What happend to the mempool on the July 22nd ?

[The Hidebehinder]

I admit I'm always looking at the mempool and at the fees more than one should, a few $ there and there when multiplied by a dozen start to sting so I have the habit of checking it all the time, and a few days ago I saw that spike in fees, I never realized what the crash in volume afterwards was, it passed but it's still bugging me a lot  that I have to ask about. Assuming both mempools dropped a lot of tx as they've reached the max size at the same time why hasn't the mempool filed up back again with transactions lover than 2sat/b as I don't doubt there are nodes with far more space than 300MB that would have still held them in their own pool and broadcast them??




*guess I'm still a newbie so images don't show unfortunately!
*should this be in technical or simple discussion?

[1715296509]

1715296509

[1715296509]

1715296509

[1715296509]

1715296509

[1715296509]

1715296509

[1715296509]

1715296509

[hugeblack]

it depend on miner/pool, If there are a few transactions to be sent, then we will definitely return to the levels of 1 sat/vByte but allowing such minimum limits means that their profits will decrease, and therefore it is not a good option for them at current status.
Now some books are mined at an average of 8 sat/vByte, if this fees remain low for a long time, some nodes may allow lower fees down to 1 sat/vByte.

It is a good opportunity for anyone who wants to reduce fees by reducing the number of inputs.

[hosseinimr93]

Assuming both mempools dropped a lot of tx as they've reached the max size at the same time why hasn't the mempool filed up back again with transactions lover than 2sat/b as I don't doubt there are nodes with far more space than 300MB that would have still held them in their own pool and broadcast them??

What do you mean by both mempools?

Nodes may drop a transaction from their mempool due to reaching their maximum size limit, but it's not that they will put that transaction again in their mempool, whenever they have space for that. Take note that each node broadcasts and relay every transaction to other nodes only one time, unless someone rebroadcasts the transaction.
If you see your transaction has been dropped from mempool of some nodoes and you want  them to put your transaction in their mempool again, you have rebroadcast your transaction.

[SquirrelJulietGarden]

It is a good opportunity for anyone who wants to reduce fees by reducing the number of inputs.

They can prepare for their future transactions with less number of inputs like your advice, by consolidating small inputs. Consolidation should be done with cheap transaction fee. If in future, they must move their bitcoin when mempools are more loaded with higher fee rate, they can accept higher fee rate but with less inputs, they will save transaction fee.

Assuming both mempools dropped a lot of tx as they've reached the max size at the same time why hasn't the mempool filed up back again with transactions lover than 2sat/b as I don't doubt there are nodes with far more space than 300MB that would have still held them in their own pool and broadcast them??

There are many mempools and each mining pools will have their full nodes and own settings for their mempools and how long they will drop unconfirmed transactions in their mempools.

Usually it is 14 days but mempools can be set to be shorter or longer.

Code:

mempoolexpiry=336

They can set it to 336 hours (14 days) and change it to longer or shorter.

[ranochigo]

The way that the transaction is represented on the graph can be quite confusing. Johoe's mempool likely didn't drop the transactions and if it did, you'll see the proportion of 1sat/byte transactions decreasing faster than the rest. Remember that the minimum relay fee basically means evicting transactions below a certain fee rate threshold which is not what we are observing here. Hence, the expected behavior for this would be having the 1sat transactions dipping while the rest are largely consistent.

Instead, you can see that the size of the mempool remained largely consistent, with the feerates across all levels decreasing simultaneously while the proportion of transactions paying higher fees increasing. This is possibly an effect of those transaction being replaced by another with a higher fee rate.

[The Hidebehinder]

What do you mean by both mempools?

Mempool.space and Johoe's mempools, those are the two that I check and both had the same drop.

Nodes may drop a transaction from their mempool due to reaching their maximum size limit, but it's not that they will put that transaction again in their mempool, whenever they have space for that. Take note that each node broadcasts and relay every transaction to other nodes only one time, unless someone rebroadcasts the transaction.

I doubt that less of a quarter transactions that have stayed that long in the mempool since we had the ordinal spam have not been rebroadcasted over and over.
This thing for example is nearly 3 months old:
https://mempool.space/tx/2c36ea57aeced9e662a04d12ca7a184fd0af56704808bc5473ef40c43baaef72

The way that the transaction is represented on the graph can be quite confusing. Johoe's mempool likely didn't drop the transactions and if it did, you'll see the proportion of 1sat/byte transactions decreasing faster than the rest. Remember that the minimum relay fee basically means evicting transactions below a certain fee rate threshold which is not what we are observing here. Hence, the expected behavior for this would be having the 1sat transactions dipping while the rest are largely consistent.

Yes but looking at Joe's mempool
At the start of the day there were 185Mvb with 155Mvb over 1sat and 99Mvb over 2sat/vb
After the drop at 12PM,  130Mvb , 101Mvb over 1+ 82Mvb over 2sat/vb
As we speak we have 140Mvb with 109 and 92 over 1 and 2sat/vb

So in 10 full days we had just 20 blocks worth of low fee transactions added although obviously there is space in the mempool to keep them and this including rebroadcasted tx?

[ranochigo]

Yes but looking at Joe's mempool
At the start of the day there were 185Mvb with 155Mvb over 1sat and 99Mvb over 2sat/vb
After the drop at 12PM,  130Mvb , 101Mvb over 1+ 82Mvb over 2sat/vb
As we speak we have 140Mvb with 109 and 92 over 1 and 2sat/vb

So in 10 full days we had just 20 blocks worth of low fee transactions added although obviously there is space in the mempool to keep them and this including rebroadcasted tx?

Dropped transactions aren't rebroadcasted automatically. They are just discarded and someone has to rebroadcast them. The memory usage of mempool functions slightly differently from your actual memory, and there are overheads. In fact, the minrelayfee right now is around 3sat/byte instead of 1.

If purging were to be done, then it would be the lowest fee rates first. In this case, you can observe a substantial decrease in size of both 1sat/byte and 2sat/byte, which is unlikely to happen if the nodes were to only purge those transactions that only pays 1 sat/byte. As far as I know, mempool.space does have a larger mempool size than most. More likely than not, some of the transactions were RBFed and thus you can observe a corresponding spike in the high fee rate transactions.

[stompix]

The way that the transaction is represented on the graph can be quite confusing. Johoe's mempool likely didn't drop the transactions and if it did, you'll see the proportion of 1sat/byte transactions decreasing faster than the rest.

It actually dropped, they've dumped quite a ton of them, and the 1sat/b was the only one affected.
That said, yeah, Johoe's layout is not really my cup of tea.

As for OP, I can tell you what happened, a ton of ordinal inscriptions that while small in number of tx large in size pushed a lot of older transactions out of the mempool that was just clearing a bit up, so suddenly every single estimator went nuts, people stopped sending 4-5 sat/b tx and this discouraged rebroadcasting of older ones so that's why the slow build-up back for 1/sat b in the mempool.

To be honest, is understandable, you waited 3 months rebroadcasting your old tx over and over, and just when you see the light you see another wave that pushes you again out of all 300MB mempools of course they would give up just CPFP so highly probably a few of the 1sat/b have just been upgraded o 4-5 or even 10sat/b as some lost patience. I would have probably done the same, just pay 5-10$ and get it off my mind.

[ranochigo]

It actually dropped, they've dumped quite a ton of them, and the 1sat/b was the only one affected.
That said, yeah, Johoe's layout is not really my cup of tea.

Hmm? I was looking at mempool.space's far better visualization. I don't think Mempool.space actually purges the transaction in this manner, or else we wouldn't see any 1-3sat/byte transaction in their graph right now. Their site says anything below <3sat/byte is being purged but a good proportion of their mempool graph still comprises of that.

There is a 7MB decrease in the composition of 2-3sat/byte transaction around the same period of time. The alternate possibility is a bunch of RBFs that shifted the transactions in the graph while maintaining similar total size.

[stompix]

Hmm? I was looking at mempool.space's far better visualization. I don't think Mempool.space actually purges the transaction in this manner, or else we wouldn't see any 1-3sat/byte transaction in their graph right now. Their site says anything below <3sat/byte is being purged but a good proportion of their mempool graph still comprises of that.

Yeah, saw this debate, if I remember correctly the purging is for 300MB, it's an indicator for others as to what would happen, not that they do in reality, they have well over that and the mempool is expressed in graphics in vMB which is god knows how much right now in MB.

Anyhow, you can see the drop in the chart with them too
July 22, 00:00 you have 56MvB 12 hours later you have 18MvB, nearly 30 blocks worth of tx.