Missing transactions from full node

Question

I am running a full node (bitcoin-core) that connects to the peers network. I am also running and SPV node with a wallet (bcoin). I have made 2 different send transaction from the SPV wallet with very low fee (1 and 3 statoshi per vbyte). I can see these transactions on Blockchair explorer (pending of course) but i did not see those transaction on my full node. usually i see all transactions. why is my full node don't get these transactions? is it related to the very low fee? or maybe its related to the spv wallet?

thanks for your help

Answer 1

The full node you are running likely has the standard mempool size (300 MB) and default settings. At the moment (see e.g. https://mempool.space/), transactions with a fee less than 5.88 sat/vB are purged from the mempool.

In your case, the two transactions from the SPV wallet that had very low fees (1 and 3 sat/vB) were likely not accepted into your full node's mempool due to this fee threshold. However, these transactions were still propagated through the bitcoin network, as they were accepted by another node that accepted transactions with a lower minimum fee rate. This node in turn relayed the transactions to other nodes, eventually reaching the Blockchair node.

As a result, those two transactions currently exist only in the mempools of certain full nodes that have a lower minimum fee rate, which is why you can see them on the Blockchair explorer but not on your full node.

Answer 2

Bitcoin Core uses a 300 MiB limit for its mempool data structure by default. If enough transactions get submitted to the network, this limit will be reached and Bitcoin Core will start dropping transactions with the lowest feerates. At that point it will also raise its dynamic minimum feerate for admittance to its mempool.

The network's blockspace market has seen a lot more demand in the last six months. For the past month, the dynamic minimum feerate for consistently online Bitcoin Core nodes with the default limit was usually over 5 ṩ/vB. So, likely your own node didn't accept it to if memory, while blockchair is running their node with a bigger limit.