Topic: Elitehash Raven PPS Pools - stable rewards, payouts every hour! (Read 880 times)

Elitehash naturally supports the algorithm change for Raven, on October 01

Miners still have to change their config files, so the port number should probably change. Everyone just has to edit those configs, there's no way around it. Changing the port number is a good way to filter out those miners who didn't react and stil try to mine x16r. They'd be shown as rejected hashrate otherwise.

Raven is now the only pool on Elitehash, as I'm reworking the code for better scalability. Worst bugs have been fixed during the summer, and I'm sorry against any hashrate loss due to them. I sent a covering payout for most, and will do a final check of the history after I've done this final rework. There could still be a hiccup or two, but I think I've learned from a lot of mistakes here

In the meantine, Raven payout rate is steadily increasing again, and it might even go over 100% soon!

The new stratum server is now working great, still a bit of extra latency but it's not really causing noticeable rejects. Should be relatively easy to fix now. Connections to Helsinki & Paris were a bit unstable this past week, but now they are stable. Got the CPU usage down to 1% and fixed this pesky problem with stale connections.

I'm about to make a switch to this stratum server I made from scratch. It's replacing 17k lines of code I didn't write with a neat, 400 line package that isn't hard to understand, at least for me.

The dip in the hashrate graph is from my 1st test with actual miners (in the Helsinki server), which to not much surprise uncovered a bug or two Everyone was sent the *full* amount of what they should have earned during that 4 hours of bugged share counting, calculated from the previous 4 hours: https://ravencoin.network/tx/3143984c32d0ec6148fc8f171e795f262b150e676cd188c38fc571f0756cbaba. In addition to the ~50% or so of shares that weren't lost due to these bugs.

The switch should be transparent to miners, but the end result should be definitely noticeable: share difficulties won't be absurdly high anymore. During the last test, I had a constant 32 for 4 hours.

(For anyone interested, the bug was related to duplicated stratum message IDs. Easy to fix by adding an ID translator.)

Worker lists and Paris server were added this month. Not sure where the miners are going, as Raven is still at the top in WhatToMine. I'm a bit hesitant to add every new coin, at least until they've been proven somehow.

Payout rate is going up now, to help gather more miners & also because higher rates really are easier to sustain with a smaller pool. Suggestions for improvements are welcome, as always. A payout list with txids is coming, but I don't have any other big improvements planned now.

Rejects have been over 1% today, my new approach is to increase the payout rate in situations like this. It's working as intended, can't avoid the higher reject rates totally, but the higher payout rates should compensate. For a moment I thought about making the rate go up automatically in such situations, but that would so ripe for exploitation it's better done manually.

I've toned down the automatic restarts, clearly it wasn't a good approach. Now the needless restarts are kept to a minimum, while rejects are being more closely monitored with automatic restarts of specific ports in case of problems. Overall reject rate is again below 1%, which is the target. It can be higher than elsewhere on this pool, but the high payout rate acts as compensation.

As a result of the changes ensuring no extra shares get counted, all the stratum ports now automatically restart multiple times per day in a continuous rotation. Connections are automatically routed to the slowest port, so the chance that such a restart causes even a second of idling is low. But most miner programs today are too dumb to return from a backup pool on their own. These automatic restarts can trigger the switch to a backup pool, so if you have other pools than Elitehash servers defined, it might get annoying to track if rigs gets stuck on backups too often. Please let me know if this has become a bigger issue - the automatic restarts have been there since the beginning, but with much less frequency until now.

The payout rate has been temporarily dropped, as I was investigating a problem with extra shares being credited to people. Just look at those graphs on Helsinki or Virginia, lots of people have huge inexplicable spikes upwards. I'm hoping the drain is finally plugged now, watching it closely.

EDIT: It's indeed looking better, as I guess you can see from the lack of payout delays. Payout rate starts climbing back to 100% now

Payout times changed to once every 20 minutes per server, hopefully that's OK. I want to keep the amount of owed coins smaller.

We aren't currently generating our own blocks actually, difficulty got so high it made more sense to spread the hashrate to get lower variance. But I'll still keep them updated, of course, thanks for the reminder!

2.1.2 asset support update? Thanks again for the great pool.

Singapore server online. Still working on improving those loading times, they should be better now but sometimes it's not so fast

Updated all the servers earlier today

Love the pool. Good work. Just making sure our mining is voting for the asset layer hard fork. I don't see anything about this on the website. Thanks!

Unfortunately I realized that applying txids to past payouts is harder than it first seemed. Because the pools are spread over many servers, sharing the same wallet, I can't just search for all the transactions in the wallet matching an address found on a server - most likely the same address has mined on another server too, and I don't know which transaction is from which server.

The best approach I think would be go through the balances database and match the payouts with transactions where the amount, address & dates match. I'll consider doing that when I get the system ready for storing txids for future payouts.

But some payout histories are unfortunately too hard to restore. On other servers all the balance changes are intact, but I messed them up when moving the Helsinki server a couple of months ago. It still has all the shares since the beginning, but the balance database doesn't contain entries before the server move. They were lost because of my carelessnes - for anyone affected by this, and who would like their complete transaction history, I can compile it manually for their address, send me a PM.

I changed the servers to do payouts at a fixed minute every hour, to avoid a rare double-spend problem that happened today. It wasn't the first time. In total 10k RVN of the 4M+ paid out weren't actually paid out as the servers attempted to use the same inputs and one of the transactions obviously never gets confirmed. After an user mentioned a payout not making it through today, I checked the history (it's easy as the listtransactions RPC command shows them having a negative amount of confirmations) and found 10k RVN worth of failed payouts.

As the payouts now happen in turns, this can't happen anymore. I'll compile the failed transactions to a single big one and send it soon. People have occasionally requested a txid list for payments, but the simplest implementation would probably have just listed the failed txids there with no-one being any wiser. Looks like the list should also check that they have been confirmed, making it a bit more complicated.

EDIT: failed transactions have now been resent

Works for me 

It's my only way to control the size of the pool. It's true that having the rate change almost daily is different compared to other pools. At first I was a bit wary of adjusting it too often, made sure to give proper notice etc. Loosening up my own rules there has worked well in practice, though. I have realized that doing the changes without notice mostly helps drive the change I want, compared to giving notice. This might annoy some, but those who have stayed long enough realize it works both ways. I don't go drumming around the changes upwards, either. Can keep them up longer that way.

Why has the payout rate dropped to 98.5%?

I prefer hourly payouts. Every 420 seconds was a bit too "spammy".

Payouts are now hourly until the stats calculation gets a whole new dedicated server. Also the rate is now only 100.3%, the pool was getting a bit too much growth at the higher rate.

THX, all working

Fixed now. PHP code was missing curly braces at a for loop. I should really get rid of all PHP files in the pool and make it all python

I received about 30 payments - but there is no summary statistics on the EU server http://rvn.elitehash.net/RVsyqS2fuYzT5HdwsnCY5yEAh9XPo5qr93
only record - Balance
wallet RVsyqS2fuYzT5HdwsnCY5yEAh9XPo5qr93 mined only EU server

If you've never mined on a server it shows no stats from it. Unless you mean that you have mined on both and it's still not showing?

why on eu server there is no information about payments

example: http://rvn.elitehash.net/RVsyqS2fuYzT5HdwsnCY5yEAh9XPo5qr93
no record: Payouts from

example: http://rvn.elitehash.net/RHEe8HpvAeqEMFrHq4USPLLwuU6U3fX47f
no record: Payouts from on eu server, only us

US1 server was down for 1 week, but now it's been replaced with a much faster server. Mining RVN doesn't look too enticing now, but just wait until the difficulty drops again! Payout rate is still 102.4%

The payout rate is now 102.4%! (Has been for some time actually)

Just added an EU pool, and the pools are now named http://rvn-eu1.elitehash.net and http://rvn-us1.elitehash.net - of course the old http://rvn.elitehash.net works too. I'll probably make it a round-robin hostname for both servers.

Hey there! The top miners list is showing 3 hour averages. Maybe it would make more sense with a tighter timeframe. I wanted semi-stable numbers there, but I admit it also adds unnecessary confusion. Everyone can already see their 1h/3h/6h/12h/24h stats, but perhaps some miners never get there after being scared off by the initially low numbers

I just moved 260mh hashing power over.  Let's give it a try and see if it is good

Although I'm still at the last of the bunch showing I'm doing 10 mh....need to update on that refresh time yo'

Welcome Elitehash pools, http://rvn.elitehash.net

There's practically no variance on payouts as it's PPS, and the payouts start rolling almost instantly. Balances above 5 RVN are paid multiple times per hour (ask for custom thresholds!), idle balances are paid out quickly too. The pool payout rate varies but I try not to make radical changes. Obviously nothing like going to 50% payout rate suddenly. It's displayed like that because it's been over 100% - though that's unlikely to return in the foreseeable future.

If the pool loses hashrate noticeably, payout rate is probably increased, because less variance has to be absorbed.
If the pool gains hashrate rapidly, payout rate will probably decrease, unless it seems possible & reasonable to endure the increase in variance.
If the pool hashrate doesn't move enough to warrant changes, payout rate may still be adjusted in small ticks.

These principles should help understand how the rate behaves. I have a natural incentive to abstain from making shocking changes, but that doesn't mean I'm perfect. There have been 2 or 3 situations that I would've liked to have prepared better for, when I felt like I was making changes too suddenly but was forced to anyway. That number would be double if I hadn't opted to give the pool some 'loans' to bail out from bad luck. I'm glad those have been repaid already, now the focus is on better stability so it doesn't have to happen again.

Perhaps we can return to focusing on high payout rates at some point. I thought that - having started almost a year ago - it would be obvious that I have an incentive to keep things running, and hence fix any delays in payments. These happened semi-frequently when I still tried to keep the payout rate as high as possible. Experience shows that people didn't think like that; the level of panic was higher than anticipated. Today it's the opposite, I really expected more sullen comparisons agains past rates. Of course it's only fair to compare against real results from current pools, but not everyone cares

It's odd how silent all miners are now. Surely there are more features people want, assuming there are no problems (there always are...)? Naturally I'd like requests for features that don't even exist anywhere else yet. Don't let that turn you down!