The latest version of Farm still occasionally experiences interruptions in farming. I have to restart my Farm once a day because after some time, when there’s no redrawing, the CPU usage drops to 0. Without restarting, Farm remains mostly non-rewarding; I can’t get rewards. I don’t know the reason because after Farm runs for two days, there are only occasional redraws, and there are no more reward bursts.
Are you on a stable release jun-18? There were a few more optimizations since, for example included in this test build (don’t use node from there, just farmer): Snapshot build · subspace/subspace@f6ed626 · GitHub
With those changes messages between node and farmer should be greatly de-duplicated and help with performance.
Would be great if you can share logs as well.
Yes, I’m using this version: Release gemini-3h-2024-jun-18 · subspace/subspace · GitHub. As for the logs, I don’t think there’s any useful information. On the first day, approximately every hour, a log like this would appear:
2024-06-29T16:36:55.032730Z INFO {farm_index=1}: subspace_farmer::reward_signing: Successfully signed reward hash 0x2c32ab2253beb7e7302ec3da4b80a4c90f01fdf848ef0b15038d4617f4e67435
After the second day, sometimes even earlier, these logs no longer appear, indicating that farming is in a stopped state. I must restart the farm for it to resume.
I believe that issue was since fixed. There is actually an issue happening silently on the node side in case you have many farmers and farms connected to the same node (which wasn’t what is was designed to do, but I understand why people do it). Latest rest build addresses that issue to a large extent and there were positive confirmations already that it did help with this particular issue. Give it a try and let me know if it helps.
Is this memory usage（Commit Size） by node looks normal ?
6.18 release, running 12 days, connected by close to 400TIB plots from 8 farmers.
Ps: Definitely not this many at the beginning.
Please don’t post unrelated questions in existing threads about different topics, better create a separate new topic instead.
I believe the fix for this was finally found and shipped in Release gemini-3h-2024-oct-03 · autonomys/subspace · GitHub