メインコンテンツまでスキップ

Mina funnel

Configuration

The Mina funnel requires access to historical chain data which is acquired through an archive node. A connection to the database is required.

Mina:
type: mina
archiveConnectionString: "postgresql://postgres:postgres@localhost:5432/archive"
confirmationDepth: 15
delay: 3600
  • delay is in seconds.
  • confirmationDepth is in number of blocks. If not provided then the archive's node configuration is used (blocks with a chain_status of canonical).

Conceptually

This funnel has the following steps in the readData function:

  1. Fetch blocks & timestamps from the underlying funnel. This can be either a funnel for a single chain, like the block funnel, or it can be a collection of wrapped funnels involving multiple networks.
  2. Fetch the latest canonical Mina block timestamp. If confirmationDepth has a value in the configuration, this is done by getting the block at that particular height. Otherwise this is done by querying the blocks table to get the latest block which has canonical status. In this case the block confirmation parameters are defined by the archive's configuration.
  3. Query the database for events and actions in a certain timestamp range. The upper bound of this timestamp range is defined by the blocks fetched at step 1. The lower bound is the upper bound on the previous round.
  4. Merge the primitives with the underlying funnel.

Presync

The presync phase finishes after indexing all the blocks with a timestamp strictly lower than START_BLOCK_HEIGHT - delay. Each primitive is paginated individually, and progress is tracked in the cde_tracking_cursor_pagination table. The presync finishes when there are no more events in the range for any of the configured extensions.

Finalizing blocks

Since events parsed by the state transition are final, we need to be sure that we have processed all the Mina events that are associated with a certain block height before supplying those to the state transition. To guarantee this, the Mina funnel will wait for the timestamp of the latest canonical block to be greater or equal than the timestamp of the target block. This is necessary because otherwise there is no guarantee that the needed blocks are already propagated to the archive node, or in case there is a re-org.

Internally this is done by querying the blocks table for the latest block with a status of canonical.

Indexer Database

This funnel requires no change to the Mina archive node database schema. The following tables are used to retrieve the necessary information:

  • account_identifiers
  • accounts_accessed
  • blocks
  • blocks_zkapp_commands
  • zkapp_account_update
  • zkapp_account_update_body
  • zkapp_commands
  • zkapp_events
  • zkapp_field
  • zkapp_field_array

Usage is similar to the one used in Archive-Node-Api except for the following:

  • blocks are filtered by the timestamp column when needed.
  • zkapp_commands is filtered by hash to paginate the results during presync.
  • blocks is used to find the the latest finalized block, which is not a use-case supported by Archive-Node-Api.