Yeah i saw that, maybe too many connections or quarkcoind used too much cpu? Or the too many files open error, I have that on my yacpool sometimes if I forget ulimit -n 1000000.
Have you seen what p2pool reports during those times, if anything? My yacpool sometimes is not reachable for a short time (~5 min) from browsers, but p2pool keeps running fine and miners submit work if i go check, I think that happens when yacoind uses a lot of CPU.
I will did into the log and hope thats not already rotated out.
Also there are still some corrupt miners which submit shares above target, it looks like it's reporting all it calculates:
QP967T5PavPKNdFm4ctqxLdbLMhECU5uoZ
If i were better in Python i would implement a temporary IP ban for such clients, so the clients get some notice.
Or is there a way to pass some messages to the miner with stratum and/or http RPC?