The Stale rate you see is also including nodes that are running not the right p2pool software. These nodes will still connect to the network but not provide proper shares. Yet the Stale rate is still accounted to the total rate of p2pool.
Someone using obsolete/misconfigured p2pool nodes?
I've been observing this the past week, that all of a sudden a hashrate of 12mh appears with Stale/Dead on Arrival shares. Yet scanning over all the public pools revealed, that all nodes were running fine and local stale rate was below 1%.
That is not what I see when looking at the stats:
q30.qhor.net: 3.53MH/s (Stale: 14%) from 13 users
cach.cryptoprojects.eu: 3.94 MH/s (DOA NaN H/s / NaN%) - local DOA stats does not work ...
p2cache.syware.de:11.2MH/s (Stale: 11%) from 15 users
185.8.164.16: 0.00H/s (NaN% DOA) - this one is empty
p2pool.cachecoin.org: 1.12MH/s (0.16% DOA) - this one has smallest DOA rate from all the non-empty pools.
p2pool-miner.info: 1.40MH/s (Stale: 17%) from 1 users
46.4.96.166: 6.99kH/s (0.0% DOA) - almost empty
seems some small pools have DOA under 1 %, but the largest pool has 11% stales and from the 7 public pools, 3 of them have stales over 10% (and one unknown)
That is also not correct - the amount of coins found depends on the amount of blocks found by the pool. In the past week users on your pools were observing a higher return since the pool rate was most of the networks rate and thus was finding more blocks than the average block time of 15minutes... since the hashrate of 1/3 to 2/3 favoring your pool during this low difficulty high hashrate period were in favor for your pool. (But leading to possible security threads on the coin)
The payout right now on p2pool and a traditional pool will be averageing over 24hours.
Whoever finds the block has advantage for the next block, as he can begin computations on the next block immediately, but rest of the network need to wait for that block to be received by the network. This gives few seconds advantage to the block submitter and may be quite noticable on coins with less than a minute blocks, but with the block time increasing, the significance of this advantage decreases and for block times around 15 minutes it is quite insignificant (less than 1% advantage for pool nearing 100% of the hashrate).
As for PPLNS vs. PROP:
Vardiff on both my pools is set up so that one share is submitted every 45 (first pool) or 25 (second pool) seconds. This means that shares are coming (approximately) at the same rate from everybody, though share from someone with 10Mh/s has bigger value and in average block time a miner should submit (in average) 20 or 36 shares (1st/2nd pool) - therefore risk of miner not finding a share in the duration of a block (so he would be left from payout on it) is very low with current configuration.
I can easily switch one or both pools to PPLNS too (this would be a simple configuration change), if I'd believe it will help something, but I think PROP is better, so I'm keeping it as it is now
