[1423GH] ABCPool PPS - Proxy Pool For High & Steady Mining Rewards

[Inaba]

If that be the case, then 0.5% seems like it would be high.  There should be a pretty low invalid rate for reasons != stale.

[Jezzz]

Poolowner: please test cgminer, it is getting stuck often where it submits a share, loses communication, gets communication, submits a share and loses communication again in a seemingly never-ending loop

I have experienced this also, during this process, the invalid submissions go sky high.

[spiccioli]

I've tested your pool with poclbm and DiabloMiner and both have problems staying connected, while phoenix, on the same rig, works ok.

What is different between your pool and most others?

Thanks for letting us know about this. We did heavy modifications of pushpool to obtain low stale rates. We've only tested these mods with Phoenix, that's why we recommend it on our site. It now appears that some other clients may experience problems.

This is DiabloMiner

Code:

[8/18/11 10:22:18 AM] ERROR: Cannot connect to pool.abcpool.co: Bitcoin disconnected during response: 200 ok

I looked through the Diablo code, and I'm fairly certain this is caused by ABCPool timing out the LP connection with a '200 OK' after 50 seconds to get rid of disconnected clients. Most pools don't do this, but it is one of the reasons ABCPool can keep the stalerate low. Phoenix handles this excellently by silently reconnecting; just what we want. Diablominer might expect a different response than 200 OK. I've posted the issue in their forum.

poclbm keeps disconnecting, I don't have a log, but I can give one if you need it.

That would be most helpful. Can you PM them?

Hi MintCondition,

here it is poclbm:

Code:

pool.abcpool.co:8332 19/08/2011 09:45:59, warning: job finished, miner is idle  
pool.abcpool.co:8332 19/08/2011 09:46:32, warning: job finished, miner is idle  
pool.abcpool.co:8332 19/08/2011 09:46:53, warning: job finished, miner is idle  
pool.abcpool.co:8332 19/08/2011 09:47:10, long poll: IO error                   
pool.abcpool.co:8332 19/08/2011 09:47:11, LP connected to pool.abcpool.co:8332  
pool.abcpool.co:8332 19/08/2011 09:47:27, warning: job finished, miner is idle  
pool.abcpool.co:8332 19/08/2011 09:47:40, warning: job finished, miner is idle  
pool.abcpool.co:8332 19/08/2011 09:47:54, warning: job finished, miner is idle  
pool.abcpool.co:8332 19/08/2011 09:48:06, long poll: IO error                   
pool.abcpool.co:8332 19/08/2011 09:48:14, LP connected to pool.abcpool.co:8332  
pool.abcpool.co:8332 19/08/2011 09:48:27, warning: job finished, miner is idle  
pool.abcpool.co:8332 19/08/2011 09:48:39, warning: job finished, miner is idle  
pool.abcpool.co:8332 19/08/2011 09:49:04, warning: job finished, miner is idle  
pool.abcpool.co:8332 19/08/2011 09:49:09, long poll: IO error                   
pool.abcpool.co:8332 19/08/2011 09:49:12, LP connected to pool.abcpool.co:8332  
pool.abcpool.co:8332 19/08/2011 09:49:27, warning: job finished, miner is idle  
pool.abcpool.co:8332 19/08/2011 09:49:49, warning: job finished, miner is idle  
pool.abcpool.co:8332 19/08/2011 09:50:08, long poll: IO error                   
pool.abcpool.co:8332 19/08/2011 09:50:11, warning: job finished, miner is idle  
pool.abcpool.co:8332 19/08/2011 09:50:12, LP connected to pool.abcpool.co:8332  
pool.abcpool.co:8332 19/08/2011 09:50:25, warning: job finished, miner is idle  
pool.abcpool.co:8332 19/08/2011 09:50:37, warning: job finished, miner is idle  
pool.abcpool.co:8332 19/08/2011 09:50:49, warning: job finished, miner is idle

best regards.

spiccioli

[Brian DeLoach]

Seems like the pool crashed about an hour ago but since has recovered.

[rTech]

Guiminers sad thing for last few hours.
Nvidia 260GTX (216) : Shares 44 accepted, 33 stale/invalid .. I know its nonsense to use nvidia card like this but its still small income for me as electric bil isnt issue

Edit:
Reason seems to be "Phoenix" it gives out 48.2 mhash and with OpenCL miner i got only 45-47 but it gives out less invalid/stales! I tested this with my older GT 220 too and same results, lot of invalid/stales with Phoenix only.

Edit2:
Wrong alert, its gives again hellish amount of stales/invalid even with opencl. Im moving my nvidia based miners to another pool and keep just ati 5770 with your as its onlyone  has low amount of stale/invalid.

[MintCondition]

UPDATE: We are currently investigating the high invalid counts for non-Phoenix clients.

Going through the logs, the high invalid rate is for the most part due to 'invalid-time' submissions. We were seeing some of these before, that's why a few days back we added the

Code:

'X-Roll-NTime: N'

HTTP header to all getworks. This header indicates explicitly that time-rolling is not allowed. It seems that some clients are taking the presence of this header as a sign to enable time-rolling instead of disabling it. That's why we've just turned the header off again. At first glance it seems to have eliminated the bulk of time-invalid submissions.

[spiccioli]

UPDATE: We are currently investigating the high invalid counts for non-Phoenix clients.

Going through the logs, the high invalid rate is for the most part due to 'invalid-time' submissions. We were seeing some of these before, that's why a few days back we added the

Code:

'X-Roll-NTime: N'

HTTP header to all getworks. This header indicates explicitly that time-rolling is not allowed. It seems that some clients are taking the presence of this header as a sign to enable time-rolling instead of disabling it. That's why we've just turned the header off again. At first glance it seems to have eliminated the bulk of time-invalid submissions.

MintCondition,

with DiabloMiner is a lot better now:

Code:

[8/19/11 10:42:54 PM] pool.abcpool.co accepted block 20 from Cypress (#1)
[8/19/11 10:42:55 PM] pool.abcpool.co accepted block 21 from Cypress (#1)
[8/19/11 10:43:01 PM] pool.abcpool.co accepted block 22 from Cypress (#2)
[8/19/11 10:43:03 PM] ERROR: Cannot connect to pool.abcpool.co: Bitcoin disconnected during response: 200 ok
[8/19/11 10:43:09 PM] pool.abcpool.co accepted block 23 from Cypress (#1)
[8/19/11 10:43:21 PM] pool.abcpool.co accepted block 24 from Cypress (#1)
[8/19/11 10:43:27 PM] pool.abcpool.co accepted block 25 from Cypress (#2)
[8/19/11 10:43:32 PM] pool.abcpool.co accepted block 26 from Cypress (#1)
[8/19/11 10:43:53 PM] ERROR: Cannot connect to pool.abcpool.co: Bitcoin disconnected during response: 200 ok
[8/19/11 10:44:08 PM] pool.abcpool.co accepted block 27 from Cypress (#2)    
[8/19/11 10:44:25 PM] pool.abcpool.co accepted block 28 from Cypress (#1)    
[8/19/11 10:44:43 PM] ERROR: Cannot connect to pool.abcpool.co: Bitcoin disconnected during response: 200 ok
[8/19/11 10:44:46 PM] pool.abcpool.co accepted block 29 from Cypress (#1)    
[8/19/11 10:44:46 PM] pool.abcpool.co accepted block 30 from Cypress (#1)    
[8/19/11 10:44:58 PM] pool.abcpool.co accepted block 31 from Cypress (#1)    
[8/19/11 10:45:00 PM] pool.abcpool.co accepted block 32 from Cypress (#2)    
[8/19/11 10:45:06 PM] pool.abcpool.co accepted block 33 from Cypress (#2)    
mhash: 658.4/644.0 | accept: 33 | reject: 0 | hw error: 0

There still is the 50 seconds LP socket close.

I think 50 seconds is too little given that you need to communicate a new block every 10 minutes on average.
Did you try a very short keep-alive, like a couple of minutes so that the socket gets closed faster if the client pc does not answer fast enough?

spiccioli

[spiccioli]

MintCondition,

with DiabloMiner is a lot better now:

Code:

[8/19/11 10:42:54 PM] pool.abcpool.co accepted block 20 from Cypress (#1)
[8/19/11 10:42:55 PM] pool.abcpool.co accepted block 21 from Cypress (#1)

There still is the 50 seconds LP socket close.

I think 50 seconds is too little given that you need to communicate a new block every 10 minutes on average.
Did you try a very short keep-alive, like a couple of minutes so that the socket gets closed faster if the client pc does not answer fast enough?

spiccioli

No, it simply took longer,

miner looping, no more share submitted, my stats have invalid shares increasing steadly.

Code:

[8/19/11 11:42:47 PM] ERROR: Cannot connect to pool.abcpool.co: Bitcoin disconnected during response: 200 ok
[8/19/11 11:43:37 PM] ERROR: Cannot connect to pool.abcpool.co: Bitcoin disconnected during response: 200 ok
[8/19/11 11:44:27 PM] ERROR: Cannot connect to pool.abcpool.co: Bitcoin disconnected during response: 200 ok
[8/19/11 11:45:17 PM] ERROR: Cannot connect to pool.abcpool.co: Bitcoin disconnected during response: 200 ok
[8/19/11 11:46:07 PM] ERROR: Cannot connect to pool.abcpool.co: Bitcoin disconnected during response: 200 ok
[8/19/11 11:46:57 PM] ERROR: Cannot connect to pool.abcpool.co: Bitcoin disconnected during response: 200 ok
[8/19/11 11:47:47 PM] ERROR: Cannot connect to pool.abcpool.co: Bitcoin disconnected during response: 200 ok
[8/19/11 11:48:37 PM] ERROR: Cannot connect to pool.abcpool.co: Bitcoin disconnected during response: 200 ok
[8/19/11 11:49:27 PM] ERROR: Cannot connect to pool.abcpool.co: Bitcoin disconnected during response: 200 ok
[8/19/11 11:50:17 PM] ERROR: Cannot connect to pool.abcpool.co: Bitcoin disconnected during response: 200 ok
mhash: 654.7/654.1 | accept: 170 | reject: 0 | hw error: 0

[kripz]

I created my account 1 hour ago and have this at 350mh/s

Current Hashrate: ~320 MH/s
469 shares
Lifetime Valid: 469 shares
1 shares
Lifetime Stale: 1 shares
25526 shares
Lifetime Invalid: 25526 shares
0.01301434 BTC
Lifetime Earned: 0.01301434 BTC
0.01301434 BTC
Balance: 0.01301434 BTC
06:54:02 UTC

Using cgminer, cgminer isnt even reporting back invalids. Is it even possible for me to generate that many shares in about 1 hour?

[user7516]

...
Using cgminer, cgminer isnt even reporting back invalids. Is it even possible for me to generate that many shares in about 1 hour?

The same problem with cgminer 1.5.6:

http://img843.imageshack.us/img843/3534/clipboardimage1f.png

[MintCondition]

...
Using cgminer, cgminer isnt even reporting back invalids. Is it even possible for me to generate that many shares in about 1 hour?

The same problem with cgminer 1.5.6:

I think cgminer is retrying all shares that it thinks it previously failed to submit. It does not report them as invalid yet, because it has not yet given up on them. When i look through the logs, I see you submitting boatloads of 'unknown-work' shares, each for a different block. These are all very, very old shares that cgminer is constantly re-submitting. The original submission attempt might have even been successful. I suspect cgminer wants to see a certain kind of error message before it gives up, and that ABCPool is sending a different one. Each resubmission counts as one invalid share, although in most cases it's the same shares over and over again.

In the next release of our backend the error messages have been changed so they look more like those of the original pushpool. We are still testing this version, and expect to have it ready in a few days.

[Brian DeLoach]

Seems like the pool crashed about an hour ago but since has recovered.

Can you comment on the downtimes? They seem to happen every 24-48 hours, but the system recovers in less than an hour. There seems to have been another crash in the last 30 minutes.

[bal3wolf]

I can say my stales and invailds arent real bad using guiminer and poclbm.

Lifetime Valid: 169024 shares
Lifetime Stale: 250 shares
Lifetime Invalid: 616 shares

[spiccioli]

Question:

given these personal stats

Code:

Personal Stats

Current Hashrate: ~1120 MH/s
Lifetime Valid: 118440 shares
Lifetime Stale: 232 shares
Lifetime Invalid: 5947 shares
Lifetime Earned: 2.48992491 BTC
Balance: 2.48992491 BTC
Updated: 23:22:03 UTC 

and a share reward of 0,00002769 BTC each, how comes it says Earned: 2,48 when 118440 share times 0,00002769 gives 3,27 BTCs?

I did not withdraw any BTC yet.

spiccioli.

[nomnomnom]

I think the share count may be off...
Mining here since not even 2 days and it shows almost 180.000 shares,
I have ~2GH/s and I doubt that I generate that many shares

Also I have a lot of connection problems today
Miner is idle, sometimes I lose the connection
and I produce some rejects which say:
[21/08/2011 02:45:56] Server gave new work; passing to WorkQueue
[21/08/2011 02:46:03] TypeError in RPC sendResult callback
[21/08/2011 02:46:03] Result 000000001a948ef9... rejected
[21/08/2011 02:46:05] TypeError in RPC sendResult callback
[21/08/2011 02:46:05] Result 0000000077394270... rejected

That was right now, while I am posting.
Sometimes it runs fine for hours then it starts again.

[chunglam]

I also think the share count may show more than I should have...
but I am satisfied with the reward from this pool compared with other pool

[PoulGrym]

2,15760376181301e-5 per share here.. 54495 shares, It should be ~ 0.0000276900796620

something is wrong.. with the amount I've been payed for the correct pay I should only have done 45448 shares..  but it's showing 57238 shares

[kripz]

You guys dont pay out the full 8 decimal places? I received 0.2btc with 0.2 set as the minimum.

[MintCondition]

NEWS:
Dear members,

As mentioned earlier, certain clients are submitting huge numbers of invalid shares while the incompatibility of certain mining clients remains. As this practically means these clients are DOSsing our servers, we are forced to start blocking their respective users in order to ensure service quality for those that are submitting normally.

1. Users that submit more invalid shares than valid (valid + stale) shares (ie. >50% of total is invalid) will be blocked on IP-basis.
2. You will only be blocked after submitting at least 1000 shares (ie. it is no problem if you start with submitting 42 invalid shares)
3. Blocked users will not be able to continue mining
4. Blocked users can still withdraw their funds.

If you are blocked, you may request an unblock by sending us a PM explaining what steps you have taken to mitigate the problem.

Some tips:
We noticed that especially the cgminer client sends a lot of invalid shares. If you are using this client, switching to another client will probably solve the invalid-shares issue.

If you have any questions, feel free to contact us.

[MintCondition]

You guys dont pay out the full 8 decimal places? I received 0.2btc with 0.2 set as the minimum.

With automatic payout we pay out in rounded figures, and the unrounded part remains as balance on your account. That way we keep your wallet a bit cleaner. This is how it works exactly:
If your balance exceeds the minimum amount set by you, we pay out in increments of that minimum amount. So if you would have had 0.51234BTC, we would have paid 0.4BTC and the remaining balance would be 0.11229BTC (because of the 0.00005 tx fee)

This means your remaining balance is still in 8 decimals; we just subtract the amount that has been payed out. By paying out using these rounded numbers it is easier to gauge your earnings from your transaction overview in the bitcoin client. If you want to withdraw your full balance you may initiate a manual cash out at any time. Manual cashouts are always for the full remaining balance.