshack.us/v2/640x480q90/849/s1pm.png

And for comparison a healthy one:
http://imagizer.imageshack.us/v2/640x480q90/823/5opj.png

Miners are on different networks, and first miner's has been mining a little on the backup pool, so we are suspecting some network issues at the moment.
Local network ping to the miner was healthy all time, so now we are checking external network connection.
The miner has also completely stopped once and had to be rebooted. Unfortunately miner does not expose any logs, but only current stats

Thanks for sharing the issue with not hashing core, we will also try to look for that as well.

[-ck]

After running over 24 hours on Eligius with its own payout address, the Eligius reported performance matches the local performance, about 1.62 TH.

I am still running into the problem where one of the "cores" will sometimes stop hashing and total hash rate will drops by about 100 GH both locally and on Eligus.  Manifests like this in the stats.

[SNIP]

... sent inquiry on this to Cointerra

I had a quick chat to the person who can do something about this to hopefully elevate the priority of this issue.

[JoseSan]

Just to further the investigation, I'm also seeing this sort of thing happen. Here's a snippet of the stats page:



http://imgur.com/RCFFu6V.png

http://imgur.com/G3573bo.png

Restarting seems to fix it for about 8 hours.

[JoseSan]

Edit: After some fiddling around (took cgminer down temporarily, changed the power setting around, restarted everything via SSH, etc.) I'm seeing 1.7Thash locally now instead of 1.5Thash and I've switched to Eligius...the hashrate is still climbing up at the moment so we'll see where it settles-maybe Eligius works better than BTCGuild with Terraminers?  I know Cointerra was testing on Eligius so maybe this is the only pool that really deals well with these.  We'll see soon enough I guess.

Can you provide some instruction on how to adjust power settings and restart via ssh. I'm on Eligus and getting about 1.4 on pool but loally shows about 1.6

To restart via ssh and the linux command prompt:

# ssh root@[address of your miner]
<prompt for password = 'cointerra'>
# reboot

.. then wait for the machine to restart. It's very similar if you're using putty.

[byte1]

Can someone tell me how to reset the password for "admin" using SSH? I cannot seem to login anymore using the GUI interface. I still have SSH root  access.

I can do a master reset, but i want to avoid that process.

To Cointerra, when you give us an option to change a password, make sure you get it twice so people like me with fat fingers do not type in the wrong letters.


I have the same problem as others on this forum, I do not get full speed as advertised. One runs at 600GHash and the second board runs at 800 GHash.

What I paid for: 2TH
Performance claim after 2 months of delays: 1.6TH
What they finally delivered: 1.4TH

Thanks!

[JoseSan]

Just to further the investigation, I'm also seeing this sort of thing happen. Here's a snippet of the stats page:

http://imgur.com/RCFFu6V.png

http://imgur.com/G3573bo.png
Restarting seems to fix it for about 8 hours.

Interesting to point out is that these same cores that eventually stop working after a few hours are also responsible for errors while in full operation.


http://imgur.com/2g64nws.png

I plan to capture the output from /tmp/cgminer.log next time it stops hashing correctly. Currently this is what it looks like:
https://dl.dropboxusercontent.com/u/2009/cgminer.log

ckolivas, what must be done on a PC that can't be done on the Angstrom Linux/BeagleBone? If need be we could build a version of cgminer and plunk it into a running system.

byte1: /Angstrom/Cointerra/lighttpd.password defaults to the following on a fresh box:

admin:

I can't check at the moment if this changing this actually works.

[byte1]

JoseSan: I tried this, and it did not help.

# sudo su root ./lighttpd.password defaults

[-ck]

ckolivas, what must be done on a PC that can't be done on the Angstrom Linux/BeagleBone? If need be we could build a version of cgminer and plunk it into a running system.

I'm not sure what the question is. Cgminer should compile on the BBB the same as a PC since it's just running linux. The distributed firmware+cgminer on the beaglebone is just older than the code in master cgminer.

[JoseSan]

I'm not sure what the question is. Cgminer should compile on the BBB the same as a PC since it's just running linux. The distributed firmware+cgminer on the beaglebone is just older than the code in master cgminer.

Ok, in that case I'll just change out the binary at /opt/cgminer. Earlier in the thread you mentioned hooking up the units to a PC directly, and I was wondering why this was necessary. Re-reading things, I guess the reason was to isolate a potential problem with the BeagleBone itself, as well as to switch from the more liberal hashrate estimation method.

Unfortunately I haven't had any problem with a reported-vs-actual hashrate, so my doing this is probably useless.

byte1, perhaps I was a little unclear. Try this:

# cd /Angstrom/Cointerra/
# echo 'admin:' > ./lighttpd.password

[byte1]

byte1, perhaps I was a little unclear. Try this:

# cd /Angstrom/Cointerra/
# echo 'admin:' > ./lighttpd.password

JoseSan, That worked. Thank you!

[-ck]

Unfortunately I haven't had any problem with a reported-vs-actual hashrate, so my doing this is probably useless.

I have not bothered to change the hashmeter in the code yet since I can't even test it on a real device any more to see if I've changed the code correctly. I don't have a device and I don't even have remote access to one any more...

[testerx]

I'm really curious why some people see the same hashrate locally and on the pool despite us all using the same firmware and the same pools...the pools seem to show the same share submissions as the local panel so I'm really at a loss as to how there are such hugely different hash rates being calculated.  How are some people seeing 1.6 locally and 1.6 on the pool when I get these:
http://eligius.st/~wizkid057/newstats/userstats.php/1MwayhowetdFh7gQHVmKv3LYWUjfwsZ2B3 (my better December unit that claims over 1.6Thash locally...about 1.45Thash on Eligius, noticeably worse on btcguild)
http://eligius.st/~wizkid057/newstats/userstats.php/1MiHnbJ5Kiab9SskUGm5quyhx4aJGYRBYu (my January unit that likes to randomly drop one of the boards and definitely has weird connectivity or mining issues...if you're wondering it claims it's mining at 1.5Thash locally but Eligius sees 1.35Thash or so).

[Korbman]

So, when CTA0 shuts down and CTA1 is still mining how do I reboot the miner?   Just reboot CGMiner in the web interface?   Have you figured this out yet?
The fans really blow the heat a long way don't they?

I'm not sure when you would ever encounter this scenario, they basically mine as one unit, you don't direct one board to one pool or anything like that, all the hashrate is directed at whatever pool you point it at.  You can SSH in and reboot it that way and it seems to reboot everything.

I had this exact scenario this morning. Woke up to CTA0 offline, but CTA1 still going strong. I ssh'd into the unit and rebooted it, which seemws to bring everything back online (for now).

I am still running into the problem where one of the "cores" will sometimes stop hashing and total hash rate will drops by about 100 GH both locally and on Eligus.  Manifests like this in the stats.

[image snip]

ASIC1Core1 obviously the one not hashing.

Same problem, and same core for me. CoinTerra support noted their releasing new firmware soon that should help with this, but we'll see. It's bad enough I paid for 2TH/s but am only seeing 1.2.

[Azrael_PT]

So, when CTA0 shuts down and CTA1 is still mining how do I reboot the miner?   Just reboot CGMiner in the web interface?   Have you figured this out yet?
The fans really blow the heat a long way don't they?

I'm not sure when you would ever encounter this scenario, they basically mine as one unit, you don't direct one board to one pool or anything like that, all the hashrate is directed at whatever pool you point it at.  You can SSH in and reboot it that way and it seems to reboot everything.

I had this exact scenario this morning. Woke up to CTA0 offline, but CTA1 still going strong. I ssh'd into the unit and rebooted it, which seemws to bring everything back online (for now).

It happened whit me as well. I think my case is that the board that is next to the Power Source heats more.


I had to open my box to cool the terminal in the power source.


It has the ventilation hole but no fan to circulate the air. So the miner heats more and wen it gets to 95ºC it throttle down the miner.


CoreTemp2 is the one that gets to 95ºC


EDIT: Was probing the Terraminer and found the FTP port open.

[JoseSan]

It has the ventilation hole but no fan to circulate the air.

The power supply has a fan, it's just not visible. It *does* ventilate the air though rather poorly (the pictured end is the exhaust). The PSU is rated for something like 105C, so I wouldn't worry too much about *it* getting hot. Although some heat is radiated from the PSU to the heat exchangers / water radiators, it isn't a huge deal. If your cores are hitting 95C (the safety cutoff) than your ambient air temperature is too high, or your air circulation is poor.

[Minor Miner]

It has the ventilation hole but no fan to circulate the air.

The power supply has a fan, it's just not visible. It *does* ventilate the air though rather poorly (the pictured end is the exhaust). The PSU is rated for something like 105C, so I wouldn't worry too much about *it* getting hot. Although some heat is radiated from the PSU to the heat exchangers / water radiators, it isn't a huge deal. If your cores are hitting 95C (the safety cutoff) than your ambient air temperature is too high, or your air circulation is poor.

There is a lot of variability in performance and temp among the units.    I am pumping 50F air into the machines right now and some chips are at 60C and one is at 113C.   Some chips within the same unit have a spread of 30C so I would not assume his problem is just his ambient air/airflow

[JoseSan]

... and one is at 113C. 

That's mental, something is wrong. Only mil-spec or special analog chips are designed to operate that high. Someone didn't put your waterblock on right, the temperature sensor is wrong, or it skipped QA altogether. I'm not saying you're wrong, I'm just worried for your sake.