mirror of
https://gitlab.com/pulsechaincom/lighthouse-pulse.git
synced 2024-12-25 13:07:18 +00:00
61ed5f0ec6
## Issue Addressed Closes https://github.com/sigp/lighthouse/issues/3270 ## Proposed Changes Optimize the calculation of historic beacon committees in the HTTP API. This is achieved by allowing committee caches to be constructed for historic epochs, and constructing these committee caches on the fly in the API. This is much faster than reconstructing the state at the requested epoch, which usually takes upwards of 20s, and sometimes minutes with SPRP=8192. The depth of the `randao_mixes` array allows us to look back 64K epochs/0.8 years from a single state, which is pretty awesome! We always use the `state_id` provided by the caller, but will return a nice 400 error if the epoch requested is out of range for the state requested, e.g. ```bash # Prater curl "http://localhost:5052/eth/v1/beacon/states/3170304/committees?epoch=33538" ``` ```json {"code":400,"message":"BAD_REQUEST: epoch out of bounds, try state at slot 1081344","stacktraces":[]} ``` Queries will be fastest when aligned to `slot % SPRP == 0`, so the hint suggests a slot that is 0 mod 8192. |
||
|---|---|---|
| .. | ||
| beacon_chain | ||
| builder_client | ||
| client | ||
| eth1 | ||
| execution_layer | ||
| genesis | ||
| http_api | ||
| http_metrics | ||
| lighthouse_network | ||
| network | ||
| operation_pool | ||
| src | ||
| store | ||
| tests | ||
| timer | ||
| Cargo.toml | ||