Replies: 2 comments 2 replies
-
|
Good idea, but we should wait at least 30 (maybe 45) seconds to be conservative |
Beta Was this translation helpful? Give feedback.
-
|
How can a PC harvester harvest more than 100 hard disks? The current harvester does not work properly if the hard disk exceeds 50 pieces. |
Beta Was this translation helpful? Give feedback.
-
|
How exactly does it not work? It becomes slower? |
Beta Was this translation helpful? Give feedback.
-
|
I must agree I have very slow storage, it works fine when all plots are loaded, but if I update chia and restart harvester I need 24+ for it to load all plots!! |
Beta Was this translation helpful? Give feedback.
-
I recently had a failing disk cause very slow lookup times (#11280). That one bad disk rendered tens of thousands of plots unavailable because the farmer/harvester waits patiently for everything to return. I also saw the same scenario occur with unavailable network shares in the early days.
There is no scenario in which waiting for an already-late proof makes sense.
My suggestion is that harvesters should timeout after ~25 seconds and return any proofs/partials found up to that point.
Example of one plot taking 437 seconds, leading to the entire harvester taking 437 seconds. Any proof found would be invalid and multiple challenges have been skipped entirely:
It would be better to see:
"WARNING: Looking up qualities on ...plot timed out after 25 seconds."Beta Was this translation helpful? Give feedback.
All reactions