CKPOOL - Open source pool/proxy/passthrough/redirector/library in c for Linux

[norgan]

ok so firstly, my linux skills are not as good as my windows skills, i've compiled ckpool and bitcoind. i am trying to set up the web interface, i point apache to the http directory but get forbidden. apache works with default site so config is good. i have chmod 755 on both http and pool directories.

What am i missing?

[kano]

There are 2 git folders html and pool
html is the web folder.
If you read index.php (which is only 1 line of code) it should be abundantly clear where pool goes

The web/db side of the pool is not designed for the non-technical, on purpose.

[norgan]

There are 2 git folders html and pool
html is the web folder.
If you read index.php (which is only 1 line of code) it should be abundantly clear where pool goes

The web/db side of the pool is not designed for the non-technical, on purpose.

ok so i see that, looks to only work with db version. I wondered if it could be a way of seeing proxy stats. i'm guessing standalone proxy (-A -p) you wont get any stats other than the logs. i might have a go at getting the db set up.

Edit: ohhh of course i need to move pool to the web directory lol

[kingscrown]

u rock man, loving your work

[kano]

There are 2 git folders html and pool
html is the web folder.
If you read index.php (which is only 1 line of code) it should be abundantly clear where pool goes

The web/db side of the pool is not designed for the non-technical, on purpose.

ok so i see that, looks to only work with db version. I wondered if it could be a way of seeing proxy stats. i'm guessing standalone proxy (-A -p) you wont get any stats other than the logs. i might have a go at getting the db set up.

Edit: ohhh of course i need to move pool to the web directory lol

No, the web based interface is connected to ckdb.
i.e. you need to run ckpool is full mode with ckdb to be able to get full web stats out of it.

You could use the worker.php and address.php I wrote that uses Con's solo pool stats interface.
However, very much by design, ckpool itself isn't a store of mining information, just as ckdb itself isn't a work generator for a mining pool.

The separation makes for a much more powerful pool.
CKPool's job is to generate work, hand it out to miners and report the work and hash rate that the miners do.
This removes all the overhead of data storage of a pool and means ckpool concentrates on being the best that it is at handling work.

CKDB deals with all the data storage and web interface needed and thus you need ckdb also if you want all that.

[norgan]

Thanks mate, will have a play through the week.

[bicer]

Running ckpool as proxy. After putting into production commit as of Nov 8 my local p2pool started reporting higher dead share rate, from around 0.5G-5G to 30-40G. After removing proxy stats returned to the expected numbers. Screen shot illustrates this. Since I'm code illiterate not sure if this is due to code change and now proxy provides true stats or p2pool node and pron't work well together.
http://s22.postimg.org/z6so4m4r5/dead_share.png

edit:corrected date, rate.

[-ck]

Running ckpool as proxy. After putting into production commit as of Nov 8 my local p2pool started reporting higher dead share rate, from around 0.5G-5G to 30-40G. After removing proxy stats returned to the expected numbers. Screen shot illustrates this. Since I'm code illiterate not sure if this is due to code change and now proxy provides true stats or p2pool node and pron't work well together.


edit:corrected date, rate.

No I can't think of a mechanism for this, but you are adding another link into the chain. The purpose of adding the proxy would only be to add more miners to the one upstream p2pool connection. If you're just adding a proxy and have only one rig connected to it there really is zero advantage to adding the extra link in the chain. I would suggest setting the connection to the proxy to have a fixed diff though with a +diff extension to your username. I suggest your hashrate in GH / 1.7

[Dimoza]

I'm trying to run solo via the http://solo.ckpool.org/ but however just a tiny fraction of shares are accepted. Is it possible to increase prevetively the minimum difficultiy form my (miner) side? I see the +diff option was named above. However I can't fid any relevant information regarding this option. How can  I use it with my USER(i.e. BTC address) parameter?

I'm using bfgminer, if it have a sense to mention.

Thank you!

[-ck]

I'm trying to run solo via the http://solo.ckpool.org/ but however just a tiny fraction of shares are accepted. Is it possible to increase prevetively the minimum difficultiy form my (miner) side? I see the +diff option was named above. However I can't fid any relevant information regarding this option. How can  I use it with my USER(i.e. BTC address) parameter?

You're posting in the support thread for the ckpool software. The solo ckpool thread is here:
https://bitcointalk.org/index.php?topic=763510.620

However the quick answer is with solo mining it doesn't matter how many shares you get accepted since you don't get paid for shares at all and you only get a reward if you solve a block. Don't worry about what diff it's using since it's just an indirect marker of how hard you're mining.

[-ck]

There has been some instability in the master git for a little while now and it's only just been resolved, apologies for the inconvenience. It took a long time to discover that the utlist macros FOREACH_SAFE are only safe for linked list deletion but not freeing of ram within the list. I've contacted the utlist author to inform him but have added a workaround in the code in the meantime and the crashes have since resolved. In addition, although no crashes were seen in the code prior to the changes, some theoretical crashes have been fixed as well, though the solo pool was running for 42 days solid before I restarted it with the new code.

Changes to the git tree:
- Added best share support/reporting per user/worker
- Pruning of old disconnected instances and dead clients' ram.
- Repaired and upgraded the stratum reconnect support to work with any valid extranonce1 sizes.

I'll be tagging this as a milestone in the near future once I confirm the stability over a longer period.

Kano has made massive changes to the ckdb code to support markers and for pplns to be more shift-like, dramatically reducing the ram usage of last instances. Once Kano commits yet more changes I will likely tag yet another milestone in short order.

As a suggestion to ckpool software users, unless you are actively watching development in the #ckpool IRC channel, you should stick to tagged milestones or you will be unaware if there is any instability in the master tree.

[ZACHM]

I want to be able to change what pool all of my miners are pointed to easily. I assume that if I setup ckpool as a proxy and point all of miners to it, then I can use the proxy to point everything (appearing as one miner) to whatever pool I set (and failovers) in the proxy settings for ckpool.

Assuming this is all correct I attempted to build ckpool on my Ubuntu 12.04 machine (desktop installation, but essentially used as a server).

I used:

Code:

git clone https://bitbucket.org/ckolivas/ckpool.git

Now I have a directory ckpool, when I am in the directory and try:

Code:

sudo apt-get install build-essential ./configure --without-ckdb make

I get this error:

Code:

E: Command line option --without-ckdb is not understood

Am I doing something wrong?
I saw the part that says "Building from git also requires autoconf and automake" so I also tried:

Code:

sudo apt-get install build-essential autoconf automake ./autogen.sh ./configure --without-ckdb make

with the same result.

[-ck]

Read the readme directly instead of from the bitbucket page which doesn't handle the line breaks properly. Those are separate commands.

[ZACHM]

Thanks. I feel stupid now, I should have known that. 

[-ck]

After yet more extensive debugging and fine tuning potential issues that were leading to hangs/lockups or crashes I finally found an issue that client instances could be in the process of being dropped and received messages from them would be processed ever so slightly after the drop, triggering... well, stuff. Anyway it's fixed and finally exhibiting stability in our real world pool usage and I've tagged a new milestone M10 to indicate its suitability for use.

[FaSan]

I still try the CKProxy. I want to congratulate for this really great software.

But I found some problems, proxy going to crash many times :

Code:

[2015-01-29 10:57:31] ckproxy connector starting
[2015-01-29 10:57:31] ckproxy generator starting
[2015-01-29 10:57:31] ckproxy generator ready
[2015-01-29 10:57:31] ckproxy stratifier starting
[2015-01-29 10:57:31] Nonce2 length 3 means proxied clients can't be >5TH each
[2015-01-29 10:57:32] ckproxy connector ready
[2015-01-29 10:57:32] ckproxy stratifier ready
[2015-01-29 11:12:22] stratifier process dead, terminating!
[2015-01-29 11:12:23] Attempting to send message dropclient=46 to non existent process stratifier with errno 3: No such process
[2015-01-29 11:12:23] Failure in send_proc from connector.c stratifier_drop_client:241 with errno 3: No such process
[2015-01-29 11:12:23] Child process received signal 15, forwarding signal to ckproxy main process
[2015-01-29 11:12:23] Parent process ckproxy received signal 15, shutting down

and

Code:

root@devel:/opt/ckpool# src/ckpool -c /opt/ckpool/ckproxy.conf -A -p
[2015-01-29 12:05:39] ckproxy stratifier starting
[2015-01-29 12:05:39] ckproxy generator starting
[2015-01-29 12:05:39] ckproxy generator ready
[2015-01-29 12:05:39] ckproxy connector starting
[2015-01-29 12:05:39] Nonce2 length 3 means proxied clients can't be >5TH each
[2015-01-29 12:05:39] ckproxy connector ready
[2015-01-29 12:05:39] ckproxy stratifier ready
*** glibc detected *** src/ckpool: corrupted double-linked list: 0x00007f5dec004f80 ***

It's compiled without DB support in Ubuntu 12.04



Thank you in advance




FaSan

[-ck]

I know it wasn't long ago I tagged M10 but I finally found a bug which would present the way the crashes have shown up to date and have fixed it and tagged M11.

[-ck]

Last bit of instability seems to have been quashed with yet another race fix. M12 was tagged shortly after that fix to signify the latest stable point and now significant attention is being given to the proxy side of the code which has not been touched in quite a while and showed a lot of immaturity in the code.

[ArchitektoR]

First of all - thanks for all your work!

Now trying to setup a test pool on testnet - and get:

Code:

[2015-02-24 00:07:45] HTTP response not ok: HTTP/1.0 401 Authorization Required
[2015-02-24 00:07:45] Reopening socket to localhost:18332
[2015-02-24 00:07:45] Failed to get valid json response to getblocktemplate
[2015-02-24 00:07:45] CRITICAL: No bitcoinds active!

Do I need extra command or config lines?

[-ck]

First of all - thanks for all your work!

Now trying to setup a test pool on testnet - and get:

Code:

[2015-02-24 00:07:45] HTTP response not ok: HTTP/1.0 401 Authorization Required
[2015-02-24 00:07:45] Reopening socket to localhost:18332
[2015-02-24 00:07:45] Failed to get valid json response to getblocktemplate
[2015-02-24 00:07:45] CRITICAL: No bitcoinds active!

Do I need extra command or config lines?

You need to set up the rpc username and password on your bitcoind and set the same auth and password in your ckpool.conf file.