twoeths

we should call the below function once instead of 128 per block: - verify signature once bc it's the same for all DataColumnSidecar of the same block - verify inclusion...

PR in progress https://github.com/ChainSafe/lodestar/pull/7910

Using this ```typescript const keypair = createKeypairFromPeerId(conn.remotePeer); const nodeIdValue = nodeId(keypair.publicKey); const enr = network.discv5?.getKadValue(nodeIdValue); if (enr) { enrStr = enr.encodeTxt(); } ``` I can only find enr of 1...

in average we have 3.84 gossip messages per batch we rarely process single message, in average only 0.3 per slot

need to see if high rss/process_heap_bytes in #7930 is because of this branch MIN_DATA_COLUMN_SIDECAR_BATCH_SIZE = 64 seems the best option

need to investigate why it takes more time to become head than non-batch version

need to wait for #7936 and review metrics again

> need to investigate why it takes more time to become head than non-batch version that's an issue with the current implementation that we add to cache before validate so...

with `ckzg` since we always verify proofs synchronously, I imagine it'll save us ~500ms for data availability things changed with rust kzg, it always uses `async` version, I don't see...

this is confirmed in our metric (not tracked on grafana yet) on chain New Seen Attesters our New Seen Attesters of produced blocks (this is lido node 9)