I’d love some help. I’m a newbie trying my hand at plotting using the ChiaDecentral instructions - basically this:
screen -d -m -S chia1 bash -c 'cd /home/XXX/chia-blockchain && . ./activate && sleep 0h && chia plots create -k 32 -b 4000 -e -r 4 -u 128 -n 16 -t /mnt/ssd1/temp1 -2 /mnt/ssd1 -d /mnt/hdd1 | tee /home/XXX/chialogs/chia1_1_.log'
System has been churning for 24 hours. Glances shows lots of activity and the temp directories have tons of tmp files- - sort_bucket, and plot.table1.tmp, etc. But I’m not seeing anything show up in hdd1 and no *.plot files anywhere else. I first thought that the permissions on hdd1 were incorrect, but they seem fine. The log file chia1_1_.log is empty.
The rig is has 32 GiB of Memory, 8 vCPUs, a 4T SSD and a 16T HDD
Any help on diagnosing the issue?
Welcome to the forums!
You have the “-e” flag, which disabled bitfield plotting. This is no longer the recommendation as bitfield plotting is faster and more efficient. Are you using a headless Ubuntu Server?
You also don’t need to specify “-2” if its the same disk you’re using.
Are you single plotting or plotting in parallel?
Thanks for getting back to me. I don’t know much about this, but I’m plotting 8 in parallel on a 32 gig headless server – and I believe that the -t is to put things into a folder on the ssd, ahd the -2 is, in case this doesn’t work, to put them in the main dir of the ssd. Is this incorrect?
I’ll remove the -e from my batch, thanks. Anyway, the script has been running for over 25 hours now and nothing has landed on the hdd. Any thoughts?
Ah, plotting is done in four phases. The “-2” puts phase 3 and 4 on the specified directory.
Did you kick off all 8 at the same time? If you did, writing those out to the HDD will be a big bottle neck.
You said nothing has appeared on the HDD so I think it might be stuck swapping to disk because it ran out of ram. Did the plotters stop printing output?
I think I get it. The 8 processes are run on an hour delay, each - meaning sleep 0h, sleep 1h, sleep 2h, etc. - so they shouldn’t be on the same cycle. There’s plenty of space left on ssd1 and hdd1 is empty. As of this morning (May 2, 915am) it looks like it is still writing:
277819392 May 2 09:10 plot-k32-2021-04-30-21-00-f4570fdfe8f0150e5912247f6ab8e0b95524faf0e2693bb4a6ff4c565d4c4878.plot.p2.t3.sort_bucket_001.tmp
277819392 May 2 09:10 plot-k32-2021-04-30-21-00-f4570fdfe8f0150e5912247f6ab8e0b95524faf0e2693bb4a6ff4c565d4c4878.plot.p2.t3.sort_bucket_000.tmp
Yeah, its stuck doing something. Especially with a date of 4-30. It looks like these are in phase 2 of the plot process. I say dump them, clean up the temp directories, then just try one plotter to see if it does it again.
Will do. Thanks again. How long should I wait for results on 1 plotter? 24 hours?
You said “vCPUs”, is this a virtual machine? Can you also put what your plotter settings are?
I’m trying everything out on an Amazon AWS t2.2xlarge. Not sure what you mean by plotter settings - this?
chia plots create -k 32 -b 4000 -r 4 -u 128 -n 16
I’m following the instructions laid out on ChiaDeCentral
Ah, The performance on AWS won’t compare to dedicated hardware at home. Try just one plotter and verify what the plot time is before moving into parallel plots. It should end hopefully before 12 hours. If not, I don’t see plotting this way as a viable solution. Those are good settings to start with. Make “-n” just 1 so that you can examine the data.
Thanks - I’m trying to run a single plotter and will check back with results. I appreciate all the help.