[20 BTC] Multithreaded Keep-alive Implementation in Bitcoind

[Inaba]

3151 XXX      20   0  389m 167m 7556 S    8  1.4  20:04.13 bitcoind

That's about 16 hours of running, so looks good to me.

[JoelKatz]

Thanks guys.

[phorensic]

Is everything stable, yet?  Time to freeze the code?  Release Candidate #1?  No new features planned?  I really want to run this code on my pool and it sounds like you have hit a milestone.

[Matt Corallo]

Is everything stable, yet?  Time to freeze the code?  Release Candidate #1?  No new features planned?  I really want to run this code on my pool and it sounds like you have hit a milestone.

And then pull request it so it can get a more thorough look-through and get merged upstream.

[Inaba]

I think there's still the getwork skeleton issue to work out, but other than that it's a big improvement.

I'm not sure if the changes to hub mode have made it into the diff yet, as well?

As far as a pull request, it's unlikely it will be mainstreamed due to some of the features/changes are not good for clients.  Though maybe I misunderstood that part?  I'd be happy to see it in the mainstream code.

[Matt Corallo]

As far as a pull request, it's unlikely it will be mainstreamed due to some of the features/changes are not good for clients.  Though maybe I misunderstood that part?  I'd be happy to see it in the mainstream code.

Well yes, hub mode probably won't get implemented as its really a hack to get it to work with the current network segmentation, but most, if not all, of the various code optimizations and RPC multithreading/asio/etc absolutely can and will eventually get merged upstream.

[JoelKatz]

The less ugly optimizations and straight bug fixes should certainly be merge candidates. The hub mode patches and uglier optimizations are not sensible merge candidates.

Obvious merge candidates would be:

1) strRPCUser/strRPCPass variables to avoid extracting them on each RPC call.

2) Native long polling support.

3) ToHex replacing HexStr.

4) Faster Base64 decoder.

5) Leak fix in pthreads version of CreateThread.

Also, some version of the RPC/keepalive fix needs to be merged but it doesn't have to be my version.

[error]

Please send pull requests for these!

[phorensic]

As far as a pull request, it's unlikely it will be mainstreamed due to some of the features/changes are not good for clients.  Though maybe I misunderstood that part?  I'd be happy to see it in the mainstream code.

I agree with you.  This is a pool oriented fork and Jay Garzik will probably take a couple seconds to look at the code and then close the pull request.

[JoelKatz]

I agree with you.  This is a pool oriented fork and Jay Garzik will probably take a couple seconds to look at the code and then close the pull request.

I think the five things I specifically mentioned would be accepted. The long polling support is really the only maybe.

[simplecoin]

Is there a unified patch ready that is stable, or better yet working repo? I've been running the last stable 3diff against .23 I saw weeks ago.

[Furyan]

Is there a unified patch ready that is stable, or better yet working repo? I've been running the last stable 3diff against .23 I saw weeks ago.

I have 4diff running stably against 0.3.24.

[Inaba]

As do I

[backburn]

3.24 4-diff is running stable for over 24 hrs. Solved several blocks @ 175 GHash

Code:

  PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND                                                    
22895 tacos    20   0  466m 260m 6800 S    4  3.3  52:03.41 bitcoind-24prev

[TeraPool]

Can somebody please spoon feed me the process as to how to patch the source code with JoelKatz' diff .txt file?

I know how to compile bitcoind, but am a noob when it comes to patching things :S

[JoelKatz]

Can somebody please spoon feed me the process as to how to patch the source code with JoelKatz' diff .txt file?

I know how to compile bitcoind, but am a noob when it comes to patching things :S

Go to the 'src' directory, and then:

1) If you don't have the diff already, use this command:
wget http://davids.webmaster.com/~davids/bitcoin-4diff.txt

2) To patch it:
patch -p1 < bitcoin-4diff.txt

If you did it correctly, the output should be this:
patching file init.cpp
patching file main.cpp
patching file net.cpp
patching file net.h
patching file rpc.cpp
patching file util.cpp
patching file util.h

Update: Sorry about the typo in the URL.

[phorensic]

Testing bitcoind v0.3.24 + Joel Katz v0.98 4diff on our pool.  Going very well so far.

[TeraPool]

Can somebody please spoon feed me the process as to how to patch the source code with JoelKatz' diff .txt file?

I know how to compile bitcoind, but am a noob when it comes to patching things :S

Go to the 'src' directory, and then:

1) If you don't have the diff already, use this command:
wget http://davids.webmaster.com/~davids/bitcoin-4diff.txt

2) To patch it:
patch -p1 < bitcoin-4diff.txt

If you did it correctly, the output should be this:
patching file init.cpp
patching file main.cpp
patching file net.cpp
patching file net.h
patching file rpc.cpp
patching file util.cpp
patching file util.h

Update: Sorry about the typo in the URL.

THANK YOU!

P.S. My hand has been forced to make a donation!

Just sent 1 to your payb.tc address. Thanks again.

[AnnihilaT]

Testing bitcoind v0.3.24 + Joel Katz v0.99 4diff on our pool now.... so far so good in the last 15 mins...

Does anyone know the difd from 0.98 and 0.99 ?  couldnt find a copy of the older version anywhere.

[Inaba]

Why would you want the older version?

Anyway ... I think there might be a small memory leak somewhere still.  My bitcoind is up to almost 700m at this point and I'm getting some connection errors occasionally, though that may be due to some other issues not related to bitcoind.