consensus-specs/specs/altair/light-client/p2p-interface.md
Hsiao-Wei Wang 2b2ffb5229
Merge pull request #3190 from etan-status/lc-headerwrapper
Add `LightClientHeader` wrapper
2023-01-13 00:40:44 +08:00

280 lines
15 KiB
Markdown

# Altair Light Client -- Networking
**Notice**: This document is a work-in-progress for researchers and implementers.
## Table of contents
<!-- TOC -->
<!-- START doctoc generated TOC please keep comment here to allow auto update -->
<!-- DON'T EDIT THIS SECTION, INSTEAD RE-RUN doctoc TO UPDATE -->
- [Networking](#networking)
- [Configuration](#configuration)
- [The gossip domain: gossipsub](#the-gossip-domain-gossipsub)
- [Topics and messages](#topics-and-messages)
- [Global topics](#global-topics)
- [`light_client_finality_update`](#light_client_finality_update)
- [`light_client_optimistic_update`](#light_client_optimistic_update)
- [The Req/Resp domain](#the-reqresp-domain)
- [Messages](#messages)
- [GetLightClientBootstrap](#getlightclientbootstrap)
- [LightClientUpdatesByRange](#lightclientupdatesbyrange)
- [GetLightClientFinalityUpdate](#getlightclientfinalityupdate)
- [GetLightClientOptimisticUpdate](#getlightclientoptimisticupdate)
- [Light clients](#light-clients)
- [Validator assignments](#validator-assignments)
- [Beacon chain responsibilities](#beacon-chain-responsibilities)
- [Sync committee](#sync-committee)
<!-- END doctoc generated TOC please keep comment here to allow auto update -->
<!-- /TOC -->
## Networking
This section extends the [networking specification for Altair](../p2p-interface.md) with additional messages, topics and data to the Req-Resp and Gossip domains.
### Configuration
| Name | Value | Description |
| - | - | - |
| `MAX_REQUEST_LIGHT_CLIENT_UPDATES` | `2**7` (= 128) | Maximum number of `LightClientUpdate` instances in a single request |
### The gossip domain: gossipsub
Gossip meshes are added to allow light clients to stay in sync with the network.
#### Topics and messages
New global topics are added to provide light clients with the latest updates.
| name | Message Type |
| - | - |
| `light_client_finality_update` | `LightClientFinalityUpdate` |
| `light_client_optimistic_update` | `LightClientOptimisticUpdate` |
##### Global topics
###### `light_client_finality_update`
This topic is used to propagate the latest `LightClientFinalityUpdate` to light clients, allowing them to keep track of the latest `finalized_header`.
The following validations MUST pass before forwarding the `finality_update` on the network.
- _[IGNORE]_ The `finalized_header.beacon.slot` is greater than that of all previously forwarded `finality_update`s
- _[IGNORE]_ The `finality_update` is received after the block at `signature_slot` was given enough time to propagate through the network -- i.e. validate that one-third of `finality_update.signature_slot` has transpired (`SECONDS_PER_SLOT / INTERVALS_PER_SLOT` seconds after the start of the slot, with a `MAXIMUM_GOSSIP_CLOCK_DISPARITY` allowance)
For full nodes, the following validations MUST additionally pass before forwarding the `finality_update` on the network.
- _[IGNORE]_ The received `finality_update` matches the locally computed one exactly (as defined in [`create_light_client_finality_update`](./full-node.md#create_light_client_finality_update))
For light clients, the following validations MUST additionally pass before forwarding the `finality_update` on the network.
- _[REJECT]_ The `finality_update` is valid -- i.e. validate that `process_light_client_finality_update` does not indicate errors
- _[IGNORE]_ The `finality_update` advances the `finalized_header` of the local `LightClientStore` -- i.e. validate that processing `finality_update` increases `store.finalized_header.beacon.slot`
Light clients SHOULD call `process_light_client_finality_update` even if the message is ignored.
The gossip `ForkDigest`-context is determined based on `compute_fork_version(compute_epoch_at_slot(finality_update.attested_header.beacon.slot))`.
Per `context = compute_fork_digest(fork_version, genesis_validators_root)`:
[0]: # (eth2spec: skip)
| `fork_version` | Message SSZ type |
| ------------------------------- | ------------------------------------ |
| `GENESIS_FORK_VERSION` | n/a |
| `ALTAIR_FORK_VERSION` and later | `altair.LightClientFinalityUpdate` |
###### `light_client_optimistic_update`
This topic is used to propagate the latest `LightClientOptimisticUpdate` to light clients, allowing them to keep track of the latest `optimistic_header`.
The following validations MUST pass before forwarding the `optimistic_update` on the network.
- _[IGNORE]_ The `attested_header.beacon.slot` is greater than that of all previously forwarded `optimistic_update`s
- _[IGNORE]_ The `optimistic_update` is received after the block at `signature_slot` was given enough time to propagate through the network -- i.e. validate that one-third of `optimistic_update.signature_slot` has transpired (`SECONDS_PER_SLOT / INTERVALS_PER_SLOT` seconds after the start of the slot, with a `MAXIMUM_GOSSIP_CLOCK_DISPARITY` allowance)
For full nodes, the following validations MUST additionally pass before forwarding the `optimistic_update` on the network.
- _[IGNORE]_ The received `optimistic_update` matches the locally computed one exactly (as defined in [`create_light_client_optimistic_update`](./full-node.md#create_light_client_optimistic_update))
For light clients, the following validations MUST additionally pass before forwarding the `optimistic_update` on the network.
- _[REJECT]_ The `optimistic_update` is valid -- i.e. validate that `process_light_client_optimistic_update` does not indicate errors
- _[IGNORE]_ The `optimistic_update` either matches corresponding fields of the most recently forwarded `LightClientFinalityUpdate` (if any), or it advances the `optimistic_header` of the local `LightClientStore` -- i.e. validate that processing `optimistic_update` increases `store.optimistic_header.beacon.slot`
Light clients SHOULD call `process_light_client_optimistic_update` even if the message is ignored.
The gossip `ForkDigest`-context is determined based on `compute_fork_version(compute_epoch_at_slot(optimistic_update.attested_header.beacon.slot))`.
Per `context = compute_fork_digest(fork_version, genesis_validators_root)`:
[0]: # (eth2spec: skip)
| `fork_version` | Message SSZ type |
| ------------------------------- | ------------------------------------ |
| `GENESIS_FORK_VERSION` | n/a |
| `ALTAIR_FORK_VERSION` and later | `altair.LightClientOptimisticUpdate` |
### The Req/Resp domain
#### Messages
##### GetLightClientBootstrap
**Protocol ID:** `/eth2/beacon_chain/req/light_client_bootstrap/1/`
Request Content:
```
(
Root
)
```
Response Content:
```
(
LightClientBootstrap
)
```
Requests the `LightClientBootstrap` structure corresponding to a given post-Altair beacon block root.
The request MUST be encoded as an SSZ-field.
Peers SHOULD provide results as defined in [`create_light_client_bootstrap`](./full-node.md#create_light_client_bootstrap). To fulfill a request, the requested block and its post state need to be known.
When a `LightClientBootstrap` instance cannot be produced for a given block root, peers SHOULD respond with error code `3: ResourceUnavailable`.
A `ForkDigest`-context based on `compute_fork_version(compute_epoch_at_slot(bootstrap.header.beacon.slot))` is used to select the fork namespace of the Response type.
Per `context = compute_fork_digest(fork_version, genesis_validators_root)`:
[0]: # (eth2spec: skip)
| `fork_version` | Response SSZ type |
| ------------------------------- | ------------------------------------ |
| `GENESIS_FORK_VERSION` | n/a |
| `ALTAIR_FORK_VERSION` and later | `altair.LightClientBootstrap` |
##### LightClientUpdatesByRange
**Protocol ID:** `/eth2/beacon_chain/req/light_client_updates_by_range/1/`
Request Content:
```
(
start_period: uint64
count: uint64
)
```
Response Content:
```
(
List[LightClientUpdate, MAX_REQUEST_LIGHT_CLIENT_UPDATES]
)
```
Requests the `LightClientUpdate` instances in the sync committee period range `[start_period, start_period + count)`, leading up to the current head sync committee period as selected by fork choice.
The request MUST be encoded as an SSZ-container.
The response MUST consist of zero or more `response_chunk`. Each _successful_ `response_chunk` MUST contain a single `LightClientUpdate` payload.
Peers SHOULD provide results as defined in [`create_light_client_update`](./full-node.md#create_light_client_update). They MUST respond with at least the earliest known result within the requested range, and MUST send results in consecutive order (by period). The response MUST NOT contain more than `min(MAX_REQUEST_LIGHT_CLIENT_UPDATES, count)` results.
For each `response_chunk`, a `ForkDigest`-context based on `compute_fork_version(compute_epoch_at_slot(update.attested_header.beacon.slot))` is used to select the fork namespace of the Response type. Note that this `fork_version` may be different from the one used to verify the `update.sync_aggregate`, which is based on `update.signature_slot`.
Per `context = compute_fork_digest(fork_version, genesis_validators_root)`:
[0]: # (eth2spec: skip)
| `fork_version` | Response chunk SSZ type |
| ------------------------------- | ------------------------------------ |
| `GENESIS_FORK_VERSION` | n/a |
| `ALTAIR_FORK_VERSION` and later | `altair.LightClientUpdate` |
##### GetLightClientFinalityUpdate
**Protocol ID:** `/eth2/beacon_chain/req/light_client_finality_update/1/`
No Request Content.
Response Content:
```
(
LightClientFinalityUpdate
)
```
Requests the latest `LightClientFinalityUpdate` known by a peer.
Peers SHOULD provide results as defined in [`create_light_client_finality_update`](./full-node.md#create_light_client_finality_update).
When no `LightClientFinalityUpdate` is available, peers SHOULD respond with error code `3: ResourceUnavailable`.
A `ForkDigest`-context based on `compute_fork_version(compute_epoch_at_slot(finality_update.attested_header.beacon.slot))` is used to select the fork namespace of the Response type. Note that this `fork_version` may be different from the one used to verify the `finality_update.sync_aggregate`, which is based on `finality_update.signature_slot`.
Per `context = compute_fork_digest(fork_version, genesis_validators_root)`:
[0]: # (eth2spec: skip)
| `fork_version` | Response SSZ type |
| ------------------------------- | ------------------------------------ |
| `GENESIS_FORK_VERSION` | n/a |
| `ALTAIR_FORK_VERSION` and later | `altair.LightClientFinalityUpdate` |
##### GetLightClientOptimisticUpdate
**Protocol ID:** `/eth2/beacon_chain/req/light_client_optimistic_update/1/`
No Request Content.
Response Content:
```
(
LightClientOptimisticUpdate
)
```
Requests the latest `LightClientOptimisticUpdate` known by a peer.
Peers SHOULD provide results as defined in [`create_light_client_optimistic_update`](./full-node.md#create_light_client_optimistic_update).
When no `LightClientOptimisticUpdate` is available, peers SHOULD respond with error code `3: ResourceUnavailable`.
A `ForkDigest`-context based on `compute_fork_version(compute_epoch_at_slot(optimistic_update.attested_header.beacon.slot))` is used to select the fork namespace of the Response type. Note that this `fork_version` may be different from the one used to verify the `optimistic_update.sync_aggregate`, which is based on `optimistic_update.signature_slot`.
Per `context = compute_fork_digest(fork_version, genesis_validators_root)`:
[0]: # (eth2spec: skip)
| `fork_version` | Response SSZ type |
| ------------------------------- | ------------------------------------ |
| `GENESIS_FORK_VERSION` | n/a |
| `ALTAIR_FORK_VERSION` and later | `altair.LightClientOptimisticUpdate` |
## Light clients
Light clients using libp2p to stay in sync with the network SHOULD subscribe to the [`light_client_finality_update`](#light_client_finality_update) and [`light_client_optimistic_update`](#light_client_optimistic_update) pubsub topics and validate all received messages while the [light client sync process](./light-client.md#light-client-sync-process) supports processing `LightClientFinalityUpdate` and `LightClientOptimisticUpdate` structures.
Light clients MAY also collect historic light client data and make it available to other peers. If they do, they SHOULD advertise supported message endpoints in [the Req/Resp domain](#the-reqresp-domain), and MAY also update the contents of their [`Status`](../../phase0/p2p-interface.md#status) message to reflect the locally available light client data.
If only limited light client data is locally available, the light client SHOULD use data based on `genesis_block` and `GENESIS_SLOT` in its `Status` message. Hybrid peers that also implement full node functionality MUST only incorporate data based on their full node sync progress into their `Status` message.
## Validator assignments
This section extends the [honest validator specification](../validator.md) with additional responsibilities to enable light clients to sync with the network.
### Beacon chain responsibilities
All full nodes SHOULD subscribe to and provide stability on the [`light_client_finality_update`](#light_client_finality_update) and [`light_client_optimistic_update`](#light_client_optimistic_update) pubsub topics by validating all received messages.
### Sync committee
Whenever fork choice selects a new head block with a sync aggregate participation `>= MIN_SYNC_COMMITTEE_PARTICIPANTS` and a post-Altair parent block, full nodes with at least one validator assigned to the current sync committee at the block's `slot` SHOULD broadcast derived light client data as follows:
- If `finalized_header.beacon.slot` increased, a `LightClientFinalityUpdate` SHOULD be broadcasted to the pubsub topic `light_client_finality_update` if no matching message has not yet been forwarded as part of gossip validation.
- If `attested_header.beacon.slot` increased, a `LightClientOptimisticUpdate` SHOULD be broadcasted to the pubsub topic `light_client_optimistic_update` if no matching message has not yet been forwarded as part of gossip validation.
These messages SHOULD be broadcasted after one-third of `slot` has transpired (`SECONDS_PER_SLOT / INTERVALS_PER_SLOT` seconds after the start of the slot). To ensure that the corresponding block was given enough time to propagate through the network, they SHOULD NOT be sent earlier. Note that this is different from how other messages are handled, e.g., attestations, which may be sent early.