Skip to main content

Data availability

A settlement batch can commit to its data through the batch's dataHash. The data-availability (DA) backend decides where that data is made available so anyone can reconstruct and check the rollup's state. The RDK supports three settings.

BackendStatusNotes
nativeThe default for every profileNo blob can be posted yet: leave dataHash empty
celestiaPlanned (not yet active)Selectable in configs; not served by the network yet
bothPlanned (depends on Celestia)Redundant: native + Celestia

native​

Every rollup profile creates a native rollup. The chain has on-chain blob storage, but no transaction posts a blob to it yet. Because of that:

  • Leave dataHash empty. On a native (or both) rollup the chain accepts a non-empty data_hash only when a matching blob is stored, so a batch that claims one is refused as data unavailable, and the fee is spent. The batch's state roots, proof and anchor still bind it.
  • RdkTxClient.submitBatch refuses such a batch before broadcasting when it can read the rollup (a hybrid client with a REST endpoint), and dataHashError applies the same rule anywhere.
  • Keep your batch data available off chain yourself, for example next to your sequencer, so others can reconstruct your state.
import { dataHashError } from "@qorechain/rdk";

await tx.submitBatch({ rollupId, batchIndex, stateRoot, prevStateRoot, txCount });
// no dataHash: dataHashError("native", "0xab...") explains why

buildDaBlob still computes a size-checked commitment for your own records, the same SHAKE-256 (32 bytes) the chain's blob store uses (releases before 0.7.0 used SHA-256); it just must not go into a batch on a native rollup until blob posting exists. The limits it checks, maxDaBlobSize (default 2097152 bytes) and blobRetentionBlocks (default 432000), are the chain's parameters for that future storage; read them live from rdk.params().

celestia (planned, not yet active)​

Celestia DA is planned but not yet active on the network. A configuration may target celestia, but the network does not serve it yet. The RDK fails gracefully: configs that select celestia validate with a non-fatal warning, and attempting to use a non-native backend for live submission raises a clear error rather than failing silently.

import { assertDaBackendAvailable } from "@qorechain/rdk";

// "native" passes; "celestia" / "both" throw a clear, descriptive error today.
assertDaBackendAvailable("native");

Guard live submission paths with assertDaBackendAvailable so you get an explicit message if you point at a backend that is not yet active.

both (planned)​

The both setting writes redundantly to native and Celestia. Because it depends on Celestia, it is likewise planned and not yet active. For live rollups today, use native.

Choosing a backend​

Use native for anything you intend to run on the network now. Select celestia or both only when you are deliberately preparing a configuration for a future in which the Celestia backend is active — and keep the assertDaBackendAvailable guard in your submission path so it fails loudly until then.