Bitcoin Forum

It's been a while since a block hasn't come out below the 731kB softcap.     :o

https://i.imgur.com/bVa1lof.png

The next block came out at 976kB

Yes, loads of TX waiting for a confirmation lately. Even those with a fee

https://i.imgur.com/ywvHuF7.png

This is a serious issue needing a solution.

Hmm. Seems like the network is really stressed. Too much transactions "clogged"?

There don't even seem to be an excessive amount of transactions in these blocks that are approaching 1 meg.

Perhaps someone found out how to create really large transactions and clog up the network?

if that's what is happening we're all doomed , we need to make the dev's aware of this someone should contact them :>

We tried, but the fees included with our requests weren't high enough to warrant their attention.  ;D ;D ;D

I doubt that large (kB) transactions are that hard to figure out how to do.  I doubt it's malicious. 

Well, daily transactions is over 100 000 now ... all days !  8)

We are not far off the 1M max block size now. The max block size solution Gavin proposed should gain more support.

From what I can see we've seen a lot of blocks today that have been mined by pools that are using the default 750,000 byte block soft limit (these show up as 731 kBytes), while a few have been mined by pools that support the 1M bytes block limit (these show up as 976 kBytes).

Now what's actually happened is that for quite a number of hours the network was having an "unlucky" day and didn't find many blocks. It wasn't hugely unlucky (there were no hour+ gaps between blocks as happened 3 times in 24 hours a few weeks ago), but it was moderately unlucky.

Starting at 11am GMT these were the numbers of blocks per hour: 3, 3, 2, 4, 4, 8, 5, 7, 8, 4, 9.  Over the 11 hour period we had 57 blocks whereas we might nominally expect 66. 57 out of 66 is actually surprisingly common. Much less common (but not totally unexpected) is that in the first 5 hours we only had 16 blocks instead of 30. Again it's not totally unexpected but it's less common. The next 6 hours saw 41 instead of 36 though.

Now total transaction volumes for the day aren't anything unusual (they're steadily climbing but Saturdays are always quieter than during the week), but it seems that there have been some reasonably large size transactions today as the mean block size is 0.424 Mbytes. What that means is that we had a period of time where for 5 hours the transactions could back up. I ran some estimates for the effective network block capacity a few weeks ago and it's around 770,000 bytes per block, but if we're only finding 16/30ths of the normal blocks in our 5 hours then that drops to an effective capacity of nearer 410,000 bytes.

What this means is that for 5 hours we were seeing more transactions than the block selectors were prepared to process (not the network, the miners, or more precisely the mining pools!).

What's interesting are the comments that larger fees didn't help. This suggests that some block makers may not be selecting transactions based on their associated fees. Of course no fee, no matter how large, can make blocks be found any quicker so if there's a 40 minute gap between blocks then you still have to wait 40 minutes.

If anyone's not already seen them, here are the more formal write-ups:

http://hashingit.com/analysis/33-7-transactions-per-second
http://hashingit.com/analysis/34-bitcoin-traffic-bulletin
http://hashingit.com/analysis/39-the-myth-of-the-megabyte-bitcoin-block
http://hashingit.com/analysis/41-waiting-for-blocks

Thanks dave. Very informative.

By the way, the average block size is now around 400KB, slowly increased from 300KB 6 months ago.
http://btc.blockr.io/charts

Looking at the list there are huge numbers of transactions paying no fees at all, or paying dust-sized fees. That suggests a lot of people are quite happy to wait however long it takes to find a block maker who's feeling generous and mines them. I use the word "generous" advisedly as most fees don't pay enough to cover the incremental risk of having an orphan race lose the associated miners their block reward, but zero fees are a joke as far as economically sane miner is concerned.

That's the average that's lowered by transactions with no fees that wouldn't be confirmed for a long time or ever at all anyway. If many transactions included in that statistic has equal or above recommended fees, we could have a powerful "selfish" miner problem.

EDIT: Beat me by a few mins, davejh.  ;D

Dust bowl approaching :o

http://www.english.illinois.edu/maps/depression/dbimages/dust2.gif

I think it's time maybe we started looking into some solutions, to partition part of the network, just the transaction wait time thing, it definitely needs work.

The devs should think about adding lower hash rate confirms to make the program be aware of priority transactions. I was thinking of setting up queue areas for transactions that would confirm them as accepted by the network and flag that it is in the process of  verifying it. Think of it as a pre-double spend clearance, weeding out unwanted transactions for the main network to confirm.

I was thinking of holding these incoming transaction in hash-trees: Transaction comes in gets added to the initial queue, Miner individually checks the transaction, tells the network it has it in it's personal Hash-tree, if other miners have the transaction they will add it to their own queues, eliminating the double spend at a lower hash rate. We could get a whole lot more transactions waiting in memory and then just confirm the Hash-tree transactions... effectively we have to find a way to compress the transactions further and compressing whole sets of 1mb blocks worth of transactions into hashes... could let us do that.

A lot of the work for figuring out parameters is left for miners to figure out, there has to be a way for the program to become self-aware enough to adjust them itself depending on fees, time, or position in the queue.

http://keccak.noekeon.org/Sakura.pdf

Its because you are loosing miners left and right. No miners means no transactions processed, at a difficulty of over 46 billion, and a value of only 200 to 230 bucks, its just not profitable to mine bitcoin. All your small miners you guys think are worthless are now mining other coins and flipping them for bigger profits, all the big mining compuonds are bogged down with huge difficulties. This is because the big compounds have no idea what they are doing as far as securing the network, all they care about is making X amount of dollars for solving X amount of blocks so they can pay the electric and have enough left over to fuel their yacht.

The small miners are needed a lot more than this community would liek to give them credit for actually, at this point with sometimes 2 hour confirms on bitcoin transaction the bottom line is, you need more computing power, and its just not worth it to mine Bitcoins right now.

the amount of miners has nothing to do with the amount of transactions that can be processed.
more miners = more security: no more and no less

(though if miners are leaving the diff is lowering and blocktimes raise which lowers the tps a little. but i dont see this as a problem right now)

lol!

My transactions are taking a really long time to confirm. Maybe we need to do the hardfork sooner...

haha.. yuuup..


another one.  sooner than what?  my transactions confirm in the normal time.  go away.

Amazingly ignorant and incorrect on multiple levels. My last post to you before you are ignored, his post history is depressing. I recommend everyone else ignore this troll as well.

Difficulty is increasing steadily-
https://bitcoinwisdom.com/bitcoin/difficulty

Which means blocks are being found quicker than average and more miners are coming on board.

Yes, we need a hardfork to increase the block limit but there is no evidence to suggest that confirmations are slower than normal.

Is this really news? I recall several thousand tx being in the queue as the norm for Bitcoin.
In addition, all my transactions with the proper fee have been accepted as per normal.

I think if we start to see 10,000 unconfirmed fee paying tx then we might need to raise the alarm.

Like someone's earlier calculations in this thread suggest, the average fee of the unconfirmed tx in queue is really low.

yes we do, and maybe thats what he meant but his post made 2 clearly unrelated statements.  just pointing that out.

If you rate the level of network stress by how close the recent block size is to the 1M cap, I would agree we are stressed. I think the large block size could be a result of high difficulty. Miners are finding it difficult to solve blocks as quickly as before. While they take longer to solve the block, they add more transactions and hence the increase in blocksize.

Current blocks solved per hour according to bitcoincharts.com is 5.89. This is only slightly lower than the target rate of 6.0/hour, so you can't really blame difficulty.

for a hero member you sure are belligerent.

For a full member you don't seem understand how the forum works very well.

Rank means nothing, it's just how long you've been around and how much you participate on the forum.

Everyone has their own opinion, and so if you would try to not be an asshole that'd be great.

----------------------------------------------------

Related to the actual issue at hand, where would the clogging of the network stem/originate from, assuming it is a clog (However unlikely)?

And the statement of losing miners may be true for small groups of people. There are still large corporations (Such as cex.io) that still have plenty of mining power behind them, but that isn't even necessary.

THE DIFFICULTY WILL NOT INCREASE AS RAPIDLY WITH FEWER MINERS. This part of the Bitcoin protocol is very useful for scenarios such as this.

I'm not sure if I'm saying the correct things here, so feel free to correct me. It's been a little while since I was really active on this forum, so I'm not sure what has changed.

One problem that no-one has mentioned here is that in many respects it doesn't matter how large the theoretical block size limit really is, it matters what the block makers choose to accept. Right now, for example, most will only mine up to a maximum of 750,000 bytes even though the protocol allows 1,000,000 bytes (the default is 750,000 but can easily be changed). Some block makers are still using much smaller values, however.

The reality is that if pretty-much all of the major mining groups (many of which are privately held and therefore cannot be influenced anywhere near as easily as public pools) decide to keep 750,000 bytes, or adopt some other arbitrary smaller limit, then it doesn't matter if Gavin or anyone else changes the protocol to allow 20M bytes or 20G bytes, blocks will still be handled exactly as they are right now.

In practice mining 20M blocks isn't going to make any sort of sense for miners without something like the IBLT change that's also under consideration as otherwise the risks of losing an orphan race when announcing a very large block compared with another miner announcing a much smaller one will dramatically outweigh any theoretical advantage in terms of fees (which are already too low to be of any real interest anyway; larger blocks will push them down).