Topic: Terraminer IV - drops to 50% hashing after 24 - 30 hours... (Read 2833 times)

In what way does power stepping work?  Can you get better performance from problem machines by stepping down the power setting?  Details would be appreciated.

I have a Group Buy with a TerraMiner that likes to get hot and slow down.....I have our TM in a datacenter and I have a VPN from my house to the machine......I just remote in to my house from my iPhone and check on the TM....Its a pain in the A, but that way I can stay on top of the hash rate no matter where I am...(as long as I have internet/cell coverage)

Basically.  I am not going to speculate here but it is not "hacked".   Power stepping does work.  I have a hosting business and many of these units.  I have seen 4 different issues with these units that have similar symptoms. 


.02BTC
-D

Well, I've had the problem twice now over the last 5 days.  Both times I brought up the Terraminer admin page and selected Restart System.  Within 4 -5 minutes the machine was back up to the normal 1.6Th/s rate.  This seems like a workable solution when I am out of town since I can use TeamViewer to log into a local machine and execute the restart.

Thanks for all the help!

hope you get it fixed

No need for this step 2., the next one (3.) will still do this (4 is a factor of 12), but there's no harm in doing it
Nothing other than a reboot. The file is saved, but the *actual* crontab isn't loaded yet, so a reboot is necessary.
This isn't necessary either; again, if you do the following command in step 3., it doesn't matter if you perform step 2. or not.

I agree, my original post was a little terse. The unix way of *appending* the output of a program to a file is the '>>' syntax. To overwrite the entire contents of a file with the output of a program, you use the single '>' syntax. 'echo' is just a way to send some text to the "standard output" so that the '>>' or '>' syntax works.

If you run the command 'crontab -l' you'll get the output of what the system is currently using for cronjobs ("chronological jobs"). You'll notice that after you run 'echo '0 0 * * * /sbin/reboot' >> /Angstrom/Cointerra/crontab.root', you'll only see one line from 'crontab -l' before rebooting (the original one put in by Cointerra). *After* rebooting, it should have loaded the contents of what's in the file /Angstrom/Cointerra/crontab.root, which should contain the new reboot jobs.

Clear as mud? PM me if you need further clarification. Power cycling electronics isn't good, but this shitty hack might very well be worth the mild hurt.

I have a theory about the hashrate dropping... I think the PSU's are dying. For those of you suffering this problem can you tell me if your units have normal or reverse flow cooling on the PSU?

You see it seems like cointerra bought both reverse flow and normal flow power supplies and they also rely purely of the PSU fans to cool them. These PSU's are working pretty hard and might be overheating and causing the hashrate to drop. You can tell if your PSU's are reverse flow if there is an extra bit of metal sticking out the back of your case around where the power cables plug in. This is to prevent hot air being sucked back into the PSU's.

You can drastically improve the cooling of your PSU's by adding another fan on the side of the case where there is an exhaust/intake hole for reverse/normal PSU's. There is a spare fan power socket at the rear of the boards you can plug the new fan straight into there and just tape it onto the outside of your machine, an easy modification.

I think it is a good idea to include this extra fan no matter what airflow PSU's you have but if you have the reverse flow type I would say it is very important to do so as the PSU's will be alot hotter in this configuration.

@ waldohoover, As you have six machines and only there are dying can you tell us if the three that are dying are reverse or normal flow compared to the three that are alive? This would be very interesting to know.

Not me, I am great at fixing thing, but not much for coding.

I am just more annoyed my network is vulnerable to attack should they get entry. I have things locked down to make it difficult, but hackers these days, if they want it they will get it.

I tried to build it and patch it myself, however its above my ability. If it was a normal distribution it would be simple.

The question is why have they not patched SSL and released a Firmware.

No, just do this to reset to the default (what was in the file originally):
echo '*/5 * * * * /opt/restart_cgminer.sh' > /Angstrom/Cointerra/crontab.root

To have any of these changes take effect, you need to reboot.

I edited my original answer. A decent summary of cronjobs (and the crontab, which is indirectly what we're editing) can be found (via google) here:

http://www.cyberciti.biz/faq/how-do-i-add-jobs-to-cron-under-linux-or-unix-oses/

Do this instead:
1) ssh into the box, then issue this command:
 
echo '0 0 * * * /sbin/reboot' >> /Angstrom/Cointerra/crontab.root

If you want to reboot every 4 hours, do this instead:

echo '0 0,4,8,12,16,20 * * * /sbin/reboot' >> /Angstrom/Cointerra/crontab.root

This is totally untested, but I'd be surprised if it didn't work.

Thanks.  I am going to try the power step / reboot option when I get home. 

Which company? I think you forgot to name it....? 

If you really want to remote power on/off your miner, this company makes Web Power Switches that connect to your Ethernet network and has 10 outlets and you login to a web page and control your outlets remotely.

waldohoover - so are you saying that a remote reboot will get the TM back up and running at full hashing rate and that I don't need to physically unplu and replug in the PSU's?

Are you saying this is a "known" issue?.....if so, that sucks.  Since I am out of town 4 - 5 days a week it makes reseting these things difficult.  I guess I need to come up with some type of solution....

It could be one of two things, either one of your power supplies is tripping its breaker or your miner has been hacked since cointerras operating system is vulnerable to the HEARTBLEED malware.  To tell if your miner has been hacked, on the status page of the cointerras GUI if you see only CTA0 or only CTA1, almost as if its been erased from the GUI than you have been hacked.  I know this because it's happened to me and apparently alot of other cointerra customers according to the Cointerra support forum.  If that is the case than contact Cointerra customer support and they will walk you through on how to reset the embedded Linux controller, it's time consuming and you have to be comfortable with working on Linux.

Hi,

I have a Terraminer IV and have been running it for about 3 weeks.  I've noticed that after about 24 - 30 hours it drops to 50% hashing.  If I restart it (unplug both PSU's and then plug them back in) it resets and runs fine for about another 24- 30 hours.  It has been doing this consistently for about 10 days now.

Has anyone else experienced this problem?  Any ideas or suggestions?

Thanks.