[CLOSED] BTC Guild - Pays TxFees+NMC, Stratum, VarDiff, Private Servers

[BitMinerN8]

I have just restarted the Stratum beta pool, enabling the new dynamic difficulty adjustment.  The new system should drastically reduce overhead for large mining farms.

The target rates for miners are between 1 share per 3-10 seconds.  If a miner goes beyond 1 share per 3 seconds (during a 5 minute window), that miner's difficulty is doubled.  If a miner averages less than 1 share per 10 seconds (in a 5 minute window), it will be halved.  A "miner" is defined as a connection to the pool, not an individual worker.  In the case of very quick miners, there are shorter time frames (as small as 10 seconds) used for difficulty adjustments, but these require bursts of speed that could not be explained as simply lucky.

The target share rates are still extremely fast, meaning variance should not be adversely affected by this new change.  The only issue you may run into with large farms is the per-worker speed (on the BTC Guild website) may vary heavily.  However, your total mining speed should stay very close to your actual speed.


EDIT/UPDATE:  Another quick restart.  The pool now makes sure to send difficulty=1 on a new connection, so a disconnect/reconnect (pool restart for example) scenario doesn't have the miner submitting high diff shares and only getting credit for diff=1.

Now since Dynamic Difficulty adjustment went live my workers keep getting disconnected/reconnect and only do about 10 or so tasks and then randomly jump to next pool in list.

I am seeing this as well. I kept wondering why a bunch of rigs kept dropping from the proxy and ending up on the main pool.

[eleuthria]

Are the disconnects still happening?  I had to do a couple restarts (a few bugs in the difficulty adjustments) during initial deployment, but everything should be stable now.

My FPGA farm is currently showing 0 failures in the last 1h 31m (and has been adjusted to diff=4).


If you're still getting disconnected, please look at your proxy and see what happens immediately before the disconnect so we can find out what error is happening.


A possible fix:  Try downloading the latest proxy (links are on the beta stats page).

[BitMinerN8]

Are the disconnects still happening?  I had to do a couple restarts (a few bugs in the difficulty adjustments) during initial deployment, but everything should be stable now.

My FPGA farm is currently showing 0 failures in the last 1h 31m (and has been adjusted to diff=4).


If you're still getting disconnected, please look at your proxy and see what happens immediately before the disconnect so we can find out what error is happening.

Sure thing. I'll keep an eye on it.

[Sitarow]

Are the disconnects still happening?  I had to do a couple restarts (a few bugs in the difficulty adjustments) during initial deployment, but everything should be stable now.

My FPGA farm is currently showing 0 failures in the last 1h 31m (and has been adjusted to diff=4).


If you're still getting disconnected, please look at your proxy and see what happens immediately before the disconnect so we can find out what error is happening.


A possible fix:  Try downloading the latest proxy (links are on the beta stats page).

I did that before I posted.

Used both links and same results.

I start getting rejects and then the worker moves onto the next pool on the list.

Still is happening. Using windows proxy.

[eleuthria]

I did that before I posted.

Used both links and same results.

I start getting rejects and then the worker moves onto the next pool on the list.

Still is happening. Using windows proxy.

Is there an error showing up on the proxy?  If not, is there an error showing up on your miner for why it's changing pools?  I can't debug this on my miners because everything is working as expected.  The error is happening somewhere between the proxy or your local miners.

[BitMinerN8]

Are the disconnects still happening?  I had to do a couple restarts (a few bugs in the difficulty adjustments) during initial deployment, but everything should be stable now.

My FPGA farm is currently showing 0 failures in the last 1h 31m (and has been adjusted to diff=4).


If you're still getting disconnected, please look at your proxy and see what happens immediately before the disconnect so we can find out what error is happening.

Sure thing. I'll keep an eye on it.

I have not had a chance to dig into logs yet (kids, homework, life, wife, etc), but I have found this one thing in common. Of my 10 rigs, some with cgminer or bfgminer (2.7.5) on windows x64 going through the windows based proxy are all working fine. It's the 4 rigs running BAMT 0.5 pointed at that same windows proxy that keep dropping off.

At another location, I have 4 rigs, 2 on Debian with BFLs only, 1 running BAMT 0.5, 1 running bfgminer windows x64, all pointed at a Debian based proxy, all working fine.

So tomorrow I will have to read up on getting the logs for the BAMT rigs pointed at the windows proxy, or setup another Debian proxy and see if that stops them from dropping/failing to a backup pool. Maybe it's something specific with the flavor of Phoenix/Phatk2? It's stock BAMT, whatever that comes with on the image. I'll update when I have more info.

[eleuthria]

Are the disconnects still happening?  I had to do a couple restarts (a few bugs in the difficulty adjustments) during initial deployment, but everything should be stable now.

My FPGA farm is currently showing 0 failures in the last 1h 31m (and has been adjusted to diff=4).


If you're still getting disconnected, please look at your proxy and see what happens immediately before the disconnect so we can find out what error is happening.

Sure thing. I'll keep an eye on it.

I have not had a chance to dig into logs yet (kids, homework, life, wife, etc), but I have found this one thing in common. Of my 10 rigs, some with cgminer or bfgminer (2.7.5) on windows x64 going through the windows based proxy are all working fine. It's the 4 rigs running BAMT 0.5 pointed at that same windows proxy that keep dropping off.

At another location, I have 4 rigs, 2 on Debian with BFLs only, 1 running BAMT 0.5, 1 running bfgminer windows x64, all pointed at a Debian based proxy, all working fine.

So tomorrow I will have to read up on getting the logs for the BAMT rigs pointed at the windows proxy, or setup another Debian proxy and see if that stops them from dropping/failing to a backup pool. Maybe it's something specific with the flavor of Phoenix/Phatk2? It's stock BAMT, whatever that comes with on the image. I'll update when I have more info.

Sitarow is having the same problems, using Phoenix/Phatk.  I have a feeling Phoenix might be having problems with dynamic difficulty.  Unfortunately, that miner is basically unsupported at this point (it's been months since any work has been done with them, and the thread seemed void of any developer).

I suggested he use BAMT with cgminer rather than phoenix.

[vphen]

1, when server readjust difficulty to higher,  on proxy, previous works will be rejected because "Share is below expected target",  if I modify the proxy let these shares pass through, will it be accepted on pool?


2, when server readjust difficulty to lower, previous works should be accepted on proxy and pool. but I'm not sure, on pool side, will these shares get lower PPS rate?

[Joshwaa]

Some of my slower miners (215MHash/s) are taking 15+ minutes to turn in a share.  This does not seem good. Also getting HW errors on cards that have never had HW errors before.

[slush]

Joshwaa - what is your total hashrate? Actually hashrate of one particular worker doesn't mean anything, if all your rigs are still able to produce one share per few seconds.

[Joshwaa]

Total hash of all workers is 12Ghash.  Most of my workers even the quick ones are taking much longer than 3 seconds.

[slush]

Hm, any error in proxy log?

I didn't said that "some your rigs", but "all your rigs (together)". I'm almost sure that submission rate for all your rigs together is one per few seconds, which is the purpose of difficulty recalculations. Except when Eleuthria had a bug in his implementation.

[Joshwaa]

Ok, so the diff is for the farm not the individual worker? On the proxy I get a share aprox ever 5 sec. So I guess this is correct.

Just for kicks. I started a windows Stratum and connected the one worker I was having problems with. Now it is working nicely with no errors. This any help?

[slush]

Ok, so the diff is for the farm not the individual worker? On the proxy I get a share aprox ever 5 sec. So I guess this is correct.

Exactly. Difficulty is per Stratum connection, not per worker.

Just for kicks. I started a windows Stratum and connected the one worker I was having problems with. Now it is working nicely with no errors. This any help?

Are you sure your first Stratum proxy is on latest version? I'm doing some tweaks many times per day (there are so many stupid bugs in various miners!), so updating stratum frequently is quite a good idea.

I'll release 0.8.2 in few moments, where higher difficulty won't be propagated to the miners.

[Sitarow]

Some of my slower miners (215MHash/s) are taking 15+ minutes to turn in a share.  This does not seem good. Also getting HW errors on cards that have never had HW errors before.

I am having the same HW Errors. They are a pain to keep ontop of.

Also are you using phoenix miner?

[Sitarow]

Ok, so the diff is for the farm not the individual worker? On the proxy I get a share aprox ever 5 sec. So I guess this is correct.

Exactly. Difficulty is per Stratum connection, not per worker.

Just for kicks. I started a windows Stratum and connected the one worker I was having problems with. Now it is working nicely with no errors. This any help?

Are you sure your first Stratum proxy is on latest version? I'm doing some tweaks many times per day (there are so many stupid bugs in various miners!), so updating stratum frequently is quite a good idea.

I'll release 0.8.2 in few moments, where higher difficulty won't be propagated to the miners.

Okay I will download and run that see if it helps.

[slush]

Sitarow - I just released 0.8.3 which suppress annoying message about share below difficulty. Update, if it will scary you, otherwise 0.8.2 is fine :-).

[Sitarow]

Sitarow - I just released 0.8.3 which suppress annoying message about share below difficulty. Update, if it will scary you, otherwise 0.8.2 is fine :-).

I seen that lol

[gnar1ta$]

Everything looks good for me, no HW errors, with 5 rigs on cgminer 2.5.0 and win 7 hosting proxy.

[Joshwaa]

With new 8.2 I am not having anymore problems with the slower miners or the BAMT rigs.