Notice: This document is a work-in-progress for researchers and implementers.
This section extends the networking specification for Altair with additional messages, topics and data to the Req-Resp and Gossip domains.
Name | Value | Description |
---|---|---|
MAX_REQUEST_LIGHT_CLIENT_UPDATES |
2**7 (= 128) |
Maximum number of LightClientUpdate instances in a single request |
Gossip meshes are added to allow light clients to stay in sync with the network.
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 |
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 forwardedfinality_update
s, or it matches the highest previously forwarded slot and also has async_aggregate
indicating supermajority (> 2/3) sync committee participation while the previously forwardedfinality_update
for that slot did not indicate supermajority - [IGNORE] The
finality_update
is received after the block atsignature_slot
was given enough time to propagate through the network -- i.e. validate that one-third offinality_update.signature_slot
has transpired (SECONDS_PER_SLOT / INTERVALS_PER_SLOT
seconds after the start of the slot, with aMAXIMUM_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 increate_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 thatprocess_light_client_finality_update
does not indicate errors - [IGNORE] The
finality_update
advances thefinalized_header
of the localLightClientStore
-- i.e. validate that processingfinality_update
increasesstore.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)
:
fork_version |
Message SSZ type |
---|---|
GENESIS_FORK_VERSION |
n/a |
ALTAIR_FORK_VERSION and later |
altair.LightClientFinalityUpdate |
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 forwardedoptimistic_update
s - [IGNORE] The
optimistic_update
is received after the block atsignature_slot
was given enough time to propagate through the network -- i.e. validate that one-third ofoptimistic_update.signature_slot
has transpired (SECONDS_PER_SLOT / INTERVALS_PER_SLOT
seconds after the start of the slot, with aMAXIMUM_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 increate_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 thatprocess_light_client_optimistic_update
does not indicate errors - [IGNORE] The
optimistic_update
either matches corresponding fields of the most recently forwardedLightClientFinalityUpdate
(if any), or it advances theoptimistic_header
of the localLightClientStore
-- i.e. validate that processingoptimistic_update
increasesstore.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)
:
fork_version |
Message SSZ type |
---|---|
GENESIS_FORK_VERSION |
n/a |
ALTAIR_FORK_VERSION and later |
altair.LightClientOptimisticUpdate |
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
. 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)
:
fork_version |
Response SSZ type |
---|---|
GENESIS_FORK_VERSION |
n/a |
ALTAIR_FORK_VERSION and later |
altair.LightClientBootstrap |
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
. 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)
:
fork_version |
Response chunk SSZ type |
---|---|
GENESIS_FORK_VERSION |
n/a |
ALTAIR_FORK_VERSION and later |
altair.LightClientUpdate |
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
.
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)
:
fork_version |
Response SSZ type |
---|---|
GENESIS_FORK_VERSION |
n/a |
ALTAIR_FORK_VERSION and later |
altair.LightClientFinalityUpdate |
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
.
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)
:
fork_version |
Response SSZ type |
---|---|
GENESIS_FORK_VERSION |
n/a |
ALTAIR_FORK_VERSION and later |
altair.LightClientOptimisticUpdate |
Light clients using libp2p to stay in sync with the network SHOULD subscribe to the light_client_finality_update
and light_client_optimistic_update
pubsub topics and validate all received messages while the 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, and MAY also update the contents of their 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.
This section extends the honest validator specification with additional responsibilities to enable light clients to sync with the network.
All full nodes SHOULD subscribe to and provide stability on the light_client_finality_update
and light_client_optimistic_update
pubsub topics by validating all received messages.
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, aLightClientFinalityUpdate
SHOULD be broadcasted to the pubsub topiclight_client_finality_update
if no matching message has not yet been forwarded as part of gossip validation. - If
attested_header.beacon.slot
increased, aLightClientOptimisticUpdate
SHOULD be broadcasted to the pubsub topiclight_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.