jminer randomly gets stuck

[nathanowen42]

Jminer stops mining randomly, usually after a couple days of running. Has happened twice in the last week. Sadly there is no way to automatically detect this failure so I have to check it every day to ensure it has not stopped.

When it stops it simply does not print any more text. I can get it unstuck by pressing ctrl-c. When I do this I get the printouts shown in the attached image.

Any idea what might cause this or what can be done to fix this? Let me know if there is anything I can provide to help determine the cause of the issue.

I am launching with this in the run.bat:

java -jar -d64 -XX:+UseG1GC -Xms8G -Xmx12G burstcoin-jminer-0.4.10-SNAPSHOT.ja

If this issue cannot be fixed, could a timestamp be printed upon completion of each round so that it is easy to see if it is still working?

Relevant System Information:

JVM - build 1.8.0_131-b11
Windows 10 Pro - Version 1703
Nvida GeForce Driver Version 384.76
GTX 770
8 8TB plots
16GB Ram
8 core i7-7700K

Copy of my config file:
copy_of_jminer_conf_file.txt

Image of the resulting text after I press ctrl-c (much longer than pictured, next time I will copy the raw text, no useful information though so no great loss). Note that before I pressed ctrl-c it only showed the successfully printed block so there was no way to see that it was stuck without waiting for 5-10 minutes to see if it moved to the next block.

[de-luxe]

@nathanowen42 Thanks for reporting this, sadly i can just guess what may be your issue ... jminer has no function yet to prevent drives from spinning down / enter power save mode. Due to that Block #376543 was a slow block that took 18min. it may be the case, that some of your drives shut down due no access over a longer time. You should turn off all power saving settings related to drives, also on usb drives ensure they do not shut down after time ... I plan to add a 'trigger drive' feature in future versions, meanwhile you could try this https://forums.burst-team.us/topic/59/windows-keep-usb-drives-spinning-prevent-spindown-usb-drives

[nathanowen42]

Thanks for the reply. I will look into the drive power settings. I am using 8 internal SATA drives (WD80EFZX) all connected to 6GB SATA ports on a motherboard (ASUS P10S-M WS), boot drive is an M.2 NVME SSD. I had installed a raid card but have removed it as I was informed that 8 smaller plots on separate drives would perform better than one plot on a RAID 0 drive. Not the most economical system of course, but I want to use it as a server/NAS once I am done using it for Burst.

I am personally somewhat suspicious of my (admittedly somewhat outdated) graphics card. When I tried to use gpuPlotGenerator I would get an CL_INVALID_COMMAND_QUEUE after a few minutes. Though the correlation between the failure and the long round would better support your hypothesis.

Sometime in the next week or two, I plan to sit down and write a quick C program that selects on stdin and have it function as a watchdog. I will pipe the output of jminer to this program. If it gets no input after some timeout (say 30 minutes) it will automatically reboot the system (or maybe just restart jminer). Will probobly also throw in some logging and statistic display functionality. I will throw the code up on github once this is working.

[nathanowen42]

Update: I have gone into my advanced power settings and changed the hard drive power off to 'never'.
I still plan to program the aforementioned watchdog program just in case.

[aitorserra]

I also have this problem. I changed the energy plan for windows to see if that works. de-luxe perhaps you can add a wake-up option like blago have? Thank you!

[calxibe]

I have also had this issue on all versions (including 0.4.11), but it’s always stuck at wallet info https://pasteboard.co/GART5uL.png

This is sad because I love the miner, it’s blazing fast! But I can only use it when I’m watching the monitor (when I go away from the computer I switch to blago).

It stops after 2-3 blocks usually, sometimes up to 10 blocks before jMiner just hangs…

[de-luxe]

@calxibe Thats not stuck ... the 'Unable to get mining info' message just means, that getting the mining info from pool/wallet failed. The miner asks for it every 2 sec. (default) so if you get this message from time to time, it is not that big deal ... The miner would be stuck if you see that your pool contains to a new block and jminer does not start mining it.

[calxibe]

Well, after the message starts rolling (wallet is localhost btw), it never gets any new blocks.

Don't know what's causing it, it's a win2016 server with amd firepro gpu,
Have tried using less plots, hdd that never powers down etc.

Stuck for 1 hour: https://pasteboard.co/GB0Oi6b.png

[de-luxe]

@calxibe ensure you use java8 and not java9

[chrisf4lc0n]

@calxibe you are not plotting using the same GPU by any chance at the same time?

[chrisf4lc0n]

@nathanowen42 on Windows to keep the drives up all time you can try KeepAliveHD.