Chia update (requires new config), hardware change etc. Chia update is probably the main offender for having to stop plotting, updates come fairly regularly and expect that to continue.
Can’t remember which thread it was in but pretty sure someone has already written a script for stopping plot creation within a job / que so you don’t have wasted writes so should be possible I think.
Either I need to read the page again or isn’t a feature, does swar switch final destination if the target drive is full?
As of now, no. But it’s in the roadmap up top. I just haven’t been able to dedicate time to it since release since this thing blew up more than expected and I’ve been trying to answer support requests… which I will admit that I have stopped answering a lot of them since I’m getting a bit burnt out.
I’ve been using Plot Manager for about a week now and it’s really helpful. Here are some things that I think would make it even better:
A job time threshold – this would allow one to set a max job time threshold to nuke any existing chia.exe processes that might be hung. Current version 1.1.5 of the chia plotter seems to randomly suicide without actually throwing any errors in the plot logs, and it would be nice to be able to say that “after 8 hours kill the process and delete the TMP files”. Of course, the Chia devs could fix the plotter issue that’s causing this, too
Better descriptions in the config file of how the job variables work, with perhaps some examples given the logic dependencies, especially the “start_early” variables.
Overall, I think the next logical step for Plot Manager would be analysis of the plotting system’s specs, such as the Temp storage used, CPU, etc, to dynamically manage and optimize plotting. Or at least allow the user to enter in the type of hardware they have if it’s too hard to poll the system resources.
For example, telling Plot Manager in the config that “I have 24 total threads @ 4.1Ghz, 2x PCI-E 4.0 SSDs in RAID 0 with XXX IOPS R/W, 32GB RAM” etc, and then it calculates the most efficient way of doing things. Even at a basic level I think this would help a lot of people eliminate the guesswork.
There is one more thing that would be really useful:
A global max_phase _1 setting.
Currently I’m running into the situation where plots are waiting to get started because of the limit I set for phase 1 since phase 1 is more that 1/3 of plot time.
Running 3x3 plots on 3 disks with max 1plot for phase 1 on each disk.
If I could set the global max phase 1 plots to 4, then I can set each individual que to max 2.
That should reduce the waiting times will still ensuring I only get max 4 plots in phase 1, instead of 6 which would overload the cpu.
I have a very hard time wanting to make something that could potentially delete someones plots / active plots because that could lead us down a very dangerous path. My biggest hangup is the potential for it to be dangerous by a bug or user error.
I think examples would be a great idea.
This is an idea I have been doing with, but not something that’s at the forefront of my TODOs. I think it’s great in theory but there are so many other variables in play like how often do you use your PC, are these drives touched by other things, and even available threads is subjective because some CPUs just might not be able to handle some of the workload. Nonetheless, it’s a good idea to eventually implement
Just deployed to 2 plotters - 1 is view only on the existing stuff automated by my own crude scripts, the second is taking over one of my existing jobs as well
Similar in essence to mine but yours shares phased progress and looks much better
Would it be worth having a chat with @Tydeno and maybe combining both apps? You’ve both got very cool features that would compliment each others build very well and might make the dev load a bit easier to deal with. Just a suggestion, not sure how well you coder guys get on with each other!
Good idea! But its Python based so I’m out (like no experience there)
From what I can tell, both tools do roughly the same. But I can tell , Swars implementation is more advanced and got better test-coverage than my PowerShell based solution called “Ploto”.
My Module is intended for people that don’t want spend time hustling around with Python and WSL.
Does this plot manager accurately shows plots that are started with the -n 2 command?
The plotman version I have only shows phases of the first plot only and for all subsequent plots being generated by the same plotter command just shows 4:0 in the phase status.
(venv) ubuntu@ip-172-31-31-55:~/chia_scripts$ plotman status
plot id k tmp dst wall phase tmp pid stat mem user sys io
07887a7b 32 /chia/plot-temp/plot-tmp1 /chia/farm-vol2/Farm-2 22:25 4:0 0 86594 RUN 6.5G 13:54 0:45 8:00
059d2be0 32 /chia/plot-temp/plot-tmp1 /chia/farm-vol2/Farm-2 22:40 4:0 0 85997 DSK 6.5G 13:57 0:46 7:58
d522bba5 32 /chia/plot-temp/plot-tmp1 /chia/farm-vol2/Farm-2 22:55 4:0 0 85441 RUN 6.5G 14:03 0:46 7:57
@swar If i have, for example, 4 destination drives and all 4 are in the same job. What happens when 1 of these drives becomes full? Does it skip to the next drive or will the plotting hang?
