Call for reconfiguration of nodes to relay transactions with fee-rates below 1 sat/vbyte
6 messages, 31 Jan to 3 Feb 2025. Started by Greg Tonoski.
Sjors Provoost (3), Greg Tonoski (2), Murch
The minrelaytxfee default value of 1000/kvB (1 sat/vbyte) has been set in Bitcoin nodes since 2013 or earlier. In 2013, 1000 sats were priced at $0.01, today it is about 100 times more. Wouldn't you like to consider adjusting your node configuration and lower the hurdle, perhaps? I would suggest setting minrelaytxfee=0.00000001 (which translates to 0.001 sat/vbyte) as a rule of thumb. The value could be set in a bitcoin.conf file. There is negligible number of such reconfigured nodes relaying transactions with fee-rates below 1 sat/vbyte currently.
There is the intention to unblock transactions with fee-rate below 1 sat/vbyte in Bitcoin. Probably consolidation and multisig types of transactions would benefit the most.
If you want to go this route, you may also want to set -incrementalrelayfee. That way you can bump 0.001 ṩ/vB to 0.002 ṩ/vB, otherwise you'll have to go straight to 1.001 ṩ/vB.
In addition, you would need to convince at least one miner to configure a matching -blockmintxfee, otherwise these low fee transactions remain stuck in your mempool until expiration (or CPFP).
Some background: https://bitcoin.stackexchange.com/a/120782/4948
A downside to consider is that it makes any potential mempool DoS attack 1000x cheaper*, but I don't know if there are any to worry about.
Comparison with historical minimum fee rates can be tricky, especially because the lightning network wasn't around yet. There's now an increased incentive to knock out a full node if it's connected to a lightning node. So even if there's no vulnerability to attack today, it's probably good to keep fees somewhat high to ensure new attacks are expensive.
- Sjors
* = assuming the attack scales linear with the amount of transaction data, and not e.g. quadratic or exponential
Quoted text (15 lines)
> Op 31 jan 2025, om 09:49 heeft Greg Tonoski <gregtonoski@gmail•com> het volgende geschreven: > > The minrelaytxfee default value of 1000/kvB (1 sat/vbyte) has been set > in Bitcoin nodes since 2013 or earlier. In 2013, 1000 sats were priced > at $0.01, today it is about 100 times more. Wouldn't you like to > consider adjusting your node configuration and lower the hurdle, > perhaps? I would suggest setting minrelaytxfee=0.00000001 (which > translates to 0.001 sat/vbyte) as a rule of thumb. The value could be > set in a bitcoin.conf file. There is negligible number of such > reconfigured nodes relaying transactions with fee-rates below 1 > sat/vbyte currently. > > There is the intention to unblock transactions with fee-rate below 1 > sat/vbyte in Bitcoin. Probably consolidation and multisig types of > transactions would benefit the most.
I agree that -incrementalrelayfee=0 (or whatever suits a node runner) would logically supplement the minrelaytxfee=0.00000001.
I suppose that miners already use -blockmintxfee=0 or anything lower than the default value because there are transactions with fees as low as 0 (zero) in the blocks.
I can't see how minrelaytxfee=0.00000001 could increase risk of DoS attack or make it significantly cheaper or more effective. There are the default 300MB size limit for mempool and 336 hours timeout for unconfirmed txs. They limit impact of a low fee-rate txs DoS attack making it ineffective.
On Fri, Jan 31, 2025 at 1:59 PM Sjors Provoost <sjors@sprovoost•nl> wrote:
Quoted text (35 lines)
> > If you want to go this route, you may also want to set -incrementalrelayfee. That way you can bump 0.001 ṩ/vB to 0.002 ṩ/vB, otherwise you'll have to go straight to 1.001 ṩ/vB. > > In addition, you would need to convince at least one miner to configure a matching -blockmintxfee, otherwise these low fee transactions remain stuck in your mempool until expiration (or CPFP). > > Some background: https://bitcoin.stackexchange.com/a/120782/4948 > > A downside to consider is that it makes any potential mempool DoS attack 1000x cheaper*, but I don't know if there are any to worry about. > > Comparison with historical minimum fee rates can be tricky, especially because the lightning network wasn't around yet. There's now an increased incentive to knock out a full node if it's connected to a lightning node. So even if there's no vulnerability to attack today, it's probably good to keep fees somewhat high to ensure new attacks are expensive. > > - Sjors > > * = assuming the attack scales linear with the amount of transaction data, and not e.g. quadratic or exponential > > > Op 31 jan 2025, om 09:49 heeft Greg Tonoski <gregtonoski@gmail•com> het volgende geschreven: > > > > The minrelaytxfee default value of 1000/kvB (1 sat/vbyte) has been set > > in Bitcoin nodes since 2013 or earlier. In 2013, 1000 sats were priced > > at $0.01, today it is about 100 times more. Wouldn't you like to > > consider adjusting your node configuration and lower the hurdle, > > perhaps? I would suggest setting minrelaytxfee=0.00000001 (which > > translates to 0.001 sat/vbyte) as a rule of thumb. The value could be > > set in a bitcoin.conf file. There is negligible number of such > > reconfigured nodes relaying transactions with fee-rates below 1 > > sat/vbyte currently. > > > > There is the intention to unblock transactions with fee-rate below 1 > > sat/vbyte in Bitcoin. Probably consolidation and multisig types of > > transactions would benefit the most. > > -- > You received this message because you are subscribed to the Google Groups "Bitcoin Development Mailing List" group. > To unsubscribe from this group and stop receiving emails from it, send an email to bitcoindev+unsubscribe@googlegroups•com. > To view this discussion visit https://groups.google.com/d/msgid/bitcoindev/A8F96312-613D-4090-B6C9-0A4C41418986%40sprovoost.nl.
Quoted text (3 lines)
> I suppose that miners already use -blockmintxfee=0 or anything lower > than the default value because there are transactions with fees as low > as 0 (zero) in the blocks.
It seems more likely that they used the prioritisetransaction RPC [0] in order to include only those 0 fee transaction that they were paid for (via lightning, creditcard).
Quoted text (5 lines)
> I can't see how minrelaytxfee=0.00000001 could increase risk of DoS > attack or make it significantly cheaper or more effective. There are > the default 300MB size limit for mempool and 336 hours timeout for > unconfirmed txs. They limit impact of a low fee-rate txs DoS attack > making it ineffective.
I could quote Donald Rumsfeld, but instead I'll point out that you're forgetting about bandwidth, memory[1] and CPU usage.
- Sjors
[0] https://bitcoincore.org/en/doc/28.0.0/rpc/mining/prioritisetransaction/ [1] e.g. https://bitcoincore.org/en/2024/10/08/disclose-large-inv-to-send/
Quoted text (3 lines)
> > I could quote Donald Rumsfeld, but instead I'll point out that you're > forgetting about bandwidth, memory[1] and CPU usage.
Memory is of course accounted for with -maxmempool. I meant to say:
bandwidth and CPU usage. [1]
Quoted text (1 line)
> [1] e.g. https://bitcoincore.org/en/2024/10/08/disclose-large-inv-to-send/
Hi Greg,
On 2025-01-31 08:43, Greg Tonoski wrote:
Quoted text (2 lines)
> I agree that -incrementalrelayfee=0 (or whatever suits a node runner) > would logically supplement the minrelaytxfee=0.00000001.
Setting `incrementalrelayfee` to zero would mean that you allow the replacement of an original transaction without increasing the total fees in the mempool. A malicious actor could trivially exploit this setting by cycling two conflicting transactions A and A' with the same weight and same fee to waste bandwidth and CPU cycles across the nodes that follow your advice.
On 2025-01-31 08:43, Greg Tonoski wrote:
Quoted text (3 lines)
> I suppose that miners already use -blockmintxfee=0 or anything lower > than the default value because there are transactions with fees as low > as 0 (zero) in the blocks.
Do you have any evidence for this claim? The prior times that I investigated this claim, transactions with feerates lower than `minTxRelayFee` exclusively appeared at the front of the block, indicating that they were explicitly prioritized by the miner, probably due to out-of-band payment. You may now also find some TRUC transactions where a low-feerate parent is carried by a higher feerate child. Neither of these two indicates that a miner has configured a lower `-blockmintxfee`. You could convince me that a miner has configured a lower `-blockmintxfee` by showing us a block that includes transactions with feerates below `minTxRelayFee` appearing in the organic descending feerate order.
On 2025-01-31 08:43, Greg Tonoski wrote:
Quoted text (5 lines)
> I can't see how minrelaytxfee=0.00000001 could increase risk of DoS > attack or make it significantly cheaper or more effective. There are > the default 300MB size limit for mempool and 336 hours timeout for > unconfirmed txs. They limit impact of a low fee-rate txs DoS attack > making it ineffective.
Unfortunately, not knowing about an exploit falls short of proving something secure.
Murch