3
2
5
...
4
Thanks a lot for your work so far
Topic: [POOL][Scrypt][Scrypt-N][X11] Profit switching pool - wafflepool.com - page 273. (Read 465769 times)
February 08, 2014, 03:57:22 PM
1
3
2
5
...
4
Thanks a lot for your work so far
3
2
5
...
4
Thanks a lot for your work so far
February 08, 2014, 03:48:55 PM
OK, Pool is at a good spot at this time. Servers are stable, switcher is running extremely fast (made some modifications, broke it up a bit), watchdog processes are watching more than ever. I've got a few things on my list, but I want to get some public responses before I jump in and just start working on stuff, what you guys think would be best to work on.
Some options (and what I think priorities would be):
1) Reworking the earnings log (high prio). Our earnings log (what shows each miners coin earnings) is super ugly. Its not just ugly client side, its ugly server side as well, and reasonably heavy to run. I'd like to clean this up (would take about a day), but would end up being a useful thing on the stats page. I'd ideally like to do this first, while we discuss what to do on #2.
2) Adding more coins (medium prio). We've freed up some space by removing a few less profitable coins, we have the ability to add a few more without any problems. However, there are a number of things that this brings up that complicate it. Most of the new coins that I can find to mine are only traded on exchanges without useful APIs (coinedup, coinmarket, etc). This means that trading would need to be manual, and like I've said before, this opens up a ton of risk when mining these coins (they sit in an account for up to a full day before getting exchanged), it also means a LOT more work on my part. Right now, all I do is withdraw btc [2min], wait for confirmations [10min], and hit send [1 second]. With manual exchanging, it would be a matter of logging on to multiple (maybe) exchanged, manually trading, manually importing those trades into a new interface on wafflepool, and then withdrawing. I'm not against it, but it becomes a much larger portion of time (less time to work on new features).
3) Per-worker stats/graphs (medium prio). Our miner page is a bit shitty currently. I'd like to clean it up, and get per-worker stats going (graphs, stats, etc). This wouldn't take too much work, just a matter of adding a lot of logging tables (we're not currently tracking almost anything on a per-worker basis).
4) New front-end to attract more visual-based miners (low? prio). Our current site is a bit, shall we say, "barebones", and while I really prefer getting data dumped in my face, a lot of miners want pretty stats, big bright numbers, and shiny widgets. I'd suggest building a new front-end for the site that is visually appealing, and hides a lot of the more complex stats. We would show overall numbers (btc/day, hashrates, some graphs), but hide things like per-coin values, just to make it very simple, and completely summed up in 10 big numbers/graphs. This would replace the current front-end as the default, and the current front-end would be moved to something like "raw.wafflepool.com" and be linked from the new site. This is a purely "grow the pool" move, nothing more.
5) Suggestions?
I'd love to hear from you guys how you want my time spent. I want to make the pool grow, and think the best way to do that is to give miners what they want
Some options (and what I think priorities would be):
1) Reworking the earnings log (high prio). Our earnings log (what shows each miners coin earnings) is super ugly. Its not just ugly client side, its ugly server side as well, and reasonably heavy to run. I'd like to clean this up (would take about a day), but would end up being a useful thing on the stats page. I'd ideally like to do this first, while we discuss what to do on #2.
2) Adding more coins (medium prio). We've freed up some space by removing a few less profitable coins, we have the ability to add a few more without any problems. However, there are a number of things that this brings up that complicate it. Most of the new coins that I can find to mine are only traded on exchanges without useful APIs (coinedup, coinmarket, etc). This means that trading would need to be manual, and like I've said before, this opens up a ton of risk when mining these coins (they sit in an account for up to a full day before getting exchanged), it also means a LOT more work on my part. Right now, all I do is withdraw btc [2min], wait for confirmations [10min], and hit send [1 second]. With manual exchanging, it would be a matter of logging on to multiple (maybe) exchanged, manually trading, manually importing those trades into a new interface on wafflepool, and then withdrawing. I'm not against it, but it becomes a much larger portion of time (less time to work on new features).
3) Per-worker stats/graphs (medium prio). Our miner page is a bit shitty currently. I'd like to clean it up, and get per-worker stats going (graphs, stats, etc). This wouldn't take too much work, just a matter of adding a lot of logging tables (we're not currently tracking almost anything on a per-worker basis).
4) New front-end to attract more visual-based miners (low? prio). Our current site is a bit, shall we say, "barebones", and while I really prefer getting data dumped in my face, a lot of miners want pretty stats, big bright numbers, and shiny widgets. I'd suggest building a new front-end for the site that is visually appealing, and hides a lot of the more complex stats. We would show overall numbers (btc/day, hashrates, some graphs), but hide things like per-coin values, just to make it very simple, and completely summed up in 10 big numbers/graphs. This would replace the current front-end as the default, and the current front-end would be moved to something like "raw.wafflepool.com" and be linked from the new site. This is a purely "grow the pool" move, nothing more.
5) Suggestions?
I'd love to hear from you guys how you want my time spent. I want to make the pool grow, and think the best way to do that is to give miners what they want
February 08, 2014, 12:30:12 PM
yes
February 08, 2014, 12:10:24 PM
Hi, I'd like to know what these mean in the stats-page of my btc-addr.:
So is my total earnings 0.00398537+0.00346067
Quote
Bitcoins earned (not yet sent): 0.00398537
So these are already bitcoins that have been traded from alt-coins?Quote
Bitcoins unconverted (approximate): 0.00346067
Are these my earnings that aren't yet traded for bitcoins?So is my total earnings 0.00398537+0.00346067
February 08, 2014, 11:04:21 AM
Had a PM from comeonalready, which brings up a great question (and a reasonably long answer), figured I'd post it here just so everyone has the same information
These small doge blocks are just issues that need tweaking (slightly faster response time). The only times I've seen this happen are in special circumstances (where we didn't intend to mine this block, it just happened). When I've seen it, we start mining a high value block (say 700k or so), and between the time we start mining, and the time we check profitability again (can take 10+ seconds due to exchange pricing delays, block processing cpu spikes, etc), someone already mined the high-value block, and we mined a small block immediately after it. I've also seen it happen (not on doge, but on luckycoin), where we will notice the block value change, tell miners to switch, and have some miners (especially on very high intensity) still submit shares 2-3 seconds after being requested to switch. When these shares come in, we still process them for a bit (until that coin's block changes), and if someone submits a valid block for a coin we just stopped mining, we'll still submit it to the network (why the hell not!), and can get smaller block rewards.
In the first of the cases, we can fix it by watching closer (which just needs some tweaks), the second case is just a matter of a few seconds of jump time between coins, which gets solved (slightly) with more endpoints (lower latency).
In the first of the cases, we can fix it by watching closer (which just needs some tweaks), the second case is just a matter of a few seconds of jump time between coins, which gets solved (slightly) with more endpoints (lower latency).
I thought perhaps the reward estimate might have been off at the time, but I've been following the recent estimates posted on the stats page and the subsequent generation rewards in the block chain and they indeed do track properly. I then considered the possibility that we very quickly found the next block after the one deemed to be profitable -- before being able to switch to another coin, but there was a 66 second gap between the two blocks being found, so why didn't we switch to another coin? What happened? At 193,881 doge reward, the profitability for that block would have fallen below even that of litecoin, which always has sufficient exchange volume, so how is it that we were still allocating our resources to doge under those circumstances?
You're absolutely correct on those blocks. We've been tweaking (every day or so) to see what the right amount of time between coin switches is (how quickly we should react to new coins being most profitable). The answer right away might be "instantly", but there are some subtle problems with it unfortunately.
Switching coins extremely quickly (and for some coins, this is every 3-4 seconds in testing - will be better after we remove some super fast coins) causes a ton of extra load on the server in terms of bandwidth (sending all new work to a thousand clients), and this actually props up our orphan rate if we find a block during those times (and if we're switching low-diff coins quickly, we find blocks pretty rapidly) because our network upload gets a bit slow, and block propagation speed drops, leading to orphans. One solution that we've played with here is to raise our transaction fee minimums on a per-coin basis, leading to smaller blocks (quicker propagation), and less outgoing bandwidth on new work for miners, but at the same time, drops profitability since you're not getting as much in transaction fees, its also detrimental to the individual coin network.
Another issue is queued work on clients. A large number of clients, even after they've been asked to switch coins, will continue submitting work on the old coin for up to 30+ seconds (I think this is due to queued work, not 100% sure, if someone wants to test, please email me and we can setup a test!) Basically even after we've told clients "doge isn't profitable, go mine moon!", depending on the number of people, we still have a very good chance of finding a doge block in the coming 30 seconds. And if our switcher missed the block change by a split second (and didn't see it for 30 seconds later or something), we could easily be finding blocks.
That said, a LOT of this is tweaking based. And I'm constantly switching it up. New endpoints make this a lot easier too, since I can just switch one set of them (faster reaction time for USEast for example, slower for USWest) and see the results by zone pretty nicely. Just a matter of playing with it Smiley
It would help to start rejecting some of the shares that are late (but still submit valid blocks anyway). They dilute the earnings of properly configured miners who tend to submit useful shares
edit: and definitely switch instantaneously
February 08, 2014, 06:18:07 AM
Had a PM from comeonalready, which brings up a great question (and a reasonably long answer), figured I'd post it here just so everyone has the same information
These small doge blocks are just issues that need tweaking (slightly faster response time). The only times I've seen this happen are in special circumstances (where we didn't intend to mine this block, it just happened). When I've seen it, we start mining a high value block (say 700k or so), and between the time we start mining, and the time we check profitability again (can take 10+ seconds due to exchange pricing delays, block processing cpu spikes, etc), someone already mined the high-value block, and we mined a small block immediately after it. I've also seen it happen (not on doge, but on luckycoin), where we will notice the block value change, tell miners to switch, and have some miners (especially on very high intensity) still submit shares 2-3 seconds after being requested to switch. When these shares come in, we still process them for a bit (until that coin's block changes), and if someone submits a valid block for a coin we just stopped mining, we'll still submit it to the network (why the hell not!), and can get smaller block rewards.
In the first of the cases, we can fix it by watching closer (which just needs some tweaks), the second case is just a matter of a few seconds of jump time between coins, which gets solved (slightly) with more endpoints (lower latency).
In the first of the cases, we can fix it by watching closer (which just needs some tweaks), the second case is just a matter of a few seconds of jump time between coins, which gets solved (slightly) with more endpoints (lower latency).
I thought perhaps the reward estimate might have been off at the time, but I've been following the recent estimates posted on the stats page and the subsequent generation rewards in the block chain and they indeed do track properly. I then considered the possibility that we very quickly found the next block after the one deemed to be profitable -- before being able to switch to another coin, but there was a 66 second gap between the two blocks being found, so why didn't we switch to another coin? What happened? At 193,881 doge reward, the profitability for that block would have fallen below even that of litecoin, which always has sufficient exchange volume, so how is it that we were still allocating our resources to doge under those circumstances?
You're absolutely correct on those blocks. We've been tweaking (every day or so) to see what the right amount of time between coin switches is (how quickly we should react to new coins being most profitable). The answer right away might be "instantly", but there are some subtle problems with it unfortunately.
Switching coins extremely quickly (and for some coins, this is every 3-4 seconds in testing - will be better after we remove some super fast coins) causes a ton of extra load on the server in terms of bandwidth (sending all new work to a thousand clients), and this actually props up our orphan rate if we find a block during those times (and if we're switching low-diff coins quickly, we find blocks pretty rapidly) because our network upload gets a bit slow, and block propagation speed drops, leading to orphans. One solution that we've played with here is to raise our transaction fee minimums on a per-coin basis, leading to smaller blocks (quicker propagation), and less outgoing bandwidth on new work for miners, but at the same time, drops profitability since you're not getting as much in transaction fees, its also detrimental to the individual coin network.
Another issue is queued work on clients. A large number of clients, even after they've been asked to switch coins, will continue submitting work on the old coin for up to 30+ seconds (I think this is due to queued work, not 100% sure, if someone wants to test, please email me and we can setup a test!) Basically even after we've told clients "doge isn't profitable, go mine moon!", depending on the number of people, we still have a very good chance of finding a doge block in the coming 30 seconds. And if our switcher missed the block change by a split second (and didn't see it for 30 seconds later or something), we could easily be finding blocks.
That said, a LOT of this is tweaking based. And I'm constantly switching it up. New endpoints make this a lot easier too, since I can just switch one set of them (faster reaction time for USEast for example, slower for USWest) and see the results by zone pretty nicely. Just a matter of playing with it Smiley
February 08, 2014, 06:09:45 AM
Quick update from this morning.
4 coins have been removed from the pool: fastcoin, galaxycoin (we were never actually mining this, it was just tracked on the backend for testing), emerald, and hobonickels.
In total, there was about 0.009 btc value on them ($6 usd) when they were shut off. That amount will sit in the exchange account in case we ever start mining them again, otherwise, after 6mo or so, I'll consider it the first donation to the pool (free half-pizza!)
We've also fully disabled the original stratum endpoint. Anyone that hadn't updated their endpoints is being proxied to useast (same location as original). If you're one of them, you got dropped for a few seconds, and mining restarted. Update your hosts to one of the new ones
4 coins have been removed from the pool: fastcoin, galaxycoin (we were never actually mining this, it was just tracked on the backend for testing), emerald, and hobonickels.
In total, there was about 0.009 btc value on them ($6 usd) when they were shut off. That amount will sit in the exchange account in case we ever start mining them again, otherwise, after 6mo or so, I'll consider it the first donation to the pool (free half-pizza!)
We've also fully disabled the original stratum endpoint. Anyone that hadn't updated their endpoints is being proxied to useast (same location as original). If you're one of them, you got dropped for a few seconds, and mining restarted. Update your hosts to one of the new ones
I stopped using the EU stratum, since my rejected increased from 0.1% to 2% and its because of "Share Above Target" error
February 08, 2014, 05:58:18 AM
Quick update from this morning.
4 coins have been removed from the pool: fastcoin, galaxycoin (we were never actually mining this, it was just tracked on the backend for testing), emerald, and hobonickels.
In total, there was about 0.009 btc value on them ($6 usd) when they were shut off. That amount will sit in the exchange account in case we ever start mining them again, otherwise, after 6mo or so, I'll consider it the first donation to the pool (free half-pizza!)
We've also fully disabled the original stratum endpoint. Anyone that hadn't updated their endpoints is being proxied to useast (same location as original). If you're one of them, you got dropped for a few seconds, and mining restarted. Update your hosts to one of the new ones
4 coins have been removed from the pool: fastcoin, galaxycoin (we were never actually mining this, it was just tracked on the backend for testing), emerald, and hobonickels.
In total, there was about 0.009 btc value on them ($6 usd) when they were shut off. That amount will sit in the exchange account in case we ever start mining them again, otherwise, after 6mo or so, I'll consider it the first donation to the pool (free half-pizza!)
We've also fully disabled the original stratum endpoint. Anyone that hadn't updated their endpoints is being proxied to useast (same location as original). If you're one of them, you got dropped for a few seconds, and mining restarted. Update your hosts to one of the new ones
February 08, 2014, 04:11:26 AM
I have mined two days with 700Kh but didn't receive anything
Check your stats here http://wafflepool.com/miners and fill in the bitcoin address you used for your worker
You only get paid when you have 0.01 or higher
February 08, 2014, 04:09:28 AM
I have mined two days with 700Kh but didn't receive anything
February 08, 2014, 01:26:19 AM
I have 0.02685082 unconfirmed Franko on my address (hash: 71285d1e).
Haven't mined in a few days, something is stuck it seems.
Haven't mined in a few days, something is stuck it seems.
February 07, 2014, 11:11:55 PM
In that case, it should beat mining dogecoin alone...
Wafflepool has consistently (on average) far surpassed the profits of mining dogecoin alone.
This is false. Even if you mine the unprofitable blocks dogecoin is always more profitable. 0.0121BTC/mHash now and has been around 0.04 earlier
February 07, 2014, 10:48:41 PM
My two Euro cents RE: coin deletion:
Yes delete the rubbish coins. But try to replace them with some (seemingly profitable) newer coins if there are exchanges with reasonable APIs.
Also, if you're kicking your heels over the weekend (!), I would be curious to see in stats tab what coins have been mined over the last 24h & at what average hash rate.
Yes delete the rubbish coins. But try to replace them with some (seemingly profitable) newer coins if there are exchanges with reasonable APIs.
Also, if you're kicking your heels over the weekend (!), I would be curious to see in stats tab what coins have been mined over the last 24h & at what average hash rate.
February 07, 2014, 09:51:39 PM
In that case, it should beat mining dogecoin alone...
Wafflepool has consistently (on average) far surpassed the profits of mining dogecoin alone.
This is false. Even if you mine the unprofitable blocks dogecoin is always more profitable. 0.0121BTC/mHash now and has been around 0.04 earlier
February 07, 2014, 05:10:33 PM
Waffle this exchange has an API:
https://mcxnow.com/api.html
https://mcxnow.com/api.html
February 07, 2014, 05:00:59 PM
Doh, should've read that post a little better. True, when I use it on the EU server it works like a charm
February 07, 2014, 04:51:25 PM
Hmmmm, it looks like my fixed difficulty setting is being ignored. -p d=256, but yet I'm still starting @ 16 towards 1024 diff. 
Are you using one of the new stratum end-points? e.g. uswest, useast, or eu? If not, it won't work. PoolWaffle has said earlier in the thread that the difficulty system only works on the new end-points.
February 07, 2014, 04:43:29 PM
Hmmmm, it looks like my fixed difficulty setting is being ignored. -p d=256, but yet I'm still starting @ 16 towards 1024 diff.
February 07, 2014, 04:28:18 PM
In that case, it should beat mining dogecoin alone...
Wafflepool has consistently (on average) far surpassed the profits of mining dogecoin alone. Anyway, mining only dogecoin is a bit of a fool's game now, because you waste tons of time mining near worthless low-reward blocks. And then smart auto-switching pools like WafflePool hop on and snag a chunk of the valuable blocks, leaving pure-dogecoin pools kinda in the dust. This is likely the reason that dogecoin pools seem to consistently have <90% luck.
Dogecoin's design is pretty ridiculous, frankly. But the whole thing is just a profitable joke, so oh well.
I'll take a profitable joke over an unprofitable one any day.
February 07, 2014, 02:38:50 PM
In that case, it should beat mining dogecoin alone...
Wafflepool has consistently (on average) far surpassed the profits of mining dogecoin alone. Anyway, mining only dogecoin is a bit of a fool's game now, because you waste tons of time mining near worthless low-reward blocks. And then smart auto-switching pools like WafflePool hop on and snag a chunk of the valuable blocks, leaving pure-dogecoin pools kinda in the dust. This is likely the reason that dogecoin pools seem to consistently have <90% luck.
Dogecoin's design is pretty ridiculous, frankly. But the whole thing is just a profitable joke, so oh well.
Jump to: