Core: Can't customize RBF fee increase?

[jnano]

The "increase transaction fee" option doesn't offer a way to customize the new fee.
How come, and how's the automatic new fee calculated?

[Anycoin Direct Support]

Hello jnano,

To overcome the problem you are referring to you could press the increase transaction fee as many times as you like. In this way you can make sure the transaction will definitely get confirmed within a short amount of time. Of course this is a work-around to the issue you are addressing rather than a fix. Therefore we suggest that you use one of the following methods.

Go to your list of transactions, right click your transaction and choose abandon transaction. Now you can send the btc again with via the send menu with the fee of your liking.

If you cannot abandon the transaction, you could go to the datadir and delete the mempool.dat file when you restart your wallet you should be able to abandon and resend the transaction. 

If both these methods don't work start the Bitcoin core wallet again and use the -zapwallettxes- option to clear your unconfirmed transactions and resend the coins with a fee of your liking.

Hopefully this will help speed your transaction along.

With kind regards,
The Anycoin Direct team       

[jnano]

Thanks.

Is "abandon transaction", then a manual resend, really the same as the action performed by "increase transaction fee"?

Using "increase" multiple doesn't really let you customize the fee. I think it would also spam the mempool with a rapid succession of conflicting transactions?

[Anycoin Direct Support]

Thanks.

Is "abandon transaction", then a manual resend, really the same as the action performed by "increase transaction fee"?

Using "increase" multiple doesn't really let you customize the fee. I think it would also spam the mempool with a rapid succession of conflicting transactions?

You are correct it is not the same, as mentioned before it is a work-around and not a fix to your problem. However, by manually abandoning transaction and resending the order you will have the highest amount of customizability.

[Thirdspace]

Is "abandon transaction", then a manual resend, really the same as the action performed by "increase transaction fee"?

not exactly, "abandon then resend" will work smoothly only if your original tx has not been broadcasted widely
your new tx will be rejected by nodes that still have your original tx in their mempools

to make use RBF, your original tx must have already signal (opt-in) RBF in the first place
read this for detailed reference https://bitcoin.org/en/developer-reference#bumpfee

[HCP]

Using "increase" multiple doesn't really let you customize the fee. I think it would also spam the mempool with a rapid succession of conflicting transactions?

No... that is the whole point of "Replace-By-Fee".

If your original transaction is marked as RBF... then nodes that support RBF will simply drop your original transaction, when they receive the replacement transaction.

[jnano]

I know that's the "replace" in RBF, but I'm not sure what happens when you abandon+resend, as opposed to just using "increase".

The original had the RBF flag set, otherwise "increase transaction fee" wouldn't be available.

My guess is that the automatic fee for "increase" is similar to what you'd get doing an initial send using a "recommended" fee with a small number of blocks for "confirmation time target".

And still I'm curious, why no way to customize the fee?

[DarkStar_]

And still I'm curious, why no way to customize the fee?

It's doable, but not in the GUI. See this if you're using Bitcoin Core. It's a limitation of the program and not the RBF protocol.

[jnano]

Yeah, I supposed it's just a GUI issue. Thanks for finding it on GitHub.

Looks like the GUI just uses bumpfee in its default mode, which says:

By default, the new fee will be calculated automatically using estimatefee.

According to the docs here the estimation defaults to targeting the next 6 blocks.

It's always a surprise how incomplete Core's GUI is after all these years.

[HCP]

I know that's the "replace" in RBF, but I'm not sure what happens when you abandon+resend, as opposed to just using "increase".

Abandon simply marks the transaction as abandoned in YOUR wallet and drops it from your local mempool.

The end result being that your wallet will "free up" the coins internally and allow you to re-spend them. However, it has NO effect on other nodes or their mempools. So, there is no guarantee that you will in fact be able to re-spend those coins, as most nodes are likely to reject any new transaction as a conflicting "double-spend". Especially since the default "mempool expiry" time was pushed out to 14 days. Obviously, if the mempool expiry time has passed and the node has already dropped your old transaction, then it should accept your new one.

Without RBF (Abandon/Resend):
A node sees your new transaction, identifies it as a "double-spend" and rejects it/refuses to relay it. Your original low-fee transaction remains in the node's mempool

With RBF:
A node sees your new transaction, identifies it as an "RBF", drops the original low-fee transaction and replaces is with your new higher-fee transaction.

It's always a surprise how incomplete Core's GUI is after all these years.

I always get the feeling that the focus is on implementing features and bugfixes within bitcoind. Whether or not BitcoinQt is able to utilise these features is a different kettle of fish.

[jnano]

But it's abandon-resend of an RBF tx.

If the change address ends up being different then I guess it will be considered a double-spend. But if after "abandon" the change address is freed and will be reused for the next tx, when sending the same UTXOs to the same recepient wouldn't it effective be the same as an "increase" tx?

Whatever the reason for Core's GUI deficiencies, it's unfortunate. There are a lot of low-hanging fruits.

[bob123]

If the change address ends up being different then I guess it will be considered a double-spend.

Yes, thats correct. At least as a double-spend attempt.

But if after "abandon" the change address is freed and will be reused for the next tx, when sending the same UTXOs to the same recepient wouldn't it effective be the same as an "increase" tx?

Yes.
A RBF-transaction is basically a double-spend transaction which uses the same UTXO from previous tx (additional UTXO's are allowed) to the same recipient with a higher fee.