Why are we bleeding nodes?
70 messages, 7 Apr to 20 May 2014. Started by Mike Hearn.
Greg Maxwell (11), Mike Hearn (8), Tamas Blummer (8), Jameson Lopp (4), Mark Friedenbach (4), Justus Ranvier (3), Tier Nolan (3), Jeff Garzik (3), +22 more
At the start of February we had 10,000 bitcoin nodes. Now we have 8,500 and still falling:
http://getaddr.bitnodes.io/dashboard/chart/?days=60
I know all the reasons why people *might* stop running a node (uses too much disk space, bandwidth, lost interest etc). But does anyone have any idea how we might get more insight into what's really going on? It'd be convenient if the subVer contained the operating system, as then we could tell if the bleed was mostly from desktops/laptops (Windows/Mac), which would be expected, or from virtual servers (Linux), which would be more concerning.
When you set up a Tor node, you can add your email address to the config file and the Tor project sends you emails from time to time about things you should know about. If we did the same, we could have a little exit survey: if your node disappears for long enough, we could email the operator and ask why they stopped.
phasing out of bitcoinqt into spv wallets?
2014-04-07 12:34 GMT+01:00 Mike Hearn <mike@plan99•net>:
Quoted text (29 lines)
> At the start of February we had 10,000 bitcoin nodes. Now we have 8,500 and > still falling: > > http://getaddr.bitnodes.io/dashboard/chart/?days=60 > > I know all the reasons why people might stop running a node (uses too much > disk space, bandwidth, lost interest etc). But does anyone have any idea how > we might get more insight into what's really going on? It'd be convenient if > the subVer contained the operating system, as then we could tell if the > bleed was mostly from desktops/laptops (Windows/Mac), which would be > expected, or from virtual servers (Linux), which would be more concerning. > > When you set up a Tor node, you can add your email address to the config > file and the Tor project sends you emails from time to time about things you > should know about. If we did the same, we could have a little exit survey: > if your node disappears for long enough, we could email the operator and ask > why they stopped. > > ------------------------------------------------------------------------------ > Put Bad Developers to Shame > Dominate Development with Jenkins Continuous Integration > Continuously Automate Build, Test & Deployment > Start a new project now. Try Jenkins in the cloud. > http://p.sf.net/sfu/13600_Cloudbees_APR > _______________________________________________ > Bitcoin-development mailing list > Bitcoin-development@lists•sourceforge•net > https://lists.sourceforge.net/lists/listinfo/bitcoin-development >
I'm glad to see that I'm not the only one concerned about the consistent dropping of nodes. Though I think that the fundamental question should be: how many nodes do we really need? Obviously more is better, but it's difficult to say how concerned we should be without more information. I posted my thoughts last month: http://coinchomp.com/2014/03/19/bitcoin-nodes-many-enough/
I have begun working on my node monitoring project and will post updates if it results in me gaining any new insights about the network.
- Jameson
On 04/07/2014 07:34 AM, Mike Hearn wrote:
Quoted text (35 lines)
> At the start of February we had 10,000 bitcoin nodes. Now we have 8,500 and > still falling: > > http://getaddr.bitnodes.io/dashboard/chart/?days=60 > > I know all the reasons why people *might* stop running a node (uses too > much disk space, bandwidth, lost interest etc). But does anyone have any > idea how we might get more insight into what's really going on? It'd be > convenient if the subVer contained the operating system, as then we could > tell if the bleed was mostly from desktops/laptops (Windows/Mac), which > would be expected, or from virtual servers (Linux), which would be more > concerning. > > When you set up a Tor node, you can add your email address to the config > file and the Tor project sends you emails from time to time about things > you should know about. If we did the same, we could have a little exit > survey: if your node disappears for long enough, we could email the > operator and ask why they stopped. > > > > ------------------------------------------------------------------------------ > Put Bad Developers to Shame > Dominate Development with Jenkins Continuous Integration > Continuously Automate Build, Test & Deployment > Start a new project now. Try Jenkins in the cloud. > http://p.sf.net/sfu/13600_Cloudbees_APR > > > > _______________________________________________ > Bitcoin-development mailing list > Bitcoin-development@lists•sourceforge•net > https://lists.sourceforge.net/lists/listinfo/bitcoin-development >
On Mon, Apr 7, 2014 at 2:19 PM, Jameson Lopp <jameson.lopp@gmail•com> wrote:
Quoted text (1 line)
> I'm glad to see that I'm not the only one concerned about the consistent dropping of nodes. Though I think that the fundamental question should be: how many nodes do we really need? Obviously more is better, but it's difficult to say how concerned we should be without more information. I posted my thoughts last month: http://coinchomp.com/2014/03/19/bitcoin-nodes-many-enough/
In my opinion, the number of full nodes doesn't matter (as long as it's enough to satisfy demand by other nodes).
What matters is how hard it is to run one. If someone is interesting in verifying that nobody is cheating on the network, can they, and can they without significant investment? Whether they actually will depends also no how interesting the currency and its digital transfers are.
Quoted text (5 lines)
> On 04/07/2014 07:34 AM, Mike Hearn wrote: >> At the start of February we had 10,000 bitcoin nodes. Now we have 8,500 and >> still falling: >> >> http://getaddr.bitnodes.io/dashboard/chart/?days=60
My own network crawler (which feeds my DNS seeder) hasn't seen any significant drop that I remember, but I don't have actual logs. It's seeing around 6000 "well reachable nodes" currently, which is the highest number I've ever seen (though it's been around 6000 for quite a while now).
Quoted text (4 lines)
> > In my opinion, the number of full nodes doesn't matter (as long as > it's enough to satisfy demand by other nodes). >
Correct. Still, a high number of nodes has a few other benefits:
1) The more nodes there are, the cheaper it should be to run each one, given that the bandwidth and CPU for serving the chain will be spread over more people.
2) It makes Bitcoin *seem* bigger, more robust and more decentralised, because there are more people uniting to run it. So there's a psychological benefit.
Also, we don't have a good way to measure capacity vs demand at the moment. Whether we have enough capacity is rather a shot in the dark right now.
Quoted text (2 lines)
> What matters is how hard it is to run one. >
Which is why I'm interested to learn the reason behind the drop. Is it insufficient interest, or is running a node too painful?
For this purpose I'd like to exclude people running Bitcoin Core on laptops or non-dedicated desktops. I don't think full nodes will ever make sense for consumer wallets again, and I see the bleeding off of those people as natural and expected (as Satoshi did). But if someone feels it's too hard to run on a cheap server then that'd concern me.
Quoted text (2 lines)
> My own network crawler (which feeds my DNS seeder) hasn't seen any > significant drop
It would be good to explain the difference, but I suspect your definition of "well reachable" excludes people running Core at home. From the diurnal cycle we see in Addy's graphs it's clear some nodes are being shut down when people go to bed. So if we have 6000 nodes on servers and 2000 at home, then I'd expect Addy's graphs and yours to slowly come into alignment as people give up using Core as a consumer wallet.
On 04/07/2014 08:26 AM, Pieter Wuille wrote:
Quoted text (2 lines)
> In my opinion, the number of full nodes doesn't matter (as long as > it's enough to satisfy demand by other nodes).
I agree, but if we don't quantify "demand" then we are practically blind. What is the plan? To wait until SPV clients start lagging / timing out because their requests cannot be handled by the nodes?
For all I know, the network would run just fine on 100 nodes. But not knowing really irks me as an engineer.
- Jameson
They're _not_ phasing out into SPV wallets from what I can say. At around the 24th of February, there has been a sharp change of the "current installs" graph of Bitcoin Wallet. That number used to grow at about 20.000 per month. After that date until now, it just barely moves horizontal.
My guess is that a large number of users have lost interest after they lost their money in MtGox. The 24th of February coincides with the "final" shutdown, according to
http://en.wikipedia.org/wiki/Mt._Gox#February_2014_shutdown_and_bankruptcy
On 04/07/2014 02:17 PM, Ricardo Filipe wrote:
Quoted text (40 lines)
> phasing out of bitcoinqt into spv wallets? > > 2014-04-07 12:34 GMT+01:00 Mike Hearn <mike@plan99•net>: >> At the start of February we had 10,000 bitcoin nodes. Now we have 8,500 and >> still falling: >> >> http://getaddr.bitnodes.io/dashboard/chart/?days=60 >> >> I know all the reasons why people might stop running a node (uses too much >> disk space, bandwidth, lost interest etc). But does anyone have any idea how >> we might get more insight into what's really going on? It'd be convenient if >> the subVer contained the operating system, as then we could tell if the >> bleed was mostly from desktops/laptops (Windows/Mac), which would be >> expected, or from virtual servers (Linux), which would be more concerning. >> >> When you set up a Tor node, you can add your email address to the config >> file and the Tor project sends you emails from time to time about things you >> should know about. If we did the same, we could have a little exit survey: >> if your node disappears for long enough, we could email the operator and ask >> why they stopped. >> >> ------------------------------------------------------------------------------ >> Put Bad Developers to Shame >> Dominate Development with Jenkins Continuous Integration >> Continuously Automate Build, Test & Deployment >> Start a new project now. Try Jenkins in the cloud. >> http://p.sf.net/sfu/13600_Cloudbees_APR >> _______________________________________________ >> Bitcoin-development mailing list >> Bitcoin-development@lists•sourceforge•net >> https://lists.sourceforge.net/lists/listinfo/bitcoin-development >> > > ------------------------------------------------------------------------------ > Put Bad Developers to Shame > Dominate Development with Jenkins Continuous Integration > Continuously Automate Build, Test & Deployment > Start a new project now. Try Jenkins in the cloud. > http://p.sf.net/sfu/13600_Cloudbees_APR >
On Mon, Apr 7, 2014 at 4:34 AM, Mike Hearn <mike@plan99•net> wrote:
Quoted text (2 lines)
> At the start of February we had 10,000 bitcoin nodes. Now we have 8,500 and > still falling:
FWIW, A few months before that we had even less than 8500 by the bitnodes count.
Bitcoin.org recommends people away from running Bitcoin-QT now, so I'm not sure that we should generally find that trend surprising.
On Mon, Apr 7, 2014 at 6:50 AM, Gregory Maxwell <gmaxwell@gmail•com> wrote:
Quoted text (1 line)
> FWIW, A few months before that we had even less than 8500 by the bitnodes count.
Gah, accidentally send.... I wanted to continue here that it was less than 8500 and had been falling pretty consistently for months, basically since the bitcoin.org change. Unfortunately it looks like the old bitnodes.io data isn't available anymore, so I'm going off my memory here.
The Bitnodes counts have always been somewhat higher than my or sipa's node counts too, fwiw.
The Bitnodes project updated their counting algorithm a month or so ago. It used to be slower and less accurate - prior to their update, it was reporting in excess of 100,000 nodes.
- Jameson
On 04/07/2014 09:53 AM, Gregory Maxwell wrote:
Quoted text (23 lines)
> On Mon, Apr 7, 2014 at 6:50 AM, Gregory Maxwell <gmaxwell@gmail•com> wrote: >> FWIW, A few months before that we had even less than 8500 by the bitnodes count. > > Gah, accidentally send.... I wanted to continue here that it was less > than 8500 and had been falling pretty consistently for months, > basically since the bitcoin.org change. Unfortunately it looks like > the old bitnodes.io data isn't available anymore, so I'm going off my > memory here. > > The Bitnodes counts have always been somewhat higher than my or sipa's > node counts too, fwiw. > > ------------------------------------------------------------------------------ > Put Bad Developers to Shame > Dominate Development with Jenkins Continuous Integration > Continuously Automate Build, Test & Deployment > Start a new project now. Try Jenkins in the cloud. > http://p.sf.net/sfu/13600_Cloudbees_APR > _______________________________________________ > Bitcoin-development mailing list > Bitcoin-development@lists•sourceforge•net > https://lists.sourceforge.net/lists/listinfo/bitcoin-development >
On Mon, Apr 7, 2014 at 6:58 AM, Jameson Lopp <jameson.lopp@gmail•com> wrote:
Quoted text (1 line)
> The Bitnodes project updated their counting algorithm a month or so ago. It used to be slower and less accurate - prior to their update, it was reporting in excess of 100,000 nodes.
Nah. It reported multiple metrics. The "100,000" number was an mostly useless metric that just counted the number of distinct addr messages floating around the network, which contains a lot of junk. They also reported an actual connectable node count previously and while they may have tweaked things here and there as far as I can tell it has been consistent with the numbers they are using in the headlines now.
Quoted text (4 lines)
> > My guess is that a large number of users have lost interest after they > lost their money in MtGox. The 24th of February coincides with the > "final" shutdown
Sigh. It would not be surprising if MtGox has indeed dealt the community a critical blow in this regard. TX traffic is down since then too:
Judging from comments and the leaked user db, it seems a lot of well known people lost money there (not me fortunately). I wish I could say people have learned but from the size of the deposit base at Bitstamp they clearly have not. A lot of Bitcoin users don't seem to be ready to be their own bank, yet still want to own some on the assumption everyone else either is or soon will be. So it's really only a matter of time until something goes wrong with some large bitbank again, either Bitstamp or Coinbase.
Some days I wonder if Bitcoin will be killed off by people who just refuse to use it properly before it ever gets a chance to shine. The general public doesn't distinguish between "Bitcoin users" who deposit with a third party and the real Bitcoin users who don't.
We need to make it so mind-numbingly simple to "run Bitcoin correctly" that the average user doesn't find reasons to do so in the course of normal use. Right now, Coinbase and Bitstamp are winning in the user experience battle, which technically endanger the user, and by proxy the Bitcoin network.
Multi-sig as a default is a start. It won't succeed unless the user experience is simply better than trusted third parties, but we need to start the education process with the very basic fundamental: trusting a third-party with full access to your Bitcoin is just replacing one centralized banking system with another.
Eric Martindale Developer Evangelist, BitPay +1 (919) 374-2020 On Apr 7, 2014 7:05 AM, "Mike Hearn" <mike@plan99•net> wrote:
Quoted text (37 lines)
> My guess is that a large number of users have lost interest after they >> lost their money in MtGox. The 24th of February coincides with the >> "final" shutdown > > > Sigh. It would not be surprising if MtGox has indeed dealt the community a > critical blow in this regard. TX traffic is down since then too: > > > https://blockchain.info/charts/n-transactions-excluding-popular?timespan=60days&showDataPoints=false&daysAverageString=1&show_header=true&scale=0&address= > > Judging from comments and the leaked user db, it seems a lot of well known > people lost money there (not me fortunately). I wish I could say people > have learned but from the size of the deposit base at Bitstamp they clearly > have not. A lot of Bitcoin users don't seem to be ready to be their own > bank, yet still want to own some on the assumption everyone else either is > or soon will be. So it's really only a matter of time until something goes > wrong with some large bitbank again, either Bitstamp or Coinbase. > > Some days I wonder if Bitcoin will be killed off by people who just refuse > to use it properly before it ever gets a chance to shine. The general > public doesn't distinguish between "Bitcoin users" who deposit with a third > party and the real Bitcoin users who don't. > > > ------------------------------------------------------------------------------ > Put Bad Developers to Shame > Dominate Development with Jenkins Continuous Integration > Continuously Automate Build, Test & Deployment > Start a new project now. Try Jenkins in the cloud. > http://p.sf.net/sfu/13600_Cloudbees_APR > _______________________________________________ > Bitcoin-development mailing list > Bitcoin-development@lists•sourceforge•net > https://lists.sourceforge.net/lists/listinfo/bitcoin-development > >
Indeed, fully agreed. The only way to really make progress here is to make the UX of being your own bank not only as good as trusting a third party, but better.
I've been encouraged by the rise of risk analysis services, but we need to integrate them into wallets more widely for them to have much impact. Otherwise people get to pick between a variety of wallets, none of which have *all* the features they want. And TREZOR is cool, albeit, something that's going to be for committed users only.
On Mon, Apr 7, 2014 at 4:15 PM, Eric Martindale <eric@ericmartindale•com>wrote:
Quoted text (54 lines)
> We need to make it so mind-numbingly simple to "run Bitcoin correctly" > that the average user doesn't find reasons to do so in the course of normal > use. Right now, Coinbase and Bitstamp are winning in the user experience > battle, which technically endanger the user, and by proxy the Bitcoin > network. > > Multi-sig as a default is a start. It won't succeed unless the user > experience is simply better than trusted third parties, but we need to > start the education process with the very basic fundamental: trusting a > third-party with full access to your Bitcoin is just replacing one > centralized banking system with another. > > Eric Martindale > Developer Evangelist, BitPay > +1 (919) 374-2020 > On Apr 7, 2014 7:05 AM, "Mike Hearn" <mike@plan99•net> wrote: > >> My guess is that a large number of users have lost interest after they >>> lost their money in MtGox. The 24th of February coincides with the >>> "final" shutdown >> >> >> Sigh. It would not be surprising if MtGox has indeed dealt the community >> a critical blow in this regard. TX traffic is down since then too: >> >> >> https://blockchain.info/charts/n-transactions-excluding-popular?timespan=60days&showDataPoints=false&daysAverageString=1&show_header=true&scale=0&address= >> >> Judging from comments and the leaked user db, it seems a lot of well >> known people lost money there (not me fortunately). I wish I could say >> people have learned but from the size of the deposit base at Bitstamp they >> clearly have not. A lot of Bitcoin users don't seem to be ready to be their >> own bank, yet still want to own some on the assumption everyone else either >> is or soon will be. So it's really only a matter of time until something >> goes wrong with some large bitbank again, either Bitstamp or Coinbase. >> >> Some days I wonder if Bitcoin will be killed off by people who just >> refuse to use it properly before it ever gets a chance to shine. The >> general public doesn't distinguish between "Bitcoin users" who deposit with >> a third party and the real Bitcoin users who don't. >> >> >> ------------------------------------------------------------------------------ >> Put Bad Developers to Shame >> Dominate Development with Jenkins Continuous Integration >> Continuously Automate Build, Test & Deployment >> Start a new project now. Try Jenkins in the cloud. >> http://p.sf.net/sfu/13600_Cloudbees_APR >> _______________________________________________ >> Bitcoin-development mailing list >> Bitcoin-development@lists•sourceforge•net >> https://lists.sourceforge.net/lists/listinfo/bitcoin-development >> >>
On 4/7/2014 7:05 AM, Mike Hearn wrote:
Quoted text (5 lines)
> Some days I wonder if Bitcoin will be killed off by people who just > refuse to use it properly before it ever gets a chance to shine. The > general public doesn't distinguish between "Bitcoin users" who deposit > with a third party and the real Bitcoin users who don't. >
A Mt-Gox-scale incident was inevitable and there are probably more hard lessons in the future. But it has made bitcoin much better already. I'm referring to things like malleability fixes, cleaning house at the foundation, and public education (hard as the lesson was). A lot more people than before understand the distinction you're making, and they are sharing the lesson.
On 04/07/2014 11:34 AM, Mike Hearn wrote:
Quoted text (12 lines)
> At the start of February we had 10,000 bitcoin nodes. Now we have 8,500 and > still falling: > > http://getaddr.bitnodes.io/dashboard/chart/?days=60 > > I know all the reasons why people *might* stop running a node (uses too > much disk space, bandwidth, lost interest etc). But does anyone have any > idea how we might get more insight into what's really going on? It'd be > convenient if the subVer contained the operating system, as then we could > tell if the bleed was mostly from desktops/laptops (Windows/Mac), which > would be expected, or from virtual servers (Linux), which would be more > concerning.
It doesn't do much good to only focus on the immediate symptoms at the exclusion of big picture trends. There are three things happening now that have nothing to do with operating systems.
1. The resource requirements of a full node are moving beyond the capabilities of casual users. This isn't inherently a problem - after all most people don't grow their own food, tailor their own clothes, or keep blacksmith tools handy in to forge their own horseshoes either.
2. The growth of small and medium-sized native Bitcoin businesses is lagging #1. Native here means their revenue and expenses are both denominated in BTC. Most business adoption we've seen so far doesn't actually handle bitcoins themselves. They use Bitcoin as a payment method whose processing they outsource.
Contributing to this is the fact that Bitcoin Core, although it has made great progress in the 0.9 release, can't be accurately described as "enterprise ready".
3. The P2P protocol used by the network is broken from an incentive perspective. Resource usage wouldn't be a problem as long as the users which consume resources pay for them and the users who provide resources are compensated, and they communicate via an efficient price discovery mechanism. Right now there is no obvious way to incorporate price discovery for bandwidth usage or storage space without a completely new P2P protocol, and the effectiveness with which progress has been blocked towards price discovery of transaction fees (the area where it is most obviously necessary) means that I'm not optimistic that this subject will ever be effectively addressed by Bitcoin Core.
- -- Support online privacy by using email encryption whenever possible. Learn how here: http://www.youtube.com/watch?v=bakOKJFtB-k
On Mon, Apr 7, 2014 at 8:45 AM, Justus Ranvier <justusranvier@gmail•com> wrote:
Quoted text (4 lines)
> 1. The resource requirements of a full node are moving beyond the > capabilities of casual users. This isn't inherently a problem - after > all most people don't grow their own food, tailor their own clothes, or > keep blacksmith tools handy in to forge their own horseshoes either.
Right now running a full node consumes about $1 in disk space non-reoccurring and costs a couple cents in power per month.
This isn't to say things are all ducky. But if you're going to say the resource requirements are beyond the capabilities of casual users I'm afraid I'm going to have to say: citation needed.
I would point to bandwidth as the most important issue to the casual user who runs a node at home. Few casual users have the know-how to set up QoS rules and thus become quite annoyed when their Internet connection is discernibly slowed.
- Jameson
On 04/07/2014 11:53 AM, Gregory Maxwell wrote:
Quoted text (24 lines)
> On Mon, Apr 7, 2014 at 8:45 AM, Justus Ranvier <justusranvier@gmail•com> wrote: >> 1. The resource requirements of a full node are moving beyond the >> capabilities of casual users. This isn't inherently a problem - after >> all most people don't grow their own food, tailor their own clothes, or >> keep blacksmith tools handy in to forge their own horseshoes either. > > Right now running a full node consumes about $1 in disk space > non-reoccurring and costs a couple cents in power per month. > > This isn't to say things are all ducky. But if you're going to say the > resource requirements are beyond the capabilities of casual users I'm > afraid I'm going to have to say: citation needed. > > ------------------------------------------------------------------------------ > Put Bad Developers to Shame > Dominate Development with Jenkins Continuous Integration > Continuously Automate Build, Test & Deployment > Start a new project now. Try Jenkins in the cloud. > http://p.sf.net/sfu/13600_Cloudbees_APR > _______________________________________________ > Bitcoin-development mailing list > Bitcoin-development@lists•sourceforge•net > https://lists.sourceforge.net/lists/listinfo/bitcoin-development >
Right now running a full-node on my home DSL connection (<1Mbps) makes other internet activity periodically unresponsive. I think we've already hit a point where resource requirements are pushing out casual users, although of course we can't be certain that accounts for all lost nodes.
On 04/07/2014 08:53 AM, Gregory Maxwell wrote:
Quoted text (25 lines)
> On Mon, Apr 7, 2014 at 8:45 AM, Justus Ranvier <justusranvier@gmail•com> wrote: >> 1. The resource requirements of a full node are moving beyond the >> capabilities of casual users. This isn't inherently a problem - after >> all most people don't grow their own food, tailor their own clothes, or >> keep blacksmith tools handy in to forge their own horseshoes either. > > Right now running a full node consumes about $1 in disk space > non-reoccurring and costs a couple cents in power per month. > > This isn't to say things are all ducky. But if you're going to say the > resource requirements are beyond the capabilities of casual users I'm > afraid I'm going to have to say: citation needed. > > ------------------------------------------------------------------------------ > Put Bad Developers to Shame > Dominate Development with Jenkins Continuous Integration > Continuously Automate Build, Test & Deployment > Start a new project now. Try Jenkins in the cloud. > http://p.sf.net/sfu/13600_Cloudbees_APR > _______________________________________________ > Bitcoin-development mailing list > Bitcoin-development@lists•sourceforge•net > https://lists.sourceforge.net/lists/listinfo/bitcoin-development > . >
On Mon, Apr 7, 2014 at 9:27 AM, Mark Friedenbach <mark@monetize•io> wrote:
Quoted text (4 lines)
> Right now running a full-node on my home DSL connection (<1Mbps) makes > other internet activity periodically unresponsive. I think we've already > hit a point where resource requirements are pushing out casual users, > although of course we can't be certain that accounts for all lost nodes.
That is an implementation issue— mostly one that arises as an indirect consequence of not having headers first and the parallel fetch, not a requirements issue.
Under the current bitcoin validity rules it should be completely reasonable to run a full contributing node with no more than 30 kb/s inbound (reviving two copies of everything, blocks + tansactions ) and 60 kbit/sec outbound (sending out four copies of everything). (So long as you're sending out >= what you're taking in you're contributing to the network's capacity). Throw in a factor of two for bursting, though not every node needs to be contributing super low latency capacity.
This is absolutely not the case with the current implementation, but it's not a requirements thing.
On 04/07/2014 09:57 AM, Gregory Maxwell wrote:
Quoted text (3 lines)
> That is an implementation issue— mostly one that arises as an indirect > consequence of not having headers first and the parallel fetch, not a > requirements issue.
Oh, absolutely. But the question "why are people not running full nodes?" has to do with the current implementation, not abstract capabilities of a future version of the bitcoind code base.
For what it's worth, the number of nodes rose dramatically during the China bullrun (I recall 45k in China alone) and dropped as dramatically as the price after the first PBOC announcement designed to cool down bitcoin trading in China.
On 7 April 2014 12:34, Mike Hearn <mike@plan99•net> wrote:
Quoted text (32 lines)
> At the start of February we had 10,000 bitcoin nodes. Now we have 8,500 > and still falling: > > http://getaddr.bitnodes.io/dashboard/chart/?days=60 > > I know all the reasons why people *might* stop running a node (uses too > much disk space, bandwidth, lost interest etc). But does anyone have any > idea how we might get more insight into what's really going on? It'd be > convenient if the subVer contained the operating system, as then we could > tell if the bleed was mostly from desktops/laptops (Windows/Mac), which > would be expected, or from virtual servers (Linux), which would be more > concerning. > > When you set up a Tor node, you can add your email address to the config > file and the Tor project sends you emails from time to time about things > you should know about. If we did the same, we could have a little exit > survey: if your node disappears for long enough, we could email the > operator and ask why they stopped. > > > ------------------------------------------------------------------------------ > Put Bad Developers to Shame > Dominate Development with Jenkins Continuous Integration > Continuously Automate Build, Test & Deployment > Start a new project now. Try Jenkins in the cloud. > http://p.sf.net/sfu/13600_Cloudbees_APR > _______________________________________________ > Bitcoin-development mailing list > Bitcoin-development@lists•sourceforge•net > https://lists.sourceforge.net/lists/listinfo/bitcoin-development > >
On Mon, Apr 7, 2014 at 10:01 AM, Mark Friedenbach <mark@monetize•io> wrote:
Quoted text (8 lines)
> On 04/07/2014 09:57 AM, Gregory Maxwell wrote: >> That is an implementation issue— mostly one that arises as an indirect >> consequence of not having headers first and the parallel fetch, not a >> requirements issue. > > Oh, absolutely. But the question "why are people not running full > nodes?" has to do with the current implementation, not abstract > capabilities of a future version of the bitcoind code base.
The distinction is very important because it's a matter of things we can and should fix vs things that cannot be fixed except by changing goals/incentives! Opposite approaches to handling them.
When I read "resource requirements of a full node are moving beyond" I didn't extract from that that "there are implementation issues that need to be improved to make it work better for low resource users" due to the word "requirements".
How difficult would it be to set up a node? Using lots of electricity at home (if required) could be an issue, but I do have a Webfaction account.
On Mon, Apr 7, 2014 at 12:16 PM, Gregory Maxwell <gmaxwell@gmail•com> wrote:
Quoted text (32 lines)
> On Mon, Apr 7, 2014 at 10:01 AM, Mark Friedenbach <mark@monetize•io> > wrote: > > On 04/07/2014 09:57 AM, Gregory Maxwell wrote: > >> That is an implementation issue-- mostly one that arises as an indirect > >> consequence of not having headers first and the parallel fetch, not a > >> requirements issue. > > > > Oh, absolutely. But the question "why are people not running full > > nodes?" has to do with the current implementation, not abstract > > capabilities of a future version of the bitcoind code base. > > The distinction is very important because it's a matter of things we > can and should fix vs things that cannot be fixed except by changing > goals/incentives! Opposite approaches to handling them. > > When I read "resource requirements of a full node are moving beyond" I > didn't extract from that that "there are implementation issues that > need to be improved to make it work better for low resource users" due > to the word "requirements". > > > ------------------------------------------------------------------------------ > Put Bad Developers to Shame > Dominate Development with Jenkins Continuous Integration > Continuously Automate Build, Test & Deployment > Start a new project now. Try Jenkins in the cloud. > http://p.sf.net/sfu/13600_Cloudbees > _______________________________________________ > Bitcoin-development mailing list > Bitcoin-development@lists•sourceforge•net > https://lists.sourceforge.net/lists/listinfo/bitcoin-development >
I’m afraid this is a highly simplistic view of the costs of running a full node.
My node consumes fantastic amounts of data traffic, which is a real cost.
In the 30 days ending Apri 6, my node:
* Received 36.8 gb of data * Sent 456.5 gb data
At my geographic service location (Singapore), this cost about $90 last month for bandwidth alone. It would be slightly cheaper if I was hosted in the US of course.
But anyone can understand that moving a half-terrabyte of data around in a month will not be cheap.
On Apr 7, 2014, at 8:53 AM, Gregory Maxwell <gmaxwell@gmail•com> wrote:
Quoted text (23 lines)
> On Mon, Apr 7, 2014 at 8:45 AM, Justus Ranvier <justusranvier@gmail•com> wrote: >> 1. The resource requirements of a full node are moving beyond the >> capabilities of casual users. This isn't inherently a problem - after >> all most people don't grow their own food, tailor their own clothes, or >> keep blacksmith tools handy in to forge their own horseshoes either. > > Right now running a full node consumes about $1 in disk space > non-reoccurring and costs a couple cents in power per month. > > This isn't to say things are all ducky. But if you're going to say the > resource requirements are beyond the capabilities of casual users I'm > afraid I'm going to have to say: citation needed. > > ------------------------------------------------------------------------------ > Put Bad Developers to Shame > Dominate Development with Jenkins Continuous Integration > Continuously Automate Build, Test & Deployment > Start a new project now. Try Jenkins in the cloud. > http://p.sf.net/sfu/13600_Cloudbees_APR > _______________________________________________ > Bitcoin-development mailing list > Bitcoin-development@lists•sourceforge•net > https://lists.sourceforge.net/lists/listinfo/bitcoin-development
It uses ~no electricity, it's not like mining.
The primary resources it needs are disk space and bandwidth, after an intensive initial day or two of building the database.
Actually, I wonder if we should start shipping (auditable) pre-baked databases calculated up to the last checkpoint so people can download them and boot up their node right away. Recalculating the entire thing from scratch every time isn't sustainable in the long run anyway.
On Mon, Apr 7, 2014 at 7:35 PM, Brent Shambaugh <brent.shambaugh@gmail•com>wrote:
Quoted text (53 lines)
> How difficult would it be to set up a node? Using lots of electricity at > home (if required) could be an issue, but I do have a Webfaction account. > > > On Mon, Apr 7, 2014 at 12:16 PM, Gregory Maxwell <gmaxwell@gmail•com>wrote: > >> On Mon, Apr 7, 2014 at 10:01 AM, Mark Friedenbach <mark@monetize•io> >> wrote: >> > On 04/07/2014 09:57 AM, Gregory Maxwell wrote: >> >> That is an implementation issue— mostly one that arises as an indirect >> >> consequence of not having headers first and the parallel fetch, not a >> >> requirements issue. >> > >> > Oh, absolutely. But the question "why are people not running full >> > nodes?" has to do with the current implementation, not abstract >> > capabilities of a future version of the bitcoind code base. >> >> The distinction is very important because it's a matter of things we >> can and should fix vs things that cannot be fixed except by changing >> goals/incentives! Opposite approaches to handling them. >> >> When I read "resource requirements of a full node are moving beyond" I >> didn't extract from that that "there are implementation issues that >> need to be improved to make it work better for low resource users" due >> to the word "requirements". >> >> >> ------------------------------------------------------------------------------ >> Put Bad Developers to Shame >> Dominate Development with Jenkins Continuous Integration >> Continuously Automate Build, Test & Deployment >> Start a new project now. Try Jenkins in the cloud. >> http://p.sf.net/sfu/13600_Cloudbees >> _______________________________________________ >> Bitcoin-development mailing list >> Bitcoin-development@lists•sourceforge•net >> https://lists.sourceforge.net/lists/listinfo/bitcoin-development >> > > > > ------------------------------------------------------------------------------ > Put Bad Developers to Shame > Dominate Development with Jenkins Continuous Integration > Continuously Automate Build, Test & Deployment > Start a new project now. Try Jenkins in the cloud. > http://p.sf.net/sfu/13600_Cloudbees > _______________________________________________ > Bitcoin-development mailing list > Bitcoin-development@lists•sourceforge•net > https://lists.sourceforge.net/lists/listinfo/bitcoin-development > >
On Mon, Apr 7, 2014 at 10:40 AM, Mike Hearn <mike@plan99•net> wrote:
Quoted text (1 line)
> Actually, I wonder
The actual validation isn't really the problem today. The slowness of the IBD is almost exclusively the lack of parallel fetching and the existence of slow peers. And making the signature gate adaptive (and deploying the 6x faster ECDSA code) would improve that future.
Go grab sipa's headers first branch, it has no problem saturating a 20mbit/sec pipe syncing up.
I rather prefer to start with SPV and upgrade to full node, if desired.
Tamas Blummer http://bitsofproof.com
On 07.04.2014, at 19:40, Mike Hearn <mike@plan99•net> wrote:
Quoted text (3 lines)
> > Actually, I wonder if we should start shipping (auditable) pre-baked databases calculated up to the last checkpoint so people can download them and boot up their node right away. Recalculating the entire thing from scratch every time isn't sustainable in the long run anyway. >
On 04/07/2014 05:16 PM, Gregory Maxwell wrote:
Quoted text (4 lines)
> When I read "resource requirements of a full node are moving > beyond" I didn't extract from that that "there are implementation > issues that need to be improved to make it work better for low > resource users" due to the word "requirements".
In order to prevent future confusion: whenever I talk about requirements (or generally use the present tense) , I'm talking about reality as it currently exists.
If I ever decide to talk about hypothetical future requirements in some imaginary world of Platonic forms, as opposed to the requirements imposed by the software that's actually available for casual users to download today, I'll mention that specifically.
- -- Support online privacy by using email encryption whenever possible. Learn how here: http://www.youtube.com/watch?v=bakOKJFtB-k
On 04/07/2014 05:40 PM, Mike Hearn wrote:
Quoted text (2 lines)
> The primary resources it needs are disk space and bandwidth, after > an intensive initial day or two of building the database.
Check out the kind of hardware causal users are running these days.
The bottleneck is not bulk disk space, but rather IOPS.
Most users don't have spare machines to dedicate to the task of running a full node, nor is it acceptable for them to not be able to use their device for other tasks while the node is bootstrapping.
- -- Support online privacy by using email encryption whenever possible. Learn how here: http://www.youtube.com/watch?v=bakOKJFtB-k
Okay awesome. It seems like I set up a Litecoin node without knowing it (because it was like this: https://bitcointalk.org/index.php?topic=128122.0) I was able to bootstrap it (https://litecoin.info/).
On Mon, Apr 7, 2014 at 12:40 PM, Mike Hearn <mike@plan99•net> wrote:
Quoted text (68 lines)
> It uses ~no electricity, it's not like mining. > > The primary resources it needs are disk space and bandwidth, after an > intensive initial day or two of building the database. > > Actually, I wonder if we should start shipping (auditable) pre-baked > databases calculated up to the last checkpoint so people can download them > and boot up their node right away. Recalculating the entire thing from > scratch every time isn't sustainable in the long run anyway. > > > On Mon, Apr 7, 2014 at 7:35 PM, Brent Shambaugh <brent.shambaugh@gmail•com > > wrote: > >> How difficult would it be to set up a node? Using lots of electricity at >> home (if required) could be an issue, but I do have a Webfaction account. >> >> >> On Mon, Apr 7, 2014 at 12:16 PM, Gregory Maxwell <gmaxwell@gmail•com>wrote: >> >>> On Mon, Apr 7, 2014 at 10:01 AM, Mark Friedenbach <mark@monetize•io> >>> wrote: >>> > On 04/07/2014 09:57 AM, Gregory Maxwell wrote: >>> >> That is an implementation issue-- mostly one that arises as an indirect >>> >> consequence of not having headers first and the parallel fetch, not a >>> >> requirements issue. >>> > >>> > Oh, absolutely. But the question "why are people not running full >>> > nodes?" has to do with the current implementation, not abstract >>> > capabilities of a future version of the bitcoind code base. >>> >>> The distinction is very important because it's a matter of things we >>> can and should fix vs things that cannot be fixed except by changing >>> goals/incentives! Opposite approaches to handling them. >>> >>> When I read "resource requirements of a full node are moving beyond" I >>> didn't extract from that that "there are implementation issues that >>> need to be improved to make it work better for low resource users" due >>> to the word "requirements". >>> >>> >>> ------------------------------------------------------------------------------ >>> Put Bad Developers to Shame >>> Dominate Development with Jenkins Continuous Integration >>> Continuously Automate Build, Test & Deployment >>> Start a new project now. Try Jenkins in the cloud. >>> http://p.sf.net/sfu/13600_Cloudbees >>> _______________________________________________ >>> Bitcoin-development mailing list >>> Bitcoin-development@lists•sourceforge•net >>> https://lists.sourceforge.net/lists/listinfo/bitcoin-development >>> >> >> >> >> ------------------------------------------------------------------------------ >> Put Bad Developers to Shame >> Dominate Development with Jenkins Continuous Integration >> Continuously Automate Build, Test & Deployment >> Start a new project now. Try Jenkins in the cloud. >> http://p.sf.net/sfu/13600_Cloudbees >> _______________________________________________ >> Bitcoin-development mailing list >> Bitcoin-development@lists•sourceforge•net >> https://lists.sourceforge.net/lists/listinfo/bitcoin-development >> >> >
Quoted text (5 lines)
> > * Sent 456.5 gb data > > At my geographic service location (Singapore), this cost about $90 last > month for bandwidth alone.
One of the reasons I initiated the (now stalled) PayFile project was in anticipation of this problem:
https://github.com/mikehearn/PayFile http://www.youtube.com/watch?v=r0BXnWlnIi4&feature=youtu.be
At some point if you want to actually download and validate the full block chain from scratch, you will have to start paying for it I'm sure.
In the meantime:
1. Getting headers-first implemented and rolled out everywhere would reduce the amount of redundant downloading and hopefully reduce transmit traffic network-wide. 2. Implementing chain pruning would allow people to control upload bandwidth consumption by reducing the amount of disk storage they allow.
Quoted text (1 line)
> The bottleneck is not bulk disk space, but rather IOPS.
Exactly. I stopped running a full node on both of my desktops machines in the last month. Both systems were simply becoming very noticeable (=unbearably) sluggish. I am also running dedicated nodes, which are fine, but on a desktop latency is a major issue (or, let's say, a different issue than on a server).
I didn't notice any network latency, but that's hard to judge as our internet is pants anyway.
Arne
PS: my machines aren't the newest, but still do a stellar job (without bitcoind running ;)
Headers first loading allows the node to run SPV from the very first minutes and it can converge to full node by time. This is BTW how newest versions of BOP can work.
Pruning however disqualifies the node as a source for bootstrapping an other full node. BTW, did we already agree on the service bits for an archive node?
Tamas Blummer http://bitsofproof.com
On 07.04.2014, at 20:23, Mike Hearn <mike@plan99•net> wrote:
Quoted text (24 lines)
> * Sent 456.5 gb data > > At my geographic service location (Singapore), this cost about $90 last month for bandwidth alone. > > One of the reasons I initiated the (now stalled) PayFile project was in anticipation of this problem: > > https://github.com/mikehearn/PayFile > http://www.youtube.com/watch?v=r0BXnWlnIi4&feature=youtu.be > > At some point if you want to actually download and validate the full block chain from scratch, you will have to start paying for it I'm sure. > > In the meantime: > Getting headers-first implemented and rolled out everywhere would reduce the amount of redundant downloading and hopefully reduce transmit traffic network-wide. > Implementing chain pruning would allow people to control upload bandwidth consumption by reducing the amount of disk storage they allow. > > ------------------------------------------------------------------------------ > Put Bad Developers to Shame > Dominate Development with Jenkins Continuous Integration > Continuously Automate Build, Test & Deployment > Start a new project now. Try Jenkins in the cloud. > http://p.sf.net/sfu/13600_Cloudbees_______________________________________________ > Bitcoin-development mailing list > Bitcoin-development@lists•sourceforge•net > https://lists.sourceforge.net/lists/listinfo/bitcoin-development
On 04/07/2014 12:00 PM, Tamas Blummer wrote:
Quoted text (6 lines)
> Once a single transaction in pruned in a block, the block is no longer > eligible to be served to other nodes. > Which transactions are pruned can be rather custom e.g. even depending > on the wallet(s) of the node, > therefore I guess it is more handy to return some bitmap of pruned/full > blocks than ranges.
The point is that the node has decided not to prune transactions from that block, so that it is capable of returning full blocks within that range.
On Mon, Apr 7, 2014 at 11:35 AM, Tamas Blummer <tamas@bitsofproof•com> wrote:
Quoted text (1 line)
> BTW, did we already agree on the service bits for an archive node?
I'm still very concerned that a binary archive bit will cause extreme load hot-spotting and the kind of binary "Use lots of resources YES or NO" I think we're currently suffering some from, but at that point enshrined in the protocol.
It would be much better to extend the addr messages so that nodes can indicate a range or two of blocks that they're serving, so that all nodes can contribute fractionally according to their means. E.g. if you want to offer up 8 GB of distributed storage and contribute to the availability of the blockchain, without having to swollow the whole 20, 30, 40 ... gigabyte pill.
Already we need that kind of distributed storage for the most recent blocks to prevent extreme bandwidth load on archives, so extending it to arbitrary ranges is only more complicated because there is currently no room to signal it.
Once a single transaction in pruned in a block, the block is no longer eligible to be served to other nodes. Which transactions are pruned can be rather custom e.g. even depending on the wallet(s) of the node, therefore I guess it is more handy to return some bitmap of pruned/full blocks than ranges.
Tamas Blummer http://bitsofproof.com
On 07.04.2014, at 20:49, Gregory Maxwell <gmaxwell@gmail•com> wrote:
Quoted text (20 lines)
> On Mon, Apr 7, 2014 at 11:35 AM, Tamas Blummer <tamas@bitsofproof•com> wrote: >> BTW, did we already agree on the service bits for an archive node? > > I'm still very concerned that a binary archive bit will cause extreme > load hot-spotting and the kind of binary "Use lots of resources YES or > NO" I think we're currently suffering some from, but at that point > enshrined in the protocol. > > It would be much better to extend the addr messages so that nodes can > indicate a range or two of blocks that they're serving, so that all > nodes can contribute fractionally according to their means. E.g. if > you want to offer up 8 GB of distributed storage and contribute to the > availability of the blockchain, without having to swollow the whole > 20, 30, 40 ... gigabyte pill. > > Already we need that kind of distributed storage for the most recent > blocks to prevent extreme bandwidth load on archives, so extending it > to arbitrary ranges is only more complicated because there is > currently no room to signal it. >
On Mon, Apr 7, 2014 at 12:00 PM, Tamas Blummer <tamas@bitsofproof•com> wrote:
Quoted text (6 lines)
> Once a single transaction in pruned in a block, the block is no longer > eligible to be served to other nodes. > Which transactions are pruned can be rather custom e.g. even depending on > the wallet(s) of the node, > therefore I guess it is more handy to return some bitmap of pruned/full > blocks than ranges.
This isn't at all how pruning works in Bitcoin-QT (nor is it how I expect pruning to work for any mature implementation). Pruning can work happily on a whole block at a time basis regardless if all the transactions in it are spent or not.
On Mon, Apr 7, 2014 at 12:00 PM, Tamas Blummer <tamas@bitsofproof•com> wrote:
Quoted text (2 lines)
> therefore I guess it is more handy to return some bitmap of pruned/full > blocks than ranges.
A bitmap also means high overhead and— if it's used to advertise non-contiguous blocks— poor locality, since blocks are fetched sequentially.
Maybe it is not a question of the maturity of the implementation but that of the person making presumptions of it.
I consider a fully pruned blockchain being equivalent to the UTXO. Block that hold no more unspent transaction are reduced to a header. There is however no harm if more retained.
Tamas Blummer http://bitsofproof.com
On 07.04.2014, at 21:02, Gregory Maxwell <gmaxwell@gmail•com> wrote:
Quoted text (13 lines)
> On Mon, Apr 7, 2014 at 12:00 PM, Tamas Blummer <tamas@bitsofproof•com> wrote: >> Once a single transaction in pruned in a block, the block is no longer >> eligible to be served to other nodes. >> Which transactions are pruned can be rather custom e.g. even depending on >> the wallet(s) of the node, >> therefore I guess it is more handy to return some bitmap of pruned/full >> blocks than ranges. > > This isn't at all how pruning works in Bitcoin-QT (nor is it how I > expect pruning to work for any mature implementation). Pruning can > work happily on a whole block at a time basis regardless if all the > transactions in it are spent or not. >
On 04/07/2014 12:20 PM, Tamas Blummer wrote:
Quoted text (2 lines)
> Validation has to be sequantial, but that step can be deferred until the > blocks before a point are loaded and continous.
And how do you find those blocks?
I have a suggestion: have nodes advertise which range of full blocks they possess, then you can perform synchronization from the adversed ranges!
On Mon, Apr 7, 2014 at 8:03 PM, Gregory Maxwell <gmaxwell@gmail•com> wrote:
Quoted text (4 lines)
> A bitmap also means high overhead and-- if it's used to advertise > non-contiguous blocks-- poor locality, since blocks are fetched > sequentially. >
A range seems like a great compromise. Putting it in the address is also a pretty cool.
If light nodes selected a random contiguous 1GB of the block-chain, then they could handle most of the download overhead, rather than the full nodes.
Another way to do it would be to have something like a routing table. If a node is queried for a block, it can reply with the IP of a node with that block instead of sending the block.
One problem is that it means that light nodes have to accept incoming connections. Otherwise, it would have to be routed through the network.
Once headers are loaded first there is no reason for sequential loading.
Validation has to be sequantial, but that step can be deferred until the blocks before a point are loaded and continous.
Tamas Blummer http://bitsofproof.com
On 07.04.2014, at 21:03, Gregory Maxwell <gmaxwell@gmail•com> wrote:
Quoted text (8 lines)
> On Mon, Apr 7, 2014 at 12:00 PM, Tamas Blummer <tamas@bitsofproof•com> wrote: >> therefore I guess it is more handy to return some bitmap of pruned/full >> blocks than ranges. > > A bitmap also means high overhead and— if it's used to advertise > non-contiguous blocks— poor locality, since blocks are fetched > sequentially. >
I hope I'm not thread-jacking here, apologies if so, but that's the approach I've taken with the node I'm working on. Headers can be downloaded and stored in any order, it'll make sense of what the winning chain is. Blocks don't need to be downloaded in any particular order and they don't need to be saved to disk, the UTXO is fully self-contained. That way the concern of storing blocks for seeding (or not) is wholly separated from syncing the UTXO. This allows me to do the initial blockchain sync in ~6 hours when I use my SSD. I only need enough disk space to store the UTXO, and then whatever amount of block data the user would want to store for the health of the network. This project is a bitcoin learning exercise for me, so I can only hope I don't have any critical design flaws in there. :)
From: tamas@bitsofproof•com Date: Mon, 7 Apr 2014 21:20:31 +0200 To: gmaxwell@gmail•com CC: bitcoin-development@lists•sourceforge•net Subject: Re: [Bitcoin-development] Why are we bleeding nodes?
Once headers are loaded first there is no reason for sequential loading. Validation has to be sequantial, but that step can be deferred until the blocks before a point are loaded and continous. Tamas Blummerhttp://bitsofproof.com
On 07.04.2014, at 21:03, Gregory Maxwell <gmaxwell@gmail•com> wrote:On Mon, Apr 7, 2014 at 12:00 PM, Tamas Blummer <tamas@bitsofproof•com> wrote: therefore I guess it is more handy to return some bitmap of pruned/full blocks than ranges.
A bitmap also means high overhead and— if it's used to advertise non-contiguous blocks— poor locality, since blocks are fetched sequentially.
You have the trunk defined by the headers. Once a range from genesis to block n is fully downloaded, you may validate upto block n. Furthermore after validation you can prune transactions spent until block n.
You would approach the highest block with validation and stop pruning say 100 blocks before it, to leave room for reorgs.
Tamas Blummer http://bitsofproof.com
On 07.04.2014, at 21:13, Mark Friedenbach <mark@monetize•io> wrote:
Quoted text (22 lines)
> > > On 04/07/2014 12:20 PM, Tamas Blummer wrote: >> Validation has to be sequantial, but that step can be deferred until the >> blocks before a point are loaded and continous. > > And how do you find those blocks? > > I have a suggestion: have nodes advertise which range of full blocks > they possess, then you can perform synchronization from the adversed ranges! > > ------------------------------------------------------------------------------ > Put Bad Developers to Shame > Dominate Development with Jenkins Continuous Integration > Continuously Automate Build, Test & Deployment > Start a new project now. Try Jenkins in the cloud. > http://p.sf.net/sfu/13600_Cloudbees > _______________________________________________ > Bitcoin-development mailing list > Bitcoin-development@lists•sourceforge•net > https://lists.sourceforge.net/lists/listinfo/bitcoin-development >
I understand the theoretical benefits of multi-sig. But if you want to make this mind-numbingly simple, do it on the *existing* single-sig.
But why in the world do we not have a *business* that offers bitcoin wallet insurance? The bitcoin world (and this list) ran around blaming MtGox and users for being 'stupid' to trust mtgox.
So start a multi-level marketing business that offers *insurance* so if your bitcoin wallet gets hacked/stolen/whatever, your 'upstream' or whomever sold you the wallet comes to your house with a new computer or installs the new wallet software, or whatever, or just makes it good.
Now, if the **insurance underwriter** decides that multisig will reduce fraud, and **tests it**, then I'd say we do multi-sig. But right now we are just a bunch of technology wizards trying to force our own opinions about what's right and 'simple' for end users without ever asking the damn end-users.
And then we call the end-users idiots because some scammer calls them and says "I'm calling from Microsoft and your computer is broke, please download this software to fix it".
Multi-sig is more magical moon-math that scammers will exploit to con your grandma out of bitcoin, and then your friends will call her a stupid luddite for falling for it.
Fix the cultural victim-blaming bullshit and you'll fix the node bleeding problem.
On Mon, Apr 07, 2014 at 10:15:15AM -0400, Eric Martindale wrote:
Quoted text (64 lines)
> We need to make it so mind-numbingly simple to "run Bitcoin correctly" that > the average user doesn't find reasons to do so in the course of normal > use. Right now, Coinbase and Bitstamp are winning in the user experience > battle, which technically endanger the user, and by proxy the Bitcoin > network. > > Multi-sig as a default is a start. It won't succeed unless the user > experience is simply better than trusted third parties, but we need to > start the education process with the very basic fundamental: trusting a > third-party with full access to your Bitcoin is just replacing one > centralized banking system with another. > > Eric Martindale > Developer Evangelist, BitPay > +1 (919) 374-2020 > On Apr 7, 2014 7:05 AM, "Mike Hearn" <mike@plan99•net> wrote: > > > My guess is that a large number of users have lost interest after they > >> lost their money in MtGox. The 24th of February coincides with the > >> "final" shutdown > > > > > > Sigh. It would not be surprising if MtGox has indeed dealt the community a > > critical blow in this regard. TX traffic is down since then too: > > > > > > https://blockchain.info/charts/n-transactions-excluding-popular?timespan=60days&showDataPoints=false&daysAverageString=1&show_header=true&scale=0&address= > > > > Judging from comments and the leaked user db, it seems a lot of well known > > people lost money there (not me fortunately). I wish I could say people > > have learned but from the size of the deposit base at Bitstamp they clearly > > have not. A lot of Bitcoin users don't seem to be ready to be their own > > bank, yet still want to own some on the assumption everyone else either is > > or soon will be. So it's really only a matter of time until something goes > > wrong with some large bitbank again, either Bitstamp or Coinbase. > > > > Some days I wonder if Bitcoin will be killed off by people who just refuse > > to use it properly before it ever gets a chance to shine. The general > > public doesn't distinguish between "Bitcoin users" who deposit with a third > > party and the real Bitcoin users who don't. > > > > > > ------------------------------------------------------------------------------ > > Put Bad Developers to Shame > > Dominate Development with Jenkins Continuous Integration > > Continuously Automate Build, Test & Deployment > > Start a new project now. Try Jenkins in the cloud. > > http://p.sf.net/sfu/13600_Cloudbees_APR > > _______________________________________________ > > Bitcoin-development mailing list > > Bitcoin-development@lists•sourceforge•net > > https://lists.sourceforge.net/lists/listinfo/bitcoin-development > > > > > ------------------------------------------------------------------------------ > Put Bad Developers to Shame > Dominate Development with Jenkins Continuous Integration > Continuously Automate Build, Test & Deployment > Start a new project now. Try Jenkins in the cloud. > http://p.sf.net/sfu/13600_Cloudbees_APR > _______________________________________________ > Bitcoin-development mailing list > Bitcoin-development@lists•sourceforge•net > https://lists.sourceforge.net/lists/listinfo/bitcoin-development
You have to load headers sequantially to be able to connect them and determine the longest chain.
Blocks can be loaded in random order once you have their order given by the headers. Computing the UTXO however will force you to at least temporarily store the blocks unless you have plenty of RAM.
Regards,
Tamas Blummer http://bitsofproof.com
On 07.04.2014, at 21:30, Paul Lyon <pmlyon@hotmail•ca> wrote:
Quoted text (34 lines)
> I hope I'm not thread-jacking here, apologies if so, but that's the approach I've taken with the node I'm working on. > > Headers can be downloaded and stored in any order, it'll make sense of what the winning chain is. Blocks don't need to be downloaded in any particular order and they don't need to be saved to disk, the UTXO is fully self-contained. That way the concern of storing blocks for seeding (or not) is wholly separated from syncing the UTXO. This allows me to do the initial blockchain sync in ~6 hours when I use my SSD. I only need enough disk space to store the UTXO, and then whatever amount of block data the user would want to store for the health of the network. > > This project is a bitcoin learning exercise for me, so I can only hope I don't have any critical design flaws in there. :) > > From: tamas@bitsofproof•com > Date: Mon, 7 Apr 2014 21:20:31 +0200 > To: gmaxwell@gmail•com > CC: bitcoin-development@lists•sourceforge•net > Subject: Re: [Bitcoin-development] Why are we bleeding nodes? > > > Once headers are loaded first there is no reason for sequential loading. > > Validation has to be sequantial, but that step can be deferred until the blocks before a point are loaded and continous. > > Tamas Blummer > http://bitsofproof.com > > On 07.04.2014, at 21:03, Gregory Maxwell <gmaxwell@gmail•com> wrote: > >> On Mon, Apr 7, 2014 at 12:00 PM, Tamas Blummer <tamas@bitsofproof•com> wrote: >>> therefore I guess it is more handy to return some bitmap of pruned/full >>> blocks than ranges. >> >> A bitmap also means high overhead and— if it's used to advertise >> non-contiguous blocks— poor locality, since blocks are fetched >> sequentially. >> > > > ------------------------------------------------------------------------------ Put Bad Developers to Shame Dominate Development with Jenkins Continuous Integration Continuously Automate Build, Test & Deployment Start a new project now. Try Jenkins in the cloud.http://p.sf.net/sfu/13600_Cloudbees > _______________________________________________ Bitcoin-development mailing list Bitcoin-development@lists•sourceforge•nethttps://lists.sourceforge.net/lists/listinfo/bitcoin-development
Or have blocks distributed through pruned nodes as a DHT.
2014-04-07 20:13 GMT+01:00 Mark Friedenbach <mark@monetize•io>:
Quoted text (21 lines)
> > > On 04/07/2014 12:20 PM, Tamas Blummer wrote: >> Validation has to be sequantial, but that step can be deferred until the >> blocks before a point are loaded and continous. > > And how do you find those blocks? > > I have a suggestion: have nodes advertise which range of full blocks > they possess, then you can perform synchronization from the adversed ranges! > > ------------------------------------------------------------------------------ > Put Bad Developers to Shame > Dominate Development with Jenkins Continuous Integration > Continuously Automate Build, Test & Deployment > Start a new project now. Try Jenkins in the cloud. > http://p.sf.net/sfu/13600_Cloudbees > _______________________________________________ > Bitcoin-development mailing list > Bitcoin-development@lists•sourceforge•net > https://lists.sourceforge.net/lists/listinfo/bitcoin-development
On Mon, Apr 7, 2014 at 8:50 PM, Tamas Blummer <tamas@bitsofproof•com> wrote:
Quoted text (3 lines)
> You have to load headers sequantially to be able to connect them and > determine the longest chain. >
The isn't strictly true. If you are connected to a some honest nodes, then you could download portions of the chain and then connect the various sub-chains together.
The protocol doesn't support it though. There is no system to ask for block headers for the main chain block with a given height,
Finding one high bandwidth peer to download the entire header chain sequentially is pretty much forced. The client can switch if there is a timeout.
Other peers could be used to parallel download the block chain while the main chain is downloading. Even if the header download stalled, it wouldn't be that big a deal.
Quoted text (1 line)
> Blocks can be loaded in random order once you have their order given by
the headers.
Quoted text (1 line)
> Computing the UTXO however will force you to at least temporarily store
the blocks unless you have plenty of RAM.
You only need to store the UTXO set, rather than the entire block chain.
It is possible to generate the UTXO set without doing any signature verification.
A lightweight node could just verify the UTXO set and then do random signature verifications.
The keeps disk space and CPU reasonably low. If an illegal transaction is added to be a block, then proof could be provided for the bad transaction.
The only slightly difficult thing is confirming inflation. That can be checked on a block by block basis when downloading the entire block chain.
Quoted text (3 lines)
> Regards, > Tamas Blummer > http://bitsofproof.com <http://bitsofproof.com>
On 07.04.2014, at 21:30, Paul Lyon <pmlyon@hotmail•ca> wrote:
I hope I'm not thread-jacking here, apologies if so, but that's the approach I've taken with the node I'm working on.
Headers can be downloaded and stored in any order, it'll make sense of what the winning chain is. Blocks don't need to be downloaded in any particular order and they don't need to be saved to disk, the UTXO is fully self-contained. That way the concern of storing blocks for seeding (or not) is wholly separated from syncing the UTXO. This allows me to do the initial blockchain sync in ~6 hours when I use my SSD. I only need enough disk space to store the UTXO, and then whatever amount of block data the user would want to store for the health of the network.
This project is a bitcoin learning exercise for me, so I can only hope I don't have any critical design flaws in there. :)
------------------------------ From: tamas@bitsofproof•com Date: Mon, 7 Apr 2014 21:20:31 +0200 To: gmaxwell@gmail•com CC: bitcoin-development@lists•sourceforge•net Subject: Re: [Bitcoin-development] Why are we bleeding nodes?
Once headers are loaded first there is no reason for sequential loading.
Validation has to be sequantial, but that step can be deferred until the blocks before a point are loaded and continous.
Tamas Blummer http://bitsofproof.com
On 07.04.2014, at 21:03, Gregory Maxwell <gmaxwell@gmail•com> wrote:
On Mon, Apr 7, 2014 at 12:00 PM, Tamas Blummer <tamas@bitsofproof•com> wrote:
therefore I guess it is more handy to return some bitmap of pruned/full blocks than ranges.
A bitmap also means high overhead and-- if it's used to advertise non-contiguous blocks-- poor locality, since blocks are fetched sequentially.
------------------------------------------------------------------------------ Put Bad Developers to Shame Dominate Development with Jenkins Continuous Integration Continuously Automate Build, Test & Deployment Start a new project now. Try Jenkins in the cloud.http://p.sf.net/sfu/13600_Cloudbees
_______________________________________________ Bitcoin-development mailing list Bitcoin-development@lists•sourceforge•net https://lists.sourceforge.net/lists/listinfo/bitcoin-development
Quoted text (13 lines)
> > ------------------------------------------------------------------------------ > Put Bad Developers to Shame > Dominate Development with Jenkins Continuous Integration > Continuously Automate Build, Test & Deployment > Start a new project now. Try Jenkins in the cloud. > http://p.sf.net/sfu/13600_Cloudbees > _______________________________________________ > Bitcoin-development mailing list > Bitcoin-development@lists•sourceforge•net > https://lists.sourceforge.net/lists/listinfo/bitcoin-development > >
I actually ask for headers from each peer I’m connected to and then dump them into the backend to be sorted out.. is this abusive to the network? I’m concerned about that as I work on this, it only dawned on me the other night that I really shouldn’t use the seed peers for downloading…
I figured with the headers being so tiny, it wouldn’t be a burden to ask for them from each peer. I won’t actually end up downloading the full blockchain’s worth of headers from every peer; I’m continually getting an updated view of the current winning chain before I send out additional header requests to peers.
From: Tier Nolan Sent: Monday, April 07, 2014 6:48 PM To: bitcoin-development@lists•sourceforge•net
On Mon, Apr 7, 2014 at 8:50 PM, Tamas Blummer <tamas@bitsofproof•com> wrote:
You have to load headers sequantially to be able to connect them and determine the longest chain.
The isn't strictly true. If you are connected to a some honest nodes, then you could download portions of the chain and then connect the various sub-chains together.
The protocol doesn't support it though. There is no system to ask for block headers for the main chain block with a given height,
Finding one high bandwidth peer to download the entire header chain sequentially is pretty much forced. The client can switch if there is a timeout.
Other peers could be used to parallel download the block chain while the main chain is downloading. Even if the header download stalled, it wouldn't be that big a deal.
Quoted text (2 lines)
> Blocks can be loaded in random order once you have their order given by the headers. > Computing the UTXO however will force you to at least temporarily store the blocks unless you have plenty of RAM.
You only need to store the UTXO set, rather than the entire block chain.
It is possible to generate the UTXO set without doing any signature verification.
A lightweight node could just verify the UTXO set and then do random signature verifications.
The keeps disk space and CPU reasonably low. If an illegal transaction is added to be a block, then proof could be provided for the bad transaction.
The only slightly difficult thing is confirming inflation. That can be checked on a block by block basis when downloading the entire block chain.
Quoted text (3 lines)
> Regards, > Tamas Blummer > http://bitsofproof.com
On 07.04.2014, at 21:30, Paul Lyon <pmlyon@hotmail•ca> wrote:
I hope I'm not thread-jacking here, apologies if so, but that's the approach I've taken with the node I'm working on.
Headers can be downloaded and stored in any order, it'll make sense of what the winning chain is. Blocks don't need to be downloaded in any particular order and they don't need to be saved to disk, the UTXO is fully self-contained. That way the concern of storing blocks for seeding (or not) is wholly separated from syncing the UTXO. This allows me to do the initial blockchain sync in ~6 hours when I use my SSD. I only need enough disk space to store the UTXO, and then whatever amount of block data the user would want to store for the health of the network.
This project is a bitcoin learning exercise for me, so I can only hope I don't have any critical design flaws in there. :)
From: tamas@bitsofproof•com Date: Mon, 7 Apr 2014 21:20:31 +0200 To: gmaxwell@gmail•com CC: bitcoin-development@lists•sourceforge•net Subject: Re: [Bitcoin-development] Why are we bleeding nodes?
Once headers are loaded first there is no reason for sequential loading.
Validation has to be sequantial, but that step can be deferred until the blocks before a point are loaded and continous.
Tamas Blummer http://bitsofproof.com
On 07.04.2014, at 21:03, Gregory Maxwell <gmaxwell@gmail•com> wrote:
On Mon, Apr 7, 2014 at 12:00 PM, Tamas Blummer <tamas@bitsofproof•com> wrote:
therefore I guess it is more handy to return some bitmap of pruned/full blocks than ranges.
A bitmap also means high overhead and— if it's used to advertise non-contiguous blocks— poor locality, since blocks are fetched sequentially.
------------------------------------------------------------------------------ Put Bad Developers to Shame Dominate Development with Jenkins Continuous Integration Continuously Automate Build, Test & Deployment Start a new project now. Try Jenkins in the cloud.http://p.sf.net/sfu/13600_Cloudbees
_______________________________________________ Bitcoin-development mailing list Bitcoin-development@lists•sourceforge•nethttps://lists.sourceforge.net/lists/listinfo/bitcoin-development
On Mon, Apr 7, 2014 at 2:48 PM, Tier Nolan <tier.nolan@gmail•com> wrote:
Quoted text (5 lines)
>> Blocks can be loaded in random order once you have their order given by >> the headers. >> Computing the UTXO however will force you to at least temporarily store >> the blocks unless you have plenty of RAM. > You only need to store the UTXO set, rather than the entire block chain.
The comment was specifically in the context of an out of order fetch. Verification must be in order. If you have fetched blocks out of order you must preserve them at least long enough to reorder them. Thats all.
On Mon, Apr 7, 2014 at 10:55 PM, Paul Lyon <pmlyon@hotmail•ca> wrote:
Quoted text (3 lines)
> I actually ask for headers from each peer I'm connected to and then dump > them into the backend to be sorted out.. is this abusive to the network? >
I think downloading from a subset of the peers and switching out any slow ones is a reasonable compromise.
Once you have a chain, you can quickly check that all peers have the same main chain.
Your backend system could have a method that gives you the hash of the last 10 headers on the longest chain it knows about. You can use the block locator hash system.
This can be used with the getheaders message and if the new peer is on a different chain, then it will just send you the headers starting at the genesis block.
If that happens, you need to download the entire chain from that peer and see if it is better than your current best.
*From:* Tier Nolan <tier.nolan@gmail•com> *Sent:* Monday, April 07, 2014 6:48 PM *To:* bitcoin-development@lists•sourceforge•net
On Mon, Apr 7, 2014 at 8:50 PM, Tamas Blummer <tamas@bitsofproof•com> wrote:
Quoted text (3 lines)
> You have to load headers sequantially to be able to connect them and > determine the longest chain. >
The isn't strictly true. If you are connected to a some honest nodes, then you could download portions of the chain and then connect the various sub-chains together.
The protocol doesn't support it though. There is no system to ask for block headers for the main chain block with a given height,
Finding one high bandwidth peer to download the entire header chain sequentially is pretty much forced. The client can switch if there is a timeout.
Other peers could be used to parallel download the block chain while the main chain is downloading. Even if the header download stalled, it wouldn't be that big a deal.
Quoted text (1 line)
> Blocks can be loaded in random order once you have their order given by
the headers.
Quoted text (1 line)
> Computing the UTXO however will force you to at least temporarily store
the blocks unless you have plenty of RAM.
You only need to store the UTXO set, rather than the entire block chain.
It is possible to generate the UTXO set without doing any signature verification.
A lightweight node could just verify the UTXO set and then do random signature verifications.
The keeps disk space and CPU reasonably low. If an illegal transaction is added to be a block, then proof could be provided for the bad transaction.
The only slightly difficult thing is confirming inflation. That can be checked on a block by block basis when downloading the entire block chain.
Quoted text (3 lines)
> Regards, > Tamas Blummer > http://bitsofproof.com <http://bitsofproof.com>
On 07.04.2014, at 21:30, Paul Lyon <pmlyon@hotmail•ca> wrote:
I hope I'm not thread-jacking here, apologies if so, but that's the approach I've taken with the node I'm working on.
Headers can be downloaded and stored in any order, it'll make sense of what the winning chain is. Blocks don't need to be downloaded in any particular order and they don't need to be saved to disk, the UTXO is fully self-contained. That way the concern of storing blocks for seeding (or not) is wholly separated from syncing the UTXO. This allows me to do the initial blockchain sync in ~6 hours when I use my SSD. I only need enough disk space to store the UTXO, and then whatever amount of block data the user would want to store for the health of the network.
This project is a bitcoin learning exercise for me, so I can only hope I don't have any critical design flaws in there. :)
------------------------------ From: tamas@bitsofproof•com Date: Mon, 7 Apr 2014 21:20:31 +0200 To: gmaxwell@gmail•com CC: bitcoin-development@lists•sourceforge•net Subject: Re: [Bitcoin-development] Why are we bleeding nodes?
Once headers are loaded first there is no reason for sequential loading.
Validation has to be sequantial, but that step can be deferred until the blocks before a point are loaded and continous.
Tamas Blummer http://bitsofproof.com
On 07.04.2014, at 21:03, Gregory Maxwell <gmaxwell@gmail•com> wrote:
On Mon, Apr 7, 2014 at 12:00 PM, Tamas Blummer <tamas@bitsofproof•com> wrote:
therefore I guess it is more handy to return some bitmap of pruned/full blocks than ranges.
A bitmap also means high overhead and-- if it's used to advertise non-contiguous blocks-- poor locality, since blocks are fetched sequentially.
------------------------------------------------------------------------------ Put Bad Developers to Shame Dominate Development with Jenkins Continuous Integration Continuously Automate Build, Test & Deployment Start a new project now. Try Jenkins in the cloud.http://p.sf.net/sfu/13600_Cloudbees
_______________________________________________ Bitcoin-development mailing list Bitcoin-development@lists•sourceforge•net https://lists.sourceforge.net/lists/listinfo/bitcoin-development
Quoted text (13 lines)
> > ------------------------------------------------------------------------------ > Put Bad Developers to Shame > Dominate Development with Jenkins Continuous Integration > Continuously Automate Build, Test & Deployment > Start a new project now. Try Jenkins in the cloud. > http://p.sf.net/sfu/13600_Cloudbees > _______________________________________________ > Bitcoin-development mailing list > Bitcoin-development@lists•sourceforge•net > https://lists.sourceforge.net/lists/listinfo/bitcoin-development > >
Multi-sig requires infrastructure. It isn't a magic wand that we can wave to make everyone secure. The protocols and techniques necessary don't exist yet, and apparently no one has much of an incentive to create them.
I mean no offense, and I don't mean to pick on you. Your post stuck out while I was reading. Secure multi-sig is what we all want, but wanting apparently isn't enough to make it happen.
Other random notes from reading this 50+ post thread:
Perhaps we should have a config flag to prevent a node from serving IBD to new nodes. IBD crushes marginal machines, particularly those with spinning disks. This has been extensively discussed elsewhere.
The ideal IBD hosts are serving the blockchain out of a RAM disk. Is there any interest in setting up a network of volunteers to host expensive servers with fast connections? It doesn't look too terribly difficult to figure out when a node has stopped asking for blocks in bulk, so we could add another config flag to eject nodes once they are done booting.
Even ignoring IBD, I think that we are gradually outgrowing cheapass hosting options. Personally, I long ago gave up on answering forum questions about running nodes on virtual servers and VPSs. It is certainly still possible to run bitcoind on small boxes, but it isn't trivial any more. (Anyone running on less than my Athlon XP 1800+ with 896 MB RAM?) If we want those nodes back, we need to optimize the hell out of the memory use, and even that might not be enough.
Eric Martindale wrote:
Quoted text (14 lines)
> > We need to make it so mind-numbingly simple to "run Bitcoin correctly" > that the average user doesn't find reasons to do so in the course of > normal use. Right now, Coinbase and Bitstamp are winning in the user > experience battle, which technically endanger the user, and by proxy > the Bitcoin network. > > Multi-sig as a default is a start. It won't succeed unless the user > experience is simply better than trusted third parties, but we need to > start the education process with the very basic fundamental: trusting > a third-party with full access to your Bitcoin is just replacing one > centralized banking system with another. > >
Being Mr. Torrent, I've held open the "80% serious" suggestion to simply refuse to serve blocks older than X (3 months?).
That forces download by other means (presumably torrent).
I do not feel it is productive for any nodes on the network to waste time/bandwidth/etc. serving static, ancient data. There remain, of course, issues of older nodes and "getting the word out" that prevents this switch from being flipped on tomorrow.
On Mon, Apr 7, 2014 at 2:49 PM, Gregory Maxwell <gmaxwell@gmail•com> wrote:
Quoted text (30 lines)
> On Mon, Apr 7, 2014 at 11:35 AM, Tamas Blummer <tamas@bitsofproof•com> wrote: >> BTW, did we already agree on the service bits for an archive node? > > I'm still very concerned that a binary archive bit will cause extreme > load hot-spotting and the kind of binary "Use lots of resources YES or > NO" I think we're currently suffering some from, but at that point > enshrined in the protocol. > > It would be much better to extend the addr messages so that nodes can > indicate a range or two of blocks that they're serving, so that all > nodes can contribute fractionally according to their means. E.g. if > you want to offer up 8 GB of distributed storage and contribute to the > availability of the blockchain, without having to swollow the whole > 20, 30, 40 ... gigabyte pill. > > Already we need that kind of distributed storage for the most recent > blocks to prevent extreme bandwidth load on archives, so extending it > to arbitrary ranges is only more complicated because there is > currently no room to signal it. > > ------------------------------------------------------------------------------ > Put Bad Developers to Shame > Dominate Development with Jenkins Continuous Integration > Continuously Automate Build, Test & Deployment > Start a new project now. Try Jenkins in the cloud. > http://p.sf.net/sfu/13600_Cloudbees > _______________________________________________ > Bitcoin-development mailing list > Bitcoin-development@lists•sourceforge•net > https://lists.sourceforge.net/lists/listinfo/bitcoin-development
Isn't that just conceding that p2p protocol A is better than p2p protocol B?
Can't Bitcoin Core's block fetching be improved to get similar performance as a torrent + import?
Currently it's hard to go wide on data fetching because headers first is still pretty 'beefy'. The headers can be compressed, which would get you about 50% savings.
Also, maybe adding a layer that groups block headers into a single hash (say, 2016 headers), and then being able to fetch those (possibly compressed) header 'blocks' from multiple sources in parallel. And fanning out block fetches even further, favoring fast nodes.
Just thinking out loud.
jp
Quoted text (63 lines)
> On Apr 7, 2014, at 8:44 PM, Jeff Garzik <jgarzik@bitpay•com> wrote: > > Being Mr. Torrent, I've held open the "80% serious" suggestion to > simply refuse to serve blocks older than X (3 months?). > > That forces download by other means (presumably torrent). > > I do not feel it is productive for any nodes on the network to waste > time/bandwidth/etc. serving static, ancient data. There remain, of > course, issues of older nodes and "getting the word out" that prevents > this switch from being flipped on tomorrow. > > > >> On Mon, Apr 7, 2014 at 2:49 PM, Gregory Maxwell <gmaxwell@gmail•com> wrote: >>> On Mon, Apr 7, 2014 at 11:35 AM, Tamas Blummer <tamas@bitsofproof•com> wrote: >>> BTW, did we already agree on the service bits for an archive node? >> >> I'm still very concerned that a binary archive bit will cause extreme >> load hot-spotting and the kind of binary "Use lots of resources YES or >> NO" I think we're currently suffering some from, but at that point >> enshrined in the protocol. >> >> It would be much better to extend the addr messages so that nodes can >> indicate a range or two of blocks that they're serving, so that all >> nodes can contribute fractionally according to their means. E.g. if >> you want to offer up 8 GB of distributed storage and contribute to the >> availability of the blockchain, without having to swollow the whole >> 20, 30, 40 ... gigabyte pill. >> >> Already we need that kind of distributed storage for the most recent >> blocks to prevent extreme bandwidth load on archives, so extending it >> to arbitrary ranges is only more complicated because there is >> currently no room to signal it. >> >> ------------------------------------------------------------------------------ >> Put Bad Developers to Shame >> Dominate Development with Jenkins Continuous Integration >> Continuously Automate Build, Test & Deployment >> Start a new project now. Try Jenkins in the cloud. >> http://p.sf.net/sfu/13600_Cloudbees >> _______________________________________________ >> Bitcoin-development mailing list >> Bitcoin-development@lists•sourceforge•net >> https://lists.sourceforge.net/lists/listinfo/bitcoin-development > > > > -- > Jeff Garzik > Bitcoin core developer and open source evangelist > BitPay, Inc. https://bitpay.com/ > > ------------------------------------------------------------------------------ > Put Bad Developers to Shame > Dominate Development with Jenkins Continuous Integration > Continuously Automate Build, Test & Deployment > Start a new project now. Try Jenkins in the cloud. > http://p.sf.net/sfu/13600_Cloudbees > _______________________________________________ > Bitcoin-development mailing list > Bitcoin-development@lists•sourceforge•net > https://lists.sourceforge.net/lists/listinfo/bitcoin-development
Quoted text (5 lines)
> > Multi-sig requires infrastructure. It isn't a magic wand that we can > wave to make everyone secure. The protocols and techniques necessary > don't exist yet, and apparently no one has much of an incentive to > create them.
It is starting to happen. If you're OK with using a specific web wallet there's BitGo and greenaddress.it already, though I think their risk analysis is just sending you an SMS code. I wrote up an integration plan for bitcoinj a few days ago:
https://groups.google.com/forum/#!topic/bitcoinj/Uxl-z40OLuQ
but guess what? It's quite complicated. As with all these features.
Specialization of nodes is ongoing most prominent with SPV wallets and mining.
There is a need I see on my own business for software that is able to serve multiple wallets, and is multi tiered, so the world facing P2P node can be in a DMZ. I target them with a hybrid model that is SPV plus mempool transaction validation against UTXO and use ‘reference’ implementations as border router. I think that this setup will be common for enterprises and hence push for a stripped down ‘reference’ border router without wallet, payment protocol, GUI, RPC calls here.
That border router could also serve as archive node evtl. also offering blocks at bulk e.g. through http. Enterprises that run a multi tiered environment have the bandwith to serve as archives.
Tamas Blummer http://bitsofproof.com
On 08.04.2014, at 05:44, Jeff Garzik <jgarzik@bitpay•com> wrote:
Quoted text (51 lines)
> Being Mr. Torrent, I've held open the "80% serious" suggestion to > simply refuse to serve blocks older than X (3 months?). > > That forces download by other means (presumably torrent). > > I do not feel it is productive for any nodes on the network to waste > time/bandwidth/etc. serving static, ancient data. There remain, of > course, issues of older nodes and "getting the word out" that prevents > this switch from being flipped on tomorrow. > > > > On Mon, Apr 7, 2014 at 2:49 PM, Gregory Maxwell <gmaxwell@gmail•com> wrote: >> On Mon, Apr 7, 2014 at 11:35 AM, Tamas Blummer <tamas@bitsofproof•com> wrote: >>> BTW, did we already agree on the service bits for an archive node? >> >> I'm still very concerned that a binary archive bit will cause extreme >> load hot-spotting and the kind of binary "Use lots of resources YES or >> NO" I think we're currently suffering some from, but at that point >> enshrined in the protocol. >> >> It would be much better to extend the addr messages so that nodes can >> indicate a range or two of blocks that they're serving, so that all >> nodes can contribute fractionally according to their means. E.g. if >> you want to offer up 8 GB of distributed storage and contribute to the >> availability of the blockchain, without having to swollow the whole >> 20, 30, 40 ... gigabyte pill. >> >> Already we need that kind of distributed storage for the most recent >> blocks to prevent extreme bandwidth load on archives, so extending it >> to arbitrary ranges is only more complicated because there is >> currently no room to signal it. >> >> ------------------------------------------------------------------------------ >> Put Bad Developers to Shame >> Dominate Development with Jenkins Continuous Integration >> Continuously Automate Build, Test & Deployment >> Start a new project now. Try Jenkins in the cloud. >> http://p.sf.net/sfu/13600_Cloudbees >> _______________________________________________ >> Bitcoin-development mailing list >> Bitcoin-development@lists•sourceforge•net >> https://lists.sourceforge.net/lists/listinfo/bitcoin-development > > > > -- > Jeff Garzik > Bitcoin core developer and open source evangelist > BitPay, Inc. https://bitpay.com/ >
On 07/04/14 15:50, Gregory Maxwell wrote:
Quoted text (2 lines)
> Bitcoin.org recommends people away from running Bitcoin-QT now, so I'm > not sure that we should generally find that trend surprising.
What options are out there for people caring about 20GB blockchain? Depending of third party server is not an option.
My node (based in Dallas, TX) has about 240 connections and is using a little under 4 Mbps in bandwidth right now.
According the hosting provider I'm at 11.85 Mbps for this week, using 95th percentile billing. The report from my provider includes my other servers though.
On Mon, Apr 7, 2014 at 12:39 PM, Chris Williams <chris@icloudtools•net>wrote:
Quoted text (60 lines)
> I’m afraid this is a highly simplistic view of the costs of running a full > node. > > My node consumes fantastic amounts of data traffic, which is a real cost. > > In the 30 days ending Apri 6, my node: > > * Received 36.8 gb of data > * Sent 456.5 gb data > > At my geographic service location (Singapore), this cost about $90 last > month for bandwidth alone. It would be slightly cheaper if I was hosted in > the US of course. > > But anyone can understand that moving a half-terrabyte of data around in a > month will not be cheap. > > > On Apr 7, 2014, at 8:53 AM, Gregory Maxwell <gmaxwell@gmail•com> wrote: > > > On Mon, Apr 7, 2014 at 8:45 AM, Justus Ranvier <justusranvier@gmail•com> > wrote: > >> 1. The resource requirements of a full node are moving beyond the > >> capabilities of casual users. This isn't inherently a problem - after > >> all most people don't grow their own food, tailor their own clothes, or > >> keep blacksmith tools handy in to forge their own horseshoes either. > > > > Right now running a full node consumes about $1 in disk space > > non-reoccurring and costs a couple cents in power per month. > > > > This isn't to say things are all ducky. But if you're going to say the > > resource requirements are beyond the capabilities of casual users I'm > > afraid I'm going to have to say: citation needed. > > > > > ------------------------------------------------------------------------------ > > Put Bad Developers to Shame > > Dominate Development with Jenkins Continuous Integration > > Continuously Automate Build, Test & Deployment > > Start a new project now. Try Jenkins in the cloud. > > http://p.sf.net/sfu/13600_Cloudbees_APR > > _______________________________________________ > > Bitcoin-development mailing list > > Bitcoin-development@lists•sourceforge•net > > https://lists.sourceforge.net/lists/listinfo/bitcoin-development > > > > ------------------------------------------------------------------------------ > Put Bad Developers to Shame > Dominate Development with Jenkins Continuous Integration > Continuously Automate Build, Test & Deployment > Start a new project now. Try Jenkins in the cloud. > http://p.sf.net/sfu/13600_Cloudbees > _______________________________________________ > Bitcoin-development mailing list > Bitcoin-development@lists•sourceforge•net > https://lists.sourceforge.net/lists/listinfo/bitcoin-development > >
On that note, I think we have every possibility to make desktop and mobile wallets mind-numbingly simple -- and perhaps even do one better. Is this now a community priority? If so, I would really appreciate some additional contributors to Hive!
https://github.com/hivewallet/hive-osx https://github.com/hivewallet/hive-android
-wendell
hivewallet.com | twitter.com/hivewallet | pgp 0x8C498718 twitter.com/bitcoinbookclub | campbitcoin.hivewallet.com
On Apr 7, 2014, at 9:15 PM, Eric Martindale wrote:
Quoted text (3 lines)
> We need to make it so mind-numbingly simple to "run Bitcoin correctly" that the average user doesn't find reasons to do so in the course of normal use. Right now, Coinbase and Bitstamp are winning in the user experience battle, which technically endanger the user, and by proxy the Bitcoin network. > > Multi-sig as a default is a start. It won't succeed unless the user experience is simply better than trusted third parties, but we need to start the education process with the very basic fundamental: trusting a third-party with full access to your Bitcoin is just replacing one centralized banking system with another.
On Wed, Apr 9, 2014 at 12:38 PM, Wendell <w@hivewallet•com> wrote:
Quoted text (6 lines)
> On that note, I think we have every possibility to make desktop and mobile > wallets mind-numbingly simple -- and perhaps even do one better. Is this > now a community priority? If so, I would really appreciate some additional > contributors to Hive! > >
How does that relate to the nodes issue?
Would packaging an optional bitcoind with your wallet be an option, which is automatically managed in the background, so that users can run a full node if they want?
Wladimir
Recently China has updated its firewall blocking bitcoin sites and pools. Whether this is simple blacklist or more sophisticated packet targeting is uncertain, however this update did spefically target VPN handshakes.
Sent: Monday, April 07, 2014 at 1:07 PM
From: Drak <drak@zikula•org>
To: "Mike Hearn" <mike@plan99•net>
Cc: "Bitcoin Dev" <bitcoin-development@lists•sourceforge•net>
Subject: Re: [Bitcoin-development] Why are we bleeding nodes?
For what it's worth, the number of nodes rose dramatically during the China bullrun (I recall 45k in China alone) and dropped as dramatically as the price after the first PBOC announcement designed to cool down bitcoin trading in China.
On 7 April 2014 12:34, Mike Hearn <mike@plan99•net> wrote:
At the start of February we had 10,000 bitcoin nodes. Now we have 8,500 and still falling:
http://getaddr.bitnodes.io/dashboard/chart/?days=60
I know all the reasons why people might stop running a node (uses too much disk space, bandwidth, lost interest etc). But does anyone have any idea how we might get more insight into what's really going on? It'd be convenient if the subVer contained the operating system, as then we could tell if the bleed was mostly from desktops/laptops (Windows/Mac), which would be expected, or from virtual servers (Linux), which would be more concerning.
When you set up a Tor node, you can add your email address to the config file and the Tor project sends you emails from time to time about things you should know about. If we did the same, we could have a little exit survey: if your node disappears for long enough, we could email the operator and ask why they stopped.
------------------------------------------------------------------------------
Put Bad Developers to Shame
Dominate Development with Jenkins Continuous Integration
Continuously Automate Build, Test & Deployment
Start a new project now. Try Jenkins in the cloud.
Yeah I'm expecting port 8333 to go away in China at some point. Actually I was expecting that years ago and was kind of surprised that the suppression was being done via banks. Guess the GFW operators were just slow to catch up. On 20 May 2014 10:16, <bitcoingrant@gmx•com> wrote:
Quoted text (58 lines)
> Recently China has updated its firewall blocking bitcoin sites and pools. > Whether this is simple blacklist or more sophisticated packet targeting > is uncertain, however this update did spefically target VPN handshakes. > > *Sent:* Monday, April 07, 2014 at 1:07 PM > *From:* Drak <drak@zikula•org> > *To:* "Mike Hearn" <mike@plan99•net> > *Cc:* "Bitcoin Dev" <bitcoin-development@lists•sourceforge•net> > *Subject:* Re: [Bitcoin-development] Why are we bleeding nodes? > For what it's worth, the number of nodes rose dramatically during the > China bullrun (I recall 45k in China alone) and dropped as dramatically as > the price after the first PBOC announcement designed to cool down bitcoin > trading in China. > > On 7 April 2014 12:34, Mike Hearn <mike@plan99•net> wrote: >> >> At the start of February we had 10,000 bitcoin nodes. Now we have 8,500 >> and still falling: >> >> http://getaddr.bitnodes.io/dashboard/chart/?days=60 >> >> I know all the reasons why people *might* stop running a node (uses too >> much disk space, bandwidth, lost interest etc). But does anyone have any >> idea how we might get more insight into what's really going on? It'd be >> convenient if the subVer contained the operating system, as then we could >> tell if the bleed was mostly from desktops/laptops (Windows/Mac), which >> would be expected, or from virtual servers (Linux), which would be more >> concerning. >> >> When you set up a Tor node, you can add your email address to the config >> file and the Tor project sends you emails from time to time about things >> you should know about. If we did the same, we could have a little exit >> survey: if your node disappears for long enough, we could email the >> operator and ask why they stopped. >> >> >> ------------------------------------------------------------------------------ >> Put Bad Developers to Shame >> Dominate Development with Jenkins Continuous Integration >> Continuously Automate Build, Test & Deployment >> Start a new project now. Try Jenkins in the cloud. >> http://p.sf.net/sfu/13600_Cloudbees_APR >> _______________________________________________ >> Bitcoin-development mailing list >> Bitcoin-development@lists•sourceforge•net >> https://lists.sourceforge.net/lists/listinfo/bitcoin-development >> > > ------------------------------------------------------------------------------ > Put Bad Developers to Shame Dominate Development with Jenkins Continuous > Integration Continuously Automate Build, Test & Deployment Start a new > project now. Try Jenkins in the cloud. > http://p.sf.net/sfu/13600_Cloudbees_______________________________________________Bitcoin-development mailing list > Bitcoin-development@lists•sourceforge•net > https://lists.sourceforge.net/lists/listinfo/bitcoin-development > > >
On Tue, May 20, 2014 at 10:15:44AM +0200, bitcoingrant@gmx•com wrote:
Quoted text (3 lines)
> Recently China has updated its firewall blocking bitcoin sites and pools. > Whether this is simple blacklist or more sophisticated packet targeting is > uncertain, however this update did spefically target VPN handshakes.
Could a blockchain fork due to network split happen?
Unlikely. I doubt any significant portion of miners in china will continue to mine on a china-specific chain, since it will certainly be outmined by non-Chinese miners, and will be orphaned eventually.
More likely is that mining interests in china will make special arrangements to circumvent the GFwOC.
Users who can't access the worldwide blockchain will notice horrendously slow confirmation times and other side effects.
Quoted text (4 lines)
> On May 20, 2014, at 10:37, Eugen Leitl <eugen@leitl•org> > > Could a blockchain fork due to network split happen? >
Quoted text (15 lines)
> > > > In my opinion, the number of full nodes doesn't matter (as long as > > it's enough to satisfy demand by other nodes). > > > > Correct. Still, a high number of nodes has a few other benefits: > > 1) The more nodes there are, the cheaper it should be to run each one, > given that the bandwidth and CPU for serving the chain will be spread over > more people. > > 2) It makes Bitcoin *seem* bigger, more robust and more decentralised, > because there are more people uniting to run it. So there's a psychological > benefit. >
Psychological benefit vs. effective benefit involves the danger of destroying trust in the Bitcoin network when there are hard facts for non-robustness while the node number looks big. Therefore, it may make sense to establish better metrics.
Quoted text (16 lines)
> Also, we don't have a good way to measure capacity vs demand at the moment. > Whether we have enough capacity is rather a shot in the dark right now. > > > > What matters is how hard it is to run one. > > > > Which is why I'm interested to learn the reason behind the drop. Is it > insufficient interest, or is running a node too painful? > > For this purpose I'd like to exclude people running Bitcoin Core on laptops > or non-dedicated desktops. I don't think full nodes will ever make sense > for consumer wallets again, and I see the bleeding off of those people as > natural and expected (as Satoshi did). But if someone feels it's too hard > to run on a cheap server then that'd concern me. >
In my opinion, the characteristic of being able to make use of non-dedicated nodes should be regarded as an advantage of the Bitcoin protocol, and not something to get rid of. Nodes being able to contribute this way may lead to even more robustness than decentralization alone, as they can do so without exposing a fixed address which could be attacked.
Best regards,
Isidor
Has there ever been serious discussion on extending the protocol to support UDP transport? That would allow for NAT traversal and for many more people to run effective nodes. I'm also curious if it could be made improve block propagation time.
On Tue, May 20, 2014 at 7:52 AM, Gmail <will.yager@gmail•com> wrote:
Quoted text (22 lines)
> Unlikely. I doubt any significant portion of miners in china will continue to mine on a china-specific chain, since it will certainly be outmined by non-Chinese miners, and will be orphaned eventually. > > More likely is that mining interests in china will make special arrangements to circumvent the GFwOC. > > Users who can't access the worldwide blockchain will notice horrendously slow confirmation times and other side effects. > >> On May 20, 2014, at 10:37, Eugen Leitl <eugen@leitl•org> >> >> Could a blockchain fork due to network split happen? >> > > ------------------------------------------------------------------------------ > "Accelerate Dev Cycles with Automated Cross-Browser Testing - For FREE > Instantly run your Selenium tests across 300+ browser/OS combos. > Get unparalleled scalability from the best Selenium testing platform available > Simple to use. Nothing to install. Get started now for free." > http://p.sf.net/sfu/SauceLabs > _______________________________________________ > Bitcoin-development mailing list > Bitcoin-development@lists•sourceforge•net > https://lists.sourceforge.net/lists/listinfo/bitcoin-development >
Yes, i spec'd out the UDP traversal of the P2P protocol. It seems reasonable especially for "inv" messages.
On Tue, May 20, 2014 at 2:46 PM, Andy Alness <andy@coinbase•com> wrote:
Quoted text (47 lines)
> Has there ever been serious discussion on extending the protocol to > support UDP transport? That would allow for NAT traversal and for many > more people to run effective nodes. I'm also curious if it could be > made improve block propagation time. > > On Tue, May 20, 2014 at 7:52 AM, Gmail <will.yager@gmail•com> wrote: >> Unlikely. I doubt any significant portion of miners in china will continue to mine on a china-specific chain, since it will certainly be outmined by non-Chinese miners, and will be orphaned eventually. >> >> More likely is that mining interests in china will make special arrangements to circumvent the GFwOC. >> >> Users who can't access the worldwide blockchain will notice horrendously slow confirmation times and other side effects. >> >>> On May 20, 2014, at 10:37, Eugen Leitl <eugen@leitl•org> >>> >>> Could a blockchain fork due to network split happen? >>> >> >> ------------------------------------------------------------------------------ >> "Accelerate Dev Cycles with Automated Cross-Browser Testing - For FREE >> Instantly run your Selenium tests across 300+ browser/OS combos. >> Get unparalleled scalability from the best Selenium testing platform available >> Simple to use. Nothing to install. Get started now for free." >> http://p.sf.net/sfu/SauceLabs >> _______________________________________________ >> Bitcoin-development mailing list >> Bitcoin-development@lists•sourceforge•net >> https://lists.sourceforge.net/lists/listinfo/bitcoin-development >> > > > > -- > Andy Alness > Software Engineer > Coinbase > San Francisco, CA > > ------------------------------------------------------------------------------ > "Accelerate Dev Cycles with Automated Cross-Browser Testing - For FREE > Instantly run your Selenium tests across 300+ browser/OS combos. > Get unparalleled scalability from the best Selenium testing platform available > Simple to use. Nothing to install. Get started now for free." > http://p.sf.net/sfu/SauceLabs > _______________________________________________ > Bitcoin-development mailing list > Bitcoin-development@lists•sourceforge•net > https://lists.sourceforge.net/lists/listinfo/bitcoin-development
Awesome! I'm assuming this is it: https://bitcointalk.org/index.php?topic=156769.0
It would be interesting (at least to me) to take this a step further and offer UDP as a full TCP replacement capable of STUN-assisted NAT traversal and possibly swarmed blockchain syncs. It would require open TCP nodes to facilitate "connection" establishment. It is obviously a non-trivial amount of work but would be an interesting experiment. Maybe BitTorrent's µTP protocol could be leveraged.
On Tue, May 20, 2014 at 12:17 PM, Jeff Garzik <jgarzik@bitpay•com> wrote:
Quoted text (58 lines)
> Yes, i spec'd out the UDP traversal of the P2P protocol. It seems > reasonable especially for "inv" messages. > > On Tue, May 20, 2014 at 2:46 PM, Andy Alness <andy@coinbase•com> wrote: >> Has there ever been serious discussion on extending the protocol to >> support UDP transport? That would allow for NAT traversal and for many >> more people to run effective nodes. I'm also curious if it could be >> made improve block propagation time. >> >> On Tue, May 20, 2014 at 7:52 AM, Gmail <will.yager@gmail•com> wrote: >>> Unlikely. I doubt any significant portion of miners in china will continue to mine on a china-specific chain, since it will certainly be outmined by non-Chinese miners, and will be orphaned eventually. >>> >>> More likely is that mining interests in china will make special arrangements to circumvent the GFwOC. >>> >>> Users who can't access the worldwide blockchain will notice horrendously slow confirmation times and other side effects. >>> >>>> On May 20, 2014, at 10:37, Eugen Leitl <eugen@leitl•org> >>>> >>>> Could a blockchain fork due to network split happen? >>>> >>> >>> ------------------------------------------------------------------------------ >>> "Accelerate Dev Cycles with Automated Cross-Browser Testing - For FREE >>> Instantly run your Selenium tests across 300+ browser/OS combos. >>> Get unparalleled scalability from the best Selenium testing platform available >>> Simple to use. Nothing to install. Get started now for free." >>> http://p.sf.net/sfu/SauceLabs >>> _______________________________________________ >>> Bitcoin-development mailing list >>> Bitcoin-development@lists•sourceforge•net >>> https://lists.sourceforge.net/lists/listinfo/bitcoin-development >>> >> >> >> >> -- >> Andy Alness >> Software Engineer >> Coinbase >> San Francisco, CA >> >> ------------------------------------------------------------------------------ >> "Accelerate Dev Cycles with Automated Cross-Browser Testing - For FREE >> Instantly run your Selenium tests across 300+ browser/OS combos. >> Get unparalleled scalability from the best Selenium testing platform available >> Simple to use. Nothing to install. Get started now for free." >> http://p.sf.net/sfu/SauceLabs >> _______________________________________________ >> Bitcoin-development mailing list >> Bitcoin-development@lists•sourceforge•net >> https://lists.sourceforge.net/lists/listinfo/bitcoin-development > > > > -- > Jeff Garzik > Bitcoin core developer and open source evangelist > BitPay, Inc. https://bitpay.com/
Indeed -- you must reinvent TCP over UDP, ultimately, to handle blocks and large TXs.
On Tue, May 20, 2014 at 4:09 PM, Andy Alness <andy@coinbase•com> wrote:
Quoted text (77 lines)
> Awesome! I'm assuming this is it: > https://bitcointalk.org/index.php?topic=156769.0 > > It would be interesting (at least to me) to take this a step further > and offer UDP as a full TCP replacement capable of STUN-assisted NAT > traversal and possibly swarmed blockchain syncs. It would require open > TCP nodes to facilitate "connection" establishment. It is obviously a > non-trivial amount of work but would be an interesting experiment. > Maybe BitTorrent's µTP protocol could be leveraged. > > On Tue, May 20, 2014 at 12:17 PM, Jeff Garzik <jgarzik@bitpay•com> wrote: >> Yes, i spec'd out the UDP traversal of the P2P protocol. It seems >> reasonable especially for "inv" messages. >> >> On Tue, May 20, 2014 at 2:46 PM, Andy Alness <andy@coinbase•com> wrote: >>> Has there ever been serious discussion on extending the protocol to >>> support UDP transport? That would allow for NAT traversal and for many >>> more people to run effective nodes. I'm also curious if it could be >>> made improve block propagation time. >>> >>> On Tue, May 20, 2014 at 7:52 AM, Gmail <will.yager@gmail•com> wrote: >>>> Unlikely. I doubt any significant portion of miners in china will continue to mine on a china-specific chain, since it will certainly be outmined by non-Chinese miners, and will be orphaned eventually. >>>> >>>> More likely is that mining interests in china will make special arrangements to circumvent the GFwOC. >>>> >>>> Users who can't access the worldwide blockchain will notice horrendously slow confirmation times and other side effects. >>>> >>>>> On May 20, 2014, at 10:37, Eugen Leitl <eugen@leitl•org> >>>>> >>>>> Could a blockchain fork due to network split happen? >>>>> >>>> >>>> ------------------------------------------------------------------------------ >>>> "Accelerate Dev Cycles with Automated Cross-Browser Testing - For FREE >>>> Instantly run your Selenium tests across 300+ browser/OS combos. >>>> Get unparalleled scalability from the best Selenium testing platform available >>>> Simple to use. Nothing to install. Get started now for free." >>>> http://p.sf.net/sfu/SauceLabs >>>> _______________________________________________ >>>> Bitcoin-development mailing list >>>> Bitcoin-development@lists•sourceforge•net >>>> https://lists.sourceforge.net/lists/listinfo/bitcoin-development >>>> >>> >>> >>> >>> -- >>> Andy Alness >>> Software Engineer >>> Coinbase >>> San Francisco, CA >>> >>> ------------------------------------------------------------------------------ >>> "Accelerate Dev Cycles with Automated Cross-Browser Testing - For FREE >>> Instantly run your Selenium tests across 300+ browser/OS combos. >>> Get unparalleled scalability from the best Selenium testing platform available >>> Simple to use. Nothing to install. Get started now for free." >>> http://p.sf.net/sfu/SauceLabs >>> _______________________________________________ >>> Bitcoin-development mailing list >>> Bitcoin-development@lists•sourceforge•net >>> https://lists.sourceforge.net/lists/listinfo/bitcoin-development >> >> >> >> -- >> Jeff Garzik >> Bitcoin core developer and open source evangelist >> BitPay, Inc. https://bitpay.com/ > > > > -- > Andy Alness > Software Engineer > Coinbase > San Francisco, CA