Status as of Sunday, June 11, 2017

[Steve Sokolowski]

Good morning! It's finally summer today, with temperatures close to the record high of 90 degrees (32.2C). Chris is going to be changing engine and transmission and brake oils on his car today, while I execute yet another release to improve performance.

• I did an estimate this morning and think that we can get about 1.15TH/s scrypt before we CPU will max out again. Therefore, I set a target so that the database can also handle that much load without falling behind.
• The improvement is simple: instead of inserting every share in the database, we insert every other share, and add a "share_multiplier." The shares to be "counted" are selected at random. In the end, this makes no difference with what we or anyone else earns, but since shares were the largest consumers of data in the database, the amount of data will decline by 10%. With less data, we can then archive the old share data for our tax records, delete it from the live server, and rewrite the shares on the faster disks. We can still compute taxes and correct balances in the case of errors by multiplying the value of a share by 2. I programmed this feature so that we can change the multiplier to 4, 8, or whatever value is needed in the future.
• One of the issues impacting payout speed recently was that Novaexchange's system had a hardcoded $15,000 withdrawal limit. The owner told us to create multiple accounts and transfer money between them to get around their software limitation, but that meant that Chris had to manually execute four separate withdrawals to get $60,000 out of the exchange every morning. They created a new "Enterprise" tier, for which we are the first customer, that has a $100,000 daily withdrawal limit. This morning marked the first day where our system was able to automatically withdrawal from a single "Enterprise" account and pay customers more quickly. Please send a tweet to Novaexchange's founder thanking him for allowing us to pay you more quickly, and give him your business. Exceptional customer service is why Novaexchange and Poloniex are growing so fast they can't keep up, while Coinbase is losing people like us as customers because they do not reply to customer service tickets.
• Someone asked to be notified about mining server restarts. This release will occur later today and I will make a comment in the chat. Two-factor authentication is too intertwined with this release to sever it at this point without a painful fork, and the system is holding up for now, so if we can't secure two-factor authentication the release might be delayed until tomorrow.
• A quick note about delayed shares: the share processing delay doesn't cause any difference in what miners earn. It simply means that the database is overloaded now, and data in memory can't be written to disk quickly enough. Eventually the disk catches up when miners leave the pool and balances are updated. No money is being lost during these delays. The purpose for the delays is not to lose money - the system was designed to fail the least critical parts, like the website, and keep the most critical parts, like mining, online at all costs. It is working as designed.
• There are a few people who sent private messages with questions like "why don't you expand rapidly," "why don't you hire someone," or "why don't you quit your job?" I'll answer these questions in a more detailed post in the "Prohashing blog" later, but the short answer is that after this bubble crashes it's difficult to see scrypt profitability remaining above 1 cent, after prices fall and mining equipment floods the market. After the last bubble, bitcoin difficulty had a sustained period of decline as miners went bankrupt. The lesson is to make your money now while times are good. I think that a lot of companies who are buying large numbers of servers and the VCs throwing money into initial coin offerings are overinvesting and will go belly up during the downtrend. By not doing this, we are ensuring that we can survive the crash and remain profitable in the long term.

[JoeTheMiner]

• A quick note about delayed shares: the share processing delay doesn't cause any difference in what miners earn. It simply means that the database is overloaded now, and data in memory can't be written to disk quickly enough. Eventually the disk catches up when miners leave the pool and balances are updated. No money is being lost during these delays. The purpose for the delays is not to lose money - the system was designed to fail the least critical parts, like the website, and keep the most critical parts, like mining, online at all costs. It is working as designed.

It actually has a major difference though on what I am currently earning. Since I rent mining through my bot it has to be able to determine the profitability via WAMP. Since it is delayed my bot stops mining as it can't reliably determine the current profitability and whether to bid higher or lower on NiceHash. Is there any way to keep the database from overloading? This is major problem for me currently.

[lilbob]

True magicians. i thought it was science but its magic

[Steve Sokolowski]

• A quick note about delayed shares: the share processing delay doesn't cause any difference in what miners earn. It simply means that the database is overloaded now, and data in memory can't be written to disk quickly enough. Eventually the disk catches up when miners leave the pool and balances are updated. No money is being lost during these delays. The purpose for the delays is not to lose money - the system was designed to fail the least critical parts, like the website, and keep the most critical parts, like mining, online at all costs. It is working as designed.

It actually has a major difference though on what I am currently earning. Since I rent mining through my bot it has to be able to determine the profitability via WAMP. Since it is delayed my bot stops mining as it can't reliably determine the current profitability and whether to bid higher or lower on NiceHash. Is there any way to keep the database from overloading? This is major problem for me currently.

Oops, I forgot to mention that WAMP users are the people who are mainly affected by this. We added the timestamps are a stopgap measure to tell when the problem is occurring.

This issue should be resolved, at least until we hit 1TH/s or so, today or tomorrow, when we reduce the number of shares being written to the database.

[JoeTheMiner]

• A quick note about delayed shares: the share processing delay doesn't cause any difference in what miners earn. It simply means that the database is overloaded now, and data in memory can't be written to disk quickly enough. Eventually the disk catches up when miners leave the pool and balances are updated. No money is being lost during these delays. The purpose for the delays is not to lose money - the system was designed to fail the least critical parts, like the website, and keep the most critical parts, like mining, online at all costs. It is working as designed.

It actually has a major difference though on what I am currently earning. Since I rent mining through my bot it has to be able to determine the profitability via WAMP. Since it is delayed my bot stops mining as it can't reliably determine the current profitability and whether to bid higher or lower on NiceHash. Is there any way to keep the database from overloading? This is major problem for me currently.

Oops, I forgot to mention that WAMP users are the people who are mainly affected by this. We added the timestamps are a stopgap measure to tell when the problem is occurring.

This issue should be resolved, at least until we hit 1TH/s or so, today or tomorrow, when we reduce the number of shares being written to the database.

OK great!

[FRISKIE]

A thought about an "imminent" price crash . .

I was thinking today about the state of the crypto-space, and how it has matured well beyond what this whole space meant during the last serious crash. We now have a number of viable projects, Ethereum smart contracts, initiatives to develop BaaS (blockchain as a service), DApps and so on - what has been missing is that magic ingredient, broader public recognition of all this and involvement in realizing its potential.

That said, what we could possibly be seeing is not a bubble in any traditional sense, but the inflow of public engagement, some purely for profit, but a number of which will learn more, embrace the technologies, and generate more buzz.

The technologies are at the point of maturity that developers have begun to work on solutions with "real world" application, efforts which move the crypto-space out of it's Bitcoin-centric focus and into a world where a broader array of blockchain enabled technologies can see the light of day now that more people are paying attention.

Just sayin'

FRISKIE

[frphm]

I was, probably naively, unaware that cryptocurrency could crash so severely. What would bring that about? Bitcoin's UASF on August 1?

I'm inclined to agree with FRISKIE that crypto as a whole is much more useful than it was back in 2013-2014, and thus wouldn't experience a similar dive just because Bitcoin tanks. But that also might be me letting my hopes and dreams cloud the reality of the situation.

[FRISKIE]

and thus wouldn't experience a similar dive just because Bitcoin tanks

This seems reasonable to me, we'll have "corrections" along the way, some of them likely will be sharp, but I'm more inclined to believe in a mass migration from BTC >> ETH, LTC, LTCC etc. than an all out exodus from crytpocurrencies.

A disruption and major price correction resulting from possible mishandling of a Bitcoin fork is not all that unwelcome, and would be regarded on my side at least as a buying opportunity for any coin or asset that hits attractive lows.

[FRISKIE]

Another thought that came to mind as I was driving . .

There's a genuine unease with many people over the sustainability of the petrodollar system, and people have had a recent, real life example in Venezuela of a currency crisis and cryptocurrencies providing vital liquidity that enables a growing number people to carry on with their daily lives.

This has not been lost on many who fear a looming crisis with the US Dollar and now see cryptocurrencies as a hedge and alternative store of wealth.

It's one of many reasons that I don't fear a collapse in the crypto-space

[Steve Sokolowski]

I also wanted to mention that Chris seems to have fixed the nightly crashes. It seems that nThreads=2 is too much for nightly vacuuming. nThreads=1 takes longer, but doesn't interfere with the system, so he changed it.