PeerDAS - huge gossip job wait time

[twoeths]

Describe the bug

this is mornitored on peerdas devnet

Expected behavior

even on a big network like a mainnet node subscribing to all subnets, job wait time for all topics should be <2s

Steps to reproduce

No response

Additional context

No response

Operating system

Linux

Lodestar version or commit hash

peerDAS

[twoeths]

P7 (broken IWANT promises) are so big that cause the whole gossip score to be very low

[twoeths]

this is due to isProcessingCurrentSlotBlock flag we put in network processor

lodestar/packages/beacon-node/src/network/processor/index.ts

Line 273 in 83fd80d

this.isProcessingCurrentSlotBlock = true;

without it, all messages are processed immediately

the original flag was for pre-deneb and for beacon_block only which we receive exactly 1 message per slot
applying for blobs was a mistake because we have up to 6/9 messages per slot, and it causes blobs not being able to process in parallel
applying it for data_column_sidecar causes very big job wait time as we can see in this issue

I would remove this flag because after deneb, even we validate a beacon_block message, we still need to wait for all blobs. Also our attestation validation flow was improved after electra

[philknows]

Can we close via #7848 or is this waiting to make it on peerDAS branch before we close it?

[twoeths]

works fine with peerDAS_devnet_7 branch
the low gossip score was isolated in #7849