We're aware of that. This is an experiment on improving node performance for multiple miners.
Sure I got this.
Another solution (or almost the same?) is also ckpool based, this may be not so elegant as yours but looks like it may work.
One guy asked me about how to configure ckpool in proxy mode for multiple users to deal with p2pool network.
As these users do not want to install and have Core wallet but they use online wallets which not accepts generated transactions. So in this test enviroment ckpool works like a gateway and uses the single connection from ckproxy to p2pool network and single worker (payout address). Is this the same what you doing?
In this case original Payout module and pplns_process() calculations of ckpool will be useless for this exact case (p2pool) and it need to be rewritten because ckpool will never know about solved p2pool Block and payouts will be never calculated based on received payout from p2pool network to specified address. But in case of single node owner with multiple miners (own devices) payout is not a problem. For different miners (persons) ckpool still can be adopted for completely another payment module and another Block detection mechanism.
PS: I do not ask you to do it for sure, just told it is possible if someone need this solution.
So.. This is local node w/ bitcoind, p2pool and ckpool on the same server, it is running well with no DOA shares for now and I am pretty sure only because this is LOCAL node but not remote.
I still do not have statistics for long period but for 12 hours now and just 2 workers but maybe it doesn't as soon as ckproxy running well . I have made conclusion for some time of using p2pool that it is necessary to have own local node with the shortest round trip delay, in this case DOA hashrate and DOA shares are minimal. Another issue is GBT latency but this is not related to this particular task.


