mirror of
https://github.com/0glabs/0g-storage-node.git
synced 2024-12-24 23:35:18 +00:00
Sync file from neighbors by default (#295)
Some checks are pending
abi-consistent-check / build-and-compare (push) Waiting to run
code-coverage / unittest-cov (push) Waiting to run
rust / check (push) Waiting to run
rust / test (push) Waiting to run
rust / lints (push) Waiting to run
functional-test / test (push) Waiting to run
Some checks are pending
abi-consistent-check / build-and-compare (push) Waiting to run
code-coverage / unittest-cov (push) Waiting to run
rust / check (push) Waiting to run
rust / test (push) Waiting to run
rust / lints (push) Waiting to run
functional-test / test (push) Waiting to run
* sync from neighbors by default * change p2p protocol version * minor fix for crash python test * Fix pubsub message id issue * fix python tests
This commit is contained in:
parent
6b2420cac4
commit
d3a2118985
@ -390,7 +390,7 @@ impl<AppReqId: ReqId> Behaviour<AppReqId> {
|
||||
.gossipsub
|
||||
.publish(topic.clone().into(), message_data.clone())
|
||||
{
|
||||
warn!(error = ?e, "Could not publish message");
|
||||
warn!(error = ?e, topic = ?topic.kind(), "Failed to publish message");
|
||||
|
||||
// add to metrics
|
||||
if let Some(v) = metrics::get_int_gauge(
|
||||
|
@ -294,32 +294,6 @@ impl From<u8> for NetworkLoad {
|
||||
|
||||
/// Return a Lighthouse specific `GossipsubConfig` where the `message_id_fn` depends on the current fork.
|
||||
pub fn gossipsub_config(network_load: u8) -> GossipsubConfig {
|
||||
// The function used to generate a gossipsub message id
|
||||
// We use the first 8 bytes of SHA256(data) for content addressing
|
||||
let fast_gossip_message_id =
|
||||
|message: &RawGossipsubMessage| FastMessageId::from(&Sha256::digest(&message.data)[..8]);
|
||||
fn prefix(prefix: [u8; 4], message: &GossipsubMessage) -> Vec<u8> {
|
||||
let topic_bytes = message.topic.as_str().as_bytes();
|
||||
|
||||
// according to: https://github.com/ethereum/consensus-specs/blob/dev/specs/merge/p2p-interface.md#the-gossip-domain-gossipsub
|
||||
// the derivation of the message-id remains the same in the merge
|
||||
let topic_len_bytes = topic_bytes.len().to_le_bytes();
|
||||
let mut vec = Vec::with_capacity(
|
||||
prefix.len() + topic_len_bytes.len() + topic_bytes.len() + message.data.len(),
|
||||
);
|
||||
vec.extend_from_slice(&prefix);
|
||||
vec.extend_from_slice(&topic_len_bytes);
|
||||
vec.extend_from_slice(topic_bytes);
|
||||
vec.extend_from_slice(&message.data);
|
||||
vec
|
||||
}
|
||||
|
||||
let gossip_message_id = move |message: &GossipsubMessage| {
|
||||
MessageId::from(
|
||||
&Sha256::digest(prefix(MESSAGE_DOMAIN_VALID_SNAPPY, message).as_slice())[..20],
|
||||
)
|
||||
};
|
||||
|
||||
let load = NetworkLoad::from(network_load);
|
||||
|
||||
GossipsubConfigBuilder::default()
|
||||
@ -368,3 +342,41 @@ fn is_global(addr: &std::net::Ipv4Addr) -> bool {
|
||||
// Make sure the address is not in 0.0.0.0/8
|
||||
&& addr.octets()[0] != 0
|
||||
}
|
||||
|
||||
fn fast_gossip_message_id(message: &RawGossipsubMessage) -> FastMessageId {
|
||||
// source | topic | data | nonce
|
||||
let topic_bytes = message.topic.as_str().as_bytes();
|
||||
let mut buf = Vec::with_capacity(64 + topic_bytes.len() + message.data.len() + 8);
|
||||
|
||||
if let Some(peer_id) = message.source {
|
||||
buf.extend_from_slice(&peer_id.to_bytes());
|
||||
}
|
||||
|
||||
buf.extend_from_slice(&topic_bytes);
|
||||
buf.extend_from_slice(&message.data);
|
||||
|
||||
if let Some(nonce) = message.sequence_number {
|
||||
buf.extend_from_slice(&nonce.to_le_bytes());
|
||||
}
|
||||
|
||||
FastMessageId::from(&Sha256::digest(&buf)[..8])
|
||||
}
|
||||
|
||||
fn gossip_message_id(message: &GossipsubMessage) -> MessageId {
|
||||
// prefix | source | topic | data | nonce
|
||||
let topic_bytes = message.topic.as_str().as_bytes();
|
||||
let mut vec = Vec::with_capacity(
|
||||
MESSAGE_DOMAIN_VALID_SNAPPY.len() + 64 + topic_bytes.len() + message.data.len() + 8,
|
||||
);
|
||||
vec.extend_from_slice(&MESSAGE_DOMAIN_VALID_SNAPPY);
|
||||
if let Some(peer_id) = message.source {
|
||||
vec.extend_from_slice(&peer_id.to_bytes());
|
||||
}
|
||||
vec.extend_from_slice(topic_bytes);
|
||||
vec.extend_from_slice(&message.data);
|
||||
if let Some(nonce) = message.sequence_number {
|
||||
vec.extend_from_slice(&nonce.to_le_bytes());
|
||||
}
|
||||
|
||||
MessageId::from(&Sha256::digest(&vec)[..20])
|
||||
}
|
||||
|
@ -97,9 +97,11 @@ pub use service::{load_private_key, Context, Libp2pEvent, Service, NETWORK_KEY_F
|
||||
/// - v1: Broadcast FindFile & AnnounceFile messages in the whole network, which caused network too heavey.
|
||||
/// - v2: Publish NewFile to neighbors only and announce file via RPC message.
|
||||
/// - v3: Add shard config in Status message.
|
||||
/// - v4: Refactor pubsub messages.
|
||||
pub const PROTOCOL_VERSION_V1: [u8; 3] = [0, 1, 1];
|
||||
pub const PROTOCOL_VERSION_V2: [u8; 3] = [0, 2, 1];
|
||||
pub const PROTOCOL_VERSION_V3: [u8; 3] = [0, 3, 0];
|
||||
pub const PROTOCOL_VERSION_V4: [u8; 3] = [0, 4, 0];
|
||||
|
||||
/// Application level requests sent to the network.
|
||||
#[derive(Debug, Clone, Copy)]
|
||||
|
@ -343,6 +343,7 @@ impl RouterService {
|
||||
let msg = PubsubMessage::NewFile(new_file.into());
|
||||
self.libp2p.swarm.behaviour_mut().publish(vec![msg]);
|
||||
metrics::SERVICE_ROUTE_NETWORK_MESSAGE_ANNOUNCE_LOCAL_FILE.mark(1);
|
||||
debug!(?new_file, "Publish NewFile message");
|
||||
}
|
||||
NetworkMessage::UPnPMappingEstablished {
|
||||
tcp_socket,
|
||||
|
@ -43,13 +43,9 @@ impl ZgsConfig {
|
||||
chain_id,
|
||||
flow_address,
|
||||
p2p_protocol_version: ProtocolVersion {
|
||||
major: network::PROTOCOL_VERSION_V3[0],
|
||||
minor: network::PROTOCOL_VERSION_V3[1],
|
||||
build: if self.sync.neighbors_only {
|
||||
network::PROTOCOL_VERSION_V3[2] + 1
|
||||
} else {
|
||||
network::PROTOCOL_VERSION_V3[2]
|
||||
},
|
||||
major: network::PROTOCOL_VERSION_V4[0],
|
||||
minor: network::PROTOCOL_VERSION_V4[1],
|
||||
build: network::PROTOCOL_VERSION_V4[2],
|
||||
},
|
||||
};
|
||||
network_config.network_id = local_network_id.clone();
|
||||
|
@ -677,7 +677,7 @@ impl SerialSyncController {
|
||||
} else {
|
||||
// FindFile timeout
|
||||
if since.elapsed() >= self.config.peer_find_timeout {
|
||||
if self.config.neighbors_only {
|
||||
if self.goal.is_all_chunks() && self.config.neighbors_only {
|
||||
self.state = SyncState::Failed {
|
||||
reason: FailureReason::TimeoutFindFile,
|
||||
};
|
||||
@ -1694,7 +1694,10 @@ mod tests {
|
||||
let file_location_cache = create_file_location_cache(peer_id, vec![tx_id]);
|
||||
|
||||
let controller = SerialSyncController::new(
|
||||
Config::default(),
|
||||
Config {
|
||||
neighbors_only: false,
|
||||
..Default::default()
|
||||
},
|
||||
tx_id,
|
||||
0,
|
||||
FileSyncGoal::new_file(num_chunks as u64),
|
||||
|
@ -68,7 +68,7 @@ impl Default for Config {
|
||||
fn default() -> Self {
|
||||
Self {
|
||||
// sync service config
|
||||
neighbors_only: false,
|
||||
neighbors_only: true,
|
||||
heartbeat_interval: Duration::from_secs(5),
|
||||
auto_sync_enabled: false,
|
||||
max_sync_files: 8,
|
||||
|
@ -941,7 +941,13 @@ mod tests {
|
||||
}
|
||||
|
||||
async fn spawn_sync_service(&mut self, with_peer_store: bool) -> SyncSender {
|
||||
self.spawn_sync_service_with_config(with_peer_store, Config::default())
|
||||
self.spawn_sync_service_with_config(
|
||||
with_peer_store,
|
||||
Config {
|
||||
neighbors_only: false,
|
||||
..Default::default()
|
||||
},
|
||||
)
|
||||
.await
|
||||
}
|
||||
|
||||
|
@ -232,10 +232,6 @@ batcher_announcement_capacity = 100
|
||||
# all files, and sufficient disk space is required.
|
||||
auto_sync_enabled = true
|
||||
|
||||
# Indicates whether to sync file from neighbor nodes only. This is to avoid flooding file
|
||||
# announcements in the whole network, which leads to high latency or even timeout to sync files.
|
||||
neighbors_only = true
|
||||
|
||||
# Maximum number of files in sync from other peers simultaneously.
|
||||
# max_sync_files = 8
|
||||
|
||||
|
@ -244,10 +244,6 @@ batcher_announcement_capacity = 100
|
||||
# all files, and sufficient disk space is required.
|
||||
auto_sync_enabled = true
|
||||
|
||||
# Indicates whether to sync file from neighbor nodes only. This is to avoid flooding file
|
||||
# announcements in the whole network, which leads to high latency or even timeout to sync files.
|
||||
neighbors_only = true
|
||||
|
||||
# Maximum number of files in sync from other peers simultaneously.
|
||||
# max_sync_files = 8
|
||||
|
||||
|
@ -246,10 +246,6 @@
|
||||
# all files, and sufficient disk space is required.
|
||||
# auto_sync_enabled = false
|
||||
|
||||
# Indicates whether to sync file from neighbor nodes only. This is to avoid flooding file
|
||||
# announcements in the whole network, which leads to high latency or even timeout to sync files.
|
||||
neighbors_only = true
|
||||
|
||||
# Maximum number of files in sync from other peers simultaneously.
|
||||
# max_sync_files = 8
|
||||
|
||||
|
@ -20,8 +20,9 @@ class CrashTest(TestFramework):
|
||||
|
||||
segment = submit_data(self.nodes[0], chunk_data)
|
||||
self.log.info("segment: %s", segment)
|
||||
wait_until(lambda: self.nodes[0].zgs_get_file_info(data_root)["finalized"] is True)
|
||||
|
||||
for i in range(self.num_nodes):
|
||||
for i in range(1, self.num_nodes):
|
||||
wait_until(
|
||||
lambda: self.nodes[i].zgs_get_file_info(data_root) is not None
|
||||
)
|
||||
|
@ -14,7 +14,6 @@ class AutoRandomSyncV2Test(TestFramework):
|
||||
"auto_sync_enabled": True,
|
||||
"max_sequential_workers": 0,
|
||||
"max_random_workers": 3,
|
||||
"neighbors_only": True,
|
||||
}
|
||||
}
|
||||
|
||||
|
@ -45,13 +45,15 @@ class SyncTest(TestFramework):
|
||||
wait_until(lambda: client2.zgs_get_file_info(data_root) is not None)
|
||||
time.sleep(3)
|
||||
assert_equal(client2.zgs_get_file_info(data_root)["finalized"], False)
|
||||
assert(client2.admin_get_file_location(0) is None)
|
||||
# file sync use ASK_FILE & ANSWER FILE protocol, and do not cache file announcement anymore.
|
||||
# assert(client2.admin_get_file_location(0) is None)
|
||||
|
||||
# Trigger file sync by rpc
|
||||
assert(client2.admin_start_sync_file(0) is None)
|
||||
wait_until(lambda: client2.sync_status_is_completed_or_unknown(0))
|
||||
wait_until(lambda: client2.zgs_get_file_info(data_root)["finalized"])
|
||||
assert(client2.admin_get_file_location(0) is not None)
|
||||
# file sync use ASK_FILE & ANSWER FILE protocol, and do not cache file announcement anymore.
|
||||
# assert(client2.admin_get_file_location(0) is not None)
|
||||
|
||||
# Validate data
|
||||
assert_equal(
|
||||
|
Loading…
Reference in New Issue
Block a user