DOUBLE_SIGNAGE_POINT errors

[Jacek-ghub]

Is there a way on your side to recognize whether a partial was duble-signage due to plot dups or network advertising it more than once? It would potentially help farmers to squash dups, and also made those coming from multiple advertisement as not triggering errors.

I checked my farm for dups, and couldn't find any. Also, when going over logs, I saw (mostly) double entries coming from the network.

[william-gr]

We have been thinking about this trying to come up with a good solution. But unfortunately nothing in the works yet.

[Jacek-ghub]

Let's ignore the case of farm based dups.

Do you know whether network originated dups are due to some peers sending the same data, and the local node not ignoring the duplicate signage point. Or rather there was some change in the signage point, and the whole network needs to be updated?

Maybe this should be addressed by Chia (as a bug on their side? In case that is peer related, the node should just ignore the second request, where if that was a change in the signage point, maybe the protocol should be extended, and have a mark on that processed proof as an updated one (e.g., something like SIGNAGE_POINT_XYZ:r2)?

Also, in case this is the second case (revised signage point), as all nodes will be processing it, maybe the pool can also keep track of those, and that would be the way to filter those out?

Also, I assume that most farmers were initially checking for dups (seeing those errors). So, maybe an option to exclude those errors from stats could be added? For instance, I would like to see when anything else, but those dups, will show up, so I know that I have some work on my part. At this point, it is more that i assume that those are dups, so basically start ignoring those reports..

[william-gr]

I always thought of it as a bug in the chia farmer, since it makes no sense for the farmer to send a proof of the same signage point. I honestly don't know what causes it in the node, I never could get someone in Chia Network to explain to us how/why that happens.

I have little faith in it being fixed on their side, hence a fix/workaround in the pool side makes sense.

[Jacek-ghub]

I have opened a bug report on GitHub for those dups. Maybe you can chime in there, so it will be recognized not just as a one person issue.

Chia-Network/chia-blockchain#11757