Alchemy Subgraphs

Overview

Alchemy Subgraphs delivers managed GraphQL indexing with seamless compatibility with The Graph tooling. A dedicated Sei pipeline (v6.1.11+ ) adds:

• Native Sei EVM support with < 2 block lag under steady load.
• Automatic retries for synthetic transaction logs.
• Managed backfills with configurable block windows (default 2,000).
• Integrated monitoring dashboards (error rates, sync status, latency percentiles).

Quickstart

1. Create an Alchemy account  and request access to the Sei Subgraphs beta.
2. Install the CLI: npm install -g @graphprotocol/graph-cli (compatible with Alchemy’s gateway).
3. Initialise your subgraph: graph init --from-example sei/my-dapp.
4. Update the network configuration:

# subgraph.yaml excerpt
dataSources:
  - kind: ethereum/contract
    network: sei-mainnet
    source:
      address: '0x...'
      startBlock: 12000000
    mapping:
      apiVersion: 0.0.9
      language: wasm/assemblyscript

5. Authenticate: graph auth --node https://subgraphs.sei.alchemy.com/api <API_KEY>.
6. Deploy: graph deploy --node https://subgraphs.sei.alchemy.com/api sei/my-dapp.

Windows & Limits

• Default backfill window: 2,000 blocks (adaptive). Use CLI flag --block-window to customise.
• Daily compute quota: 5 hours build time on the free tier; contact support for dedicated workers.
• Concurrent backfills per project: 2.
• Reorg depth assumption: 0 (Sei finality). Payloads auto-commit once ingested; manual rewinds available via dashboard.
• Synthetic event handling: automatically tagged via synthetic=true metadata.

Backfill Strategy

• Stage subgraph versions: keep production indexing current data while backfill version replays history.
• Enable automatic snapshotting every 250k blocks to accelerate restarts.
• Use the CLI graph build locally with --ipfs /tmp/ipfs to catch ABI/schema errors before deploying.
• Pair backfills with the Indexing Best Practices guidance for retry behaviour.

Reorg Handling

Sei blocks do not revert, but Alchemy maintains a virtual reorg buffer for consistency:

• Recent blocks (last 20) stay in a mutable buffer for 5 minutes; this allows manual rewinds if upstream data is corrected.
• The dashboard shows Finalized vs Buffered heights; monitor and alert if the buffer grows unexpectedly (indicates provider lag).
• If a rewind is triggered (rare), redeploy the subgraph to the prior snapshot via the console.

Build Pipeline Snapshot

Implementation References

Troubleshooting