Brief background
In the fall of 2023 I decided to switch to compressed plots and began the process. Everything was finished at the end of November. Since then I have not received a single reward.
By December, when it became obvious that something was wrong, I began to worry and began to check that everything was in order.
“Don’t worry, you’re just in the twilight zone” was the most common answer.
And everything looked exactly like this. All indicators in the GUI were green. The log did not show any critical errors.
And yet, after almost half a year, when I got desperate and switched to mining in a pool, I found a bug.
The fact is that for unknown reasons, when using harvesting on a video card, for reasons unknown to me, the partials generation process stops, the video card goes into hold mode, but this is NOT displayed in the GUI.
Externally, the system looks completely working. The plots are checking, “no proof found” etc.
I am not an expert in chia blockchain, but I dare to assume that the mainnet is a higher-order global pool, i.e. what is true for a small pool is likely to be true for the mainnet.
Those. When this bug disable your video card, partial accepting stops and your array of plots “falls out” from the reward calculation system.
How does this shows itself?
ONLY !!! with a line with “GRResult is not GRResult_OK, received GRResult_OutOfMemory” in the log and no activity in the video card system monitor.
Win 10, 6300 plots, GTX 1060 6Gb.
p.s. that bug [Bug] GRResult error occuring a couple times a day farming a few PB of C2 compressed plots using an Nvidia P4 GPU - Bladebit · Issue #15404 · Chia-Network/chia-blockchain · GitHub
You left out that that line is an ERROR level line. You do need to scan your logs for ERROR / WARNING / Exception / … This part is on you.
Virtually everyone on the forum agrees that if in trouble, join a pool.
Also, BB plotter is the worst compressed format plotter. The bug you pointed to on github was created almost a year ago, yet Harold / chia doesn’t give a shit about fixing it.
I mean, we have seen people pushing 10x bad luck (I also was close to that). Running solo, with a bug weird like that, it is really hard to nail it down if not being connected to the pool to get some extra datapoints.
I think that back in 2021, I wrote that GUI should have something like “recent [bad] events” field, where it would show the latest issues (log lines with ERROR / WARNING / …). It is a no-brainer to do, but again there is no will to do it on CNI side.
From that error alone I can tell you are using Bladebit and an old video card, it’s a very common known error, there are scripts available to monitor the log and restart the harvester.
My advice, if you can replot change to GH, or you can even use GH to farm your existing plots without the error but you will have a fee to pay but only gain reliability.
This is normal, skill issue.
Always pool
Sorry it was a hard lesson, but now you have learned to… Just say no to anything CNI develops in the last several years. Farmers are not their priority in the least.
I used to say no to pools as well, but now realize they provide the tools the GUI lacks to ascertain farm health. Absolutely worth it for sure.
Gigahorse All the Way and then some…
Thanks a lot for reply, I’ve just cant understand, why do not make a huge button in GUI with 3 color. Green - your system is Ok, yellow - it’s Ok, but!!! & red - nothing works. That’s it!
its a greed problem issue.
you solo’ed because you didnt want to pay 0.5%-1% for pooling.
You ended up getting -100%
Pooling does not reduce rewards, it just smooths them closer to what they should expect to be.
If you have a very large farm you might not need the smoothing effect but you lose all the things the pool helps you with (making sure proof searches are operational, etc) if you want to escape the fee.
Sorry, but you are extremely quick in labeling.
You have no idea why I was mining solo.
However, the problem with the lack of payment of rewards persisted even after joining the pool; simply being in the pool allowed us to identify the bug.
So this is not greed, but rather poor quality software.