@RoboCoder - I have only spent a short time testing PSJ. Mainly to make sure it worked correctly so that I could study the codebase. My immediate goal is to gain a complete understanding of the inner working of Bitcoin (protocols, community, problems, etc). As I'm much stronger in Java than C, PSJ seemed to be the best starting point. BitcoinJ (
https://code.google.com/p/bitcoinj/) has also a great codebase to learn from.
As the original author is unreachable, I've opted to start from scratch (learning along the way) in building a fresh JAVA pool server. I've also been studying Eloipool's codebase (
https://gitorious.org/bitcoin/eloipool), which is Python3, and it seems to be pretty solid. I've got a very VERY basic implementation up and running in Java, but its not quite ready for public testing yet.
PSJ gives decent stats on getworks and inound work - but i wish it would give more stats on the shares being tested back to the daemon. I would actually lilke to see more stats in general by miner made available directly from the management interface. Would certainly be willing to help test any work you do - and would definitely be willing to contribute funds toward development of stratum support and maybe a management client with stats reporting and more detail/monitoring by worker.
I totally agree, and plan on integrating a Reporting Engine with API support, run by a lightweight web server (lighttpd or similar).
Can I ask you WHY you choose PSJ? Is it stricly because of Java or is there another reason? I'll be sure to let you know when I have something that is worth-while testing.
Howdy!
I choose PSJ primarily because i run windows servers and am not very experienced with the vagaries of Linux - and i can run PSJ (java) on windows pretty easily. I am come from a different development background for code/web - i am a ms developer so vb and c are my prefered languages and for web scripting (are you ready for this) i am old school cold fusion - been developing in that since version 2 - actually met Jeremy Allaire years ago when he ran the company and i was on a development contract for y2k.gov (done in CF) and Allaire wanted to do a case study on it! I am WAY out of the normal developer profile for the bitcoin world!
Also C <-> Java is pretty easy cross training. Also do not have a lot of time to develop large apps so wanted something where the work was already done. And mining directly against the bitcoind is somewhat unsatisfying as i use bfgminer and the only thing you see is a change in the best share difficultly solved.
I actually had it running with the bitcoin-qt .63 and my miners got shares via longpoll but the hash rate reported by BFG on submitted shares is about 5% lower than when i mine against slush's pool with stratum.
It seems to be working on feeding work to the miners but not really confident that a solution will get back to the daemon and get me the block.
I have not looked at either the bitcoin daemon code or the PSJ code (yet) but am beginning to think the best way to figure out what might be going on is to reverse engineer the code. But it is soooo time consuming. Also because my goal was an internal mining "controller" for my own rigs, i think that i should approach it by just developing a scaled down app for my uses. Decisions, decisions..
RoboCoder