After upgrading ETH v1.16.4 → v1.16.5, Arbitrum node fails to fetch blobs;

[553589912]

Description

After upgrading the ETH client from v1.16.4 to v1.16.5, the Arbitrum node started logging the following warnings:

arb-node | WARN [10-17|10:16:02.647] error reading inbox err="failed to get blobs: error fetching blobs in 23595470 l1 block: beacon client in blobSidecars got error or empty response fetching non-expired blobs in slot: 12821418, if using a prysm endpoint, try --enable-experimental-backfill flag, err: response returned with status 500 Internal Server Error, want 200 OK. body: on overflows","code":500}"

Adding the --enable-experimental-backfill flag to Prysm does not resolve the issue, and the Arbitrum node is unable to fetch historical L1 blobSidecars.

[553589912]

https://blog.ethereum.org/2025/10/15/fusaka-blob-update

[arowana-official]

@553589912 Check which specific error message that L1 beacon node ( not Geth or execution node ) returns using curl.

It is likely that your beacon node doesn't have full blobs for a blob due to how PeerDAS works and you should likely add something like --supernode option as a startup flag ( So that your L1 Beacon Node would have all blob required for sync & blob endpoint )

See https://github.com/sigp/lighthouse/releases/tag/v8.0.0-rc.1 as well.

[553589912]

@553589912 Check which specific error message that L1 beacon node ( not Geth or execution node ) returns using curl.

It is likely that your beacon node doesn't have full blobs for a blob due to how PeerDAS works and you should likely add something like --supernode option as a startup flag ( So that your L1 Beacon Node would have all blob required for sync & blob endpoint )

See https://github.com/sigp/lighthouse/releases/tag/v8.0.0-rc.1 as well.

Hi, Prysm doesn’t have the --supernode flag. Should I use --subscribe-all-data-subnets instead ?

[Tristan-Wilson]

This is probably a prysm bug, this is a known issue in prysm 6.1.2 when used with mainnet because the fusaka fork time is not set. I suggest downgrading to 6.1.1 in the meantime. 6.1.3 will fix this once it's released.

[Tristan-Wilson]

Prysm fix: OffchainLabs/prysm#15867

[Tristan-Wilson]

Nitro PR to improve the logging #3881

[Tristan-Wilson]

I'm fairly sure the prysm 6.1.2 issue is the cause because the truncated message "on overflows" matches with the end of the full error message "Internal server error: could not calculate Fulu start slot: start slot calculation overflows: multiplication overflows", so I'm going to close this. Feel free to reopen if you think this is wrong.