Retry requesting pooled transactions

[flcl42]

Fixes hive tests

• Tx still should be requested if correct and incorrect txs are being returned by different peers
• If a peer does not send tx back, it will probably block from requesting from others In tx pool presence check is enough

Changes

• Request new pooled transaction only from the first peer. But if it's not returned or returned wrong - request from all the peers announced it after a timeout (5 seconds for now)

Types of changes

What types of changes does your code introduce?

• Bugfix (a non-breaking change that fixes an issue)
• New feature (a non-breaking change that adds functionality)
• Breaking change (a change that causes existing functionality not to work as expected)
• Optimization
• Refactoring
• Documentation update
• Build-related changes
• Other: Description

Testing

Requires testing

• Yes
• No

If yes, did you write tests?

• Yes
• No (covered by hive)

[marcindsobczak]

It will end with requesting the same transactions over and over again as our default TxPool has only capacity of 2048 txs and there is way more of them in the network. So without cache we will request all underpaid transactions, it will not pass filters and we will request it again and again.
We need more complex solution - a map between tx requested and received. This way we will be able to remove tx hash from cache if we don't have a response or something else is wrong. We will also be able to disconnect peers sending to us unrequested blob/big txs or with different params than announced (size, txtype).

[flcl42]

It will end with requesting the same transactions over and over again as our default TxPool has only capacity of 2048 txs and there is way more of them in the network. So without cache we will request all underpaid transactions, it will not pass filters and we will request it again and again. We need more complex solution - a map between tx requested and received. This way we will be able to remove tx hash from cache if we don't have a response or something else is wrong. We will also be able to disconnect peers sending to us unrequested blob/big txs or with different params than announced (size, txtype).

What prevents us from maintaining just a txhash cache inside TxPool.SubmitTx? If not in the cache - continue with submitting; externally, if not in the txpool cache - request em.

[marcindsobczak]

It is fine, as TxPool.IsKnown is checking hashCache inside TxPool, so we will not request one tx many times. Sorry for a false alarm.

[benaadams]

Might be better to shrink the the cache rather than remove entirely?

[Copilot]

Pull Request Overview

This PR fixes transaction request handling in the TxPool to properly handle corner cases where different peers return correct or incorrect transactions. The main issue being addressed is ensuring transactions are still requested when both correct and incorrect transactions are being returned by different peers, and preventing peers that don't respond from blocking requests to other peers.

Key Changes

• Removes cache for requesting transaction hashes to allow retries from different peers
• Implements a new SimpleRetryCache system for tracking transaction requests with timeouts
• Updates the minimum blob priority fee from 1 GWei to 0 GWei

Reviewed Changes

Copilot reviewed 13 out of 13 changed files in this pull request and generated 10 comments.

Show a summary per file

File	Description
TxPool.cs	Adds SimpleRetryCache property and calls to track received transactions
SimpleRetryCache.cs	New class implementing retry logic with timeout-based request handling
NullTxPool.cs	Adds SimpleRetryCache property to null implementation
ITxPool.cs	Adds SimpleRetryCache property to interface
PriorityFeeTooLowFilter.cs	Changes minimum blob priority fee from 1 GWei to 0 GWei
Eth68ProtocolHandler.cs	Passes session ID to transaction requestor
Eth66ProtocolHandler.cs	Passes session ID to transaction requestor
Eth65ProtocolHandler.cs	Passes session ID to transaction requestor
PooledTxsRequestor.cs	Replaces pending hash cache with SimpleRetryCache integration
IPooledTxsRequestor.cs	Updates interface to include session ID parameter
Test files	Updates test calls to match new method signatures

[LukaszRozmej]

Should this be generalized to any p2p communication in a way? That we will re-request things from other peers if failed from one peer?

Probably conflicts with #9465 which I would prefer to go in first.

@marcindsobczak can you review?

[marcindsobczak]

Great work!

I see one issue with potential dead loop with requesting underpaid txs, described in a comment. Except of this - great PR!