Topic: KanoPool kano.is lowest 0.9% fee 🐈 since 2014 - Worldwide - 2432 blocks - page 1599. (Read 5352367 times)

No you can't game the pool

Conspiracy theories.  I still don't buy it.  No one benefits from withholding blocks.  What if it's a coincident that someone rented hash and that happened when their rental ended due to them canceling or ran out of fund.  That's the only way I would see it makes sense.

these crooks need to be punished!

is it true to assume, that the 46BTC that got away with were rightfully from other members?

thanks kano for being transparent in the matter.

Meanwhile, way to go frozencat!!! -- are you related to sabercatt, the block before you?

Thanks for the update kano.  I'm glad to hear they are no longer mining here.

Fantastic explanation kano.

Thank you sir for the update.

Thanks for the info kano. Would have been nice to know if this was the same person withholding as the other pool had. But of course no info was shared from them.

Good to know you are on the look out. Same the pool couldnt recoup more from them.

Now, regarding all the tracking withholding work I've been doing, here's an update on it:

Firstly, there's a bunch of miners in the subnet 91.218.B.* (B=8-11) that are running hardware that doesn't produce blocks.
Try 'whois 91.218.8.0'
They know it also, since they've changed around their settings, email addresses, swapped to address accounts, and even stopped mining here.
I've emailed them and got no reply.
The main account that alerted me to the problem was the username "foma" that is certainly withholding.
That says: submitted 53G diff of shares, ~690K shares, 11 were over 1G diff, smallest over 1G was 1.02G, biggest over 1G was 1.99G (avg over 1G = 1.32G)
1.99/53.05 = 3.8% luck or 53.05/1.99 = 2666% Diff ... yeah way too high ... 1000% is a good indicator that something is almost certainly wrong.
Now of course there's the actual best share number 1.99G which is less than 2^31 - so it's hardware that wont produce a block or even a share over 2^31

Their total mining since the first account (foma) was created on 19-Jan has been 52.05339 BTC
The total I've held back (over the past ~week) is: 5.96578576 BTC (of that ~52 BTC)
So they've made ~46BTC mining while withholding since January.

During that time the pool luck has been ~110%, and their mining has been very roughly 0.94% of the pool total.
If you happen to know anyone on that subnet in Russia, let them know your dissatisfaction with their mining and the fact they've been taking BTC from us without the possibility of producing a block for the pool (which they know)

--

There's a second one also.
His statistics are extremely suspect for a similar, but not exactly the same reason.
His account name is saradis (plus others) ... who happens to match an account mentioned in the BCMonster pool thread ... I'll link them this post
(I have already told them and Johnny about foma)
It's a smaller account and I need to spend more time on analysing it.
The luck is below 9% (above 1100% diff) the max share is below 2^32 (not 2^31)
I'll post more details later.

--

There are 2 more accounts that I put on hold, that I have since let out of jail.
I'll not say who they are since it would appear I jumped the gun on them and they are not guilty.
One was by association of being in a subnet related to 'foma'.
They've not lost anything, just didn't get some payouts that will be in their balance until they either ask for them or I get around to sending out dust and unpaid balances ... some time in the future ... in a galaxy far far away.
... and that's IMO the best thing about how I'm dealing with this, any payment that I hold back isn't lost, if they turn out to not be guilty of withholding.
(and as I said before any amount like the ~6BTC above that I do hold and they are withholding, I'll later distribute to the payout that should have got it, once I've setup that sort of thing to work with the pool without having to manually do it)

I guess it is normal behavior then.  I'm fairly new to the renting thing (I prefer my own equipment), but about a week ago I parked some rentals on WestHash at 5%, 10%, and 15% under market (after fees) just to see if it is worth it or not.  I don't change the rates or cancel them (except at the network diff change)... just let them bubble to the top of their price points as other rentals come and go.  I refill as necessary.  If I adjusted anything I would end up at the back of the line.  They're usually inactive except for a few hours here and there, but I figured I would give it a try.

Seeing as the work diff gets stuck at the maximum diff ever seen for the worker, I suppose that I should use multiple worker names in the future (one for each rental point).  It's not a big deal since it doesn't affect rewards though.

Kano clarified for me that the number of stale shares isn't affected which was my biggest concern without knowledge of the technical specifics.  When I stop and think about it, I shouldn't have been worried about that either.  No problem here.

It didn't change on the downward on this experiment and even when everything is done, it still shows at 5.4K diff.  Same with my last rental from a long time ago showing at 144K as it was much much higher hash rate.  When it dwindle down to 0, it didn't bring down the difficulty.   I only see if adjust when you end it and restart the rental.

If it doesn't affect rewards then I'm not bothered by it, but the work diff does seem stuck so I reported it.  It looks like I may have briefly peaked at 20T near the beginning for less than a minute (typical WestHash over-provisioning), but even that doesn't equal 44k (though I suppose it could have been 100T+ for 5 seconds which showed as 20T to me... I have no way to know).  That said, I would have expected the diff to adjust downward in the 2 1/2 hours that followed (far more than a few minutes).  Doing the math, I was averaging about 1.5 shares per minute for 2 1/2 hours.  Of course, since I saw 4 minute gaps in there, there must have been some faster bursts mixed in.  Maybe I just had a very choppy rental??

I did use the same worker name back during the >600% forever block for a bigger rental (50T range) and the stuck diff seems more in line with that rate.  Could something simply be stuck for that worker from way back then?

Block by sabercatt with 45THS!  That's his 2nd kano block!

Man, it's good to get that one out of the way...

I did a quick test for you.  This is what I'm showing via a 5TH rental pointed at Kano.



First 1minute and 50s, NICE HASH is pumping in less than 1TH but the share rate is showing much much higher on Kano as it's flooding in

Worker Name:   Work Diff   :Last Share   Shares   :Diff   :Share Rate   «Elapsed   :Invalid   Block %   :Hash Rate
nhando1977.xx   2.048k   1s   140   286,720   10.57THs   1m 54s   0.000%   0.000%   280.21GHs


5Mins later  (Work diff jump up to 5.472K).  Of course I bought only 5TH but got almost 9TH due to setting high price so it can hit quicker for this test.

Worker Name:   Work Diff   :Last Share   Shares   :Diff   :Share Rate   «Elapsed   :Invalid   Block %   :Hash Rate
nhando1977.xx   5.472k   0s   202   602,016   8.53THs   5m 0s   0.000%   0.000%   550.96GHs


10 Mins Later (Same Same)  

Worker Name:   Work Diff   :Last Share   Shares   :Diff   :Share Rate   «Elapsed   :Invalid   Block %   :Hash Rate
nhando1977.xx   5.472k   54s   292   1,094,496   7.61THs   10m 14s   0.000%   0.000%   1.21THs

I stop the rental at 9mins and 30 seconds to see if the Diff will drop.  

At the 20Mins mark, the diff stayed the same

Worker Name:   Work Diff   :Last Share   Shares   :Diff   :Share Rate   «Elapsed   :Invalid   Block %   :Hash Rate
nhando1977.xx   5.472k   11min   292   1,094,496   3.90THs   20m 0s   0.000%   0.000%   1.18THs


Conclusion?  NH must have at some point bump your hashing limit to much higher than what you set it to, so the difficulty was adjusted automatically to the higher end but it never did drop back down as your hash rate settle back to where it was set as a limit.  I know that for my Megahash rental my difficult rate is really really high but it's been a long time since I rent any hash rate for Kano.

Firstly, it takes both a few minutes and a few shares before difficulty will change.

Secondly, the difficulty has the same 'no effect' on rejected shares as it has on accepted shares.
'no effect' meaning it only affects variance, not the expected average result.

As an example, if you are mining at 100,000 diff instead of 1,000 diff, you will find 1/100th the number of shares.
Shares still have the same chance of being accepted or stale, but since you have 1/100th of them, you have the same expected average rejected diff vs accepted diff.
Since the variance is higher (fewer shares), it takes longer to get the expected average.

If you happen to have an early run of a stale share, that's, unfortunately, just bad luck.
You can have the exact opposite of course also, of having a long run of no stale shares and thus having a way better reject ratio.
That's variance. There's no expected gain or loss, just variance.

Not extra hash-- I know about that.  This is the work difficulty at the pool that I'm asking about.  It seems to be stuck at 44,196 even though the rental is only running at 6T (well, actually fluctuating between 4T and 10T as usual, but averaging around 6T).  This is what my workers page looks like (WH worker is the rental):


I checked my worker/management settings and that worker has a minimum diff set to 2,052 (so that WH doesn't complain if it goes under 2,048).  The rental has stopped now (rates went up), but while it was running it was only turning in shares once every 30s to 3 minutes.  There were only 155 shares through the entire rental duration.  

It seems the pool vardiff should have come down from 44,196, but it remained stuck there throughout the entire rental.

Edit: And now it has started back up-- WH is giving me just over 5T but the workers page says it has been 4 minutes since the last share and the diff is still stuck at 44.196k.  Last share was stale.  Delta is now -83% red.

No fix, they will dump extra hash for you regardless of what you set your limit at depending on the surplus amount.  So you got to monitor it closely.   I've seen them give me hundreds of extra TH when I'm not paying attention.   You can watch the same results on the list and see how when rental demand is lower, most everyone get more hash then what they set their limits to. 

One of my WestHash rentals pointed here that has been idle for days just came to life when the rental price broke into the 0.0028 range.  It is acting peculiar today though-- I'm only renting 6T but the worker diff has shot up to 44,196!  It looks like that worker is only turning in a share every couple minutes.

Has anyone ever seen this happen before?  Is there a fix?

I don't want to risk stale work or something.

Yesterday I added another Avalon 6 to this pool.  This one is a performer too... It easily averages around 3900GH on the pool without even being turned up to the maximum voltage it will take.

Setting this one up should have been easy, but Murphy was conspiring against me yesterday.  It took several hours to get everything working correctly.  Part of the process was reducing the speed on two of my SP20s to move them from two to one PSU to free up one PSU for the new A6.  Everything was planned well in advance, but I managed to mess it up.  In the process I embarrassingly managed to get some PSU cables crossed up, and wound up overloading a supply when tweaking a miner setting because the cables were not where I thought they were.  After resolving that issue the new A6 wouldn't detect in the rPi, which turned out to be a bad SD card (the last thing I checked, naturally).  After fixing that and getting it running I discovered a configuration issue where my Avalon 6s were trying to act as DHCP *servers* and causing problems with IPv6 and DNS resolution of PCs on the LAN.  Turning off odhcpd on the Raspberry Pi fixed that.  Then just as I finished tidying the Ethernet cables with zip ties one of them went bad and I had to undo all of the ties to replace the cable.  Of course in frustration I unplugged the wrong one......

Anyway, everything has been stable for 18 hours now.  Block please!

You are absolutely right...no one can predict what is actually going to happen.  I choose to take a positive approach...

this is so only if the market has not yet included the halving in the price...if it is already done...it will or can decrease.

Theoretically because of supply and demand...because there is less BTC coming in the BTC demand and price will go up.  That's what every miner out there is hoping for!