feat(cluster): implement distributed clustering with etcd coordination

- Integrate etcd-client for distributed coordination and leader election
- Add remote client macros with proper formatting for all services
- Implement RequestMetrics for tracking RPC performance and errors
- Add rate limiting mechanism across all service endpoints
- Create ElectionRequest and ElectionResult message types for leader election
- Add role management with primary/replica switching capabilities
- Implement health checker with automatic failover detection
- Add repository count metrics for cluster monitoring
- Update Cargo.toml with etcd-client and dashmap dependencies
- Modify RepoEntry to include read_only flag for replica handling
- Implement should_accept_election logic to prevent duplicate elections
- Add RoleChangedEvent handling for cluster role updates
This commit is contained in:
zhenyi
2026-06-08 14:31:29 +08:00
parent d243dce027
commit 8f472a0443
37 changed files with 4691 additions and 83 deletions
+59 -7
View File
@@ -3,7 +3,11 @@ use crate::pb::*;
use super::{GitksService, cache, into_status, into_stream};
remote_client!(remote_tree_client, TreeServiceClient<tonic::transport::Channel>, "tree");
remote_client!(
remote_tree_client,
TreeServiceClient<tonic::transport::Channel>,
"tree"
);
#[tonic::async_trait]
impl tree_service_server::TreeService for GitksService {
@@ -14,7 +18,9 @@ impl tree_service_server::TreeService for GitksService {
&self,
request: tonic::Request<ListTreeRequest>,
) -> Result<tonic::Response<ListTreeResponse>, tonic::Status> {
let m = crate::metrics::RequestMetrics::new("gitks.TreeService/ListTree");
let inner = request.into_inner();
let _rate = self.acquire_rate_limit(inner.repository.as_ref()).await?;
let repo = self.repo_label(inner.repository.as_ref());
let span = tracing::info_span!("tree.list_tree", %repo);
let _enter = span.enter();
@@ -24,11 +30,16 @@ impl tree_service_server::TreeService for GitksService {
if let Some(mut client) =
remote_tree_client(self, inner.repository.as_ref(), false).await?
{
m.record("ok");
return client.list_tree(inner).await;
}
crate::metrics::record_rpc_error(&m, &err);
return Err(err);
}
Err(err) => {
crate::metrics::record_rpc_error(&m, &err);
return Err(err);
}
Err(err) => return Err(err),
};
let resp = if cache::selector_is_oid(&inner.revision) {
cache::cached_response("tree.list_tree", &inner, || {
@@ -38,6 +49,7 @@ impl tree_service_server::TreeService for GitksService {
gb.list_tree(inner).map_err(into_status)?
};
tracing::info!(%repo, count = resp.entries.len(), "list_tree done");
m.record("ok");
Ok(tonic::Response::new(resp))
}
@@ -45,7 +57,9 @@ impl tree_service_server::TreeService for GitksService {
&self,
request: tonic::Request<GetTreeRequest>,
) -> Result<tonic::Response<Tree>, tonic::Status> {
let m = crate::metrics::RequestMetrics::new("gitks.TreeService/GetTree");
let inner = request.into_inner();
let _rate = self.acquire_rate_limit(inner.repository.as_ref()).await?;
let repo = self.repo_label(inner.repository.as_ref());
let span = tracing::info_span!("tree.get_tree", %repo);
let _enter = span.enter();
@@ -55,11 +69,16 @@ impl tree_service_server::TreeService for GitksService {
if let Some(mut client) =
remote_tree_client(self, inner.repository.as_ref(), false).await?
{
m.record("ok");
return client.get_tree(inner).await;
}
crate::metrics::record_rpc_error(&m, &err);
return Err(err);
}
Err(err) => {
crate::metrics::record_rpc_error(&m, &err);
return Err(err);
}
Err(err) => return Err(err),
};
let resp = if cache::selector_is_oid(&inner.revision) {
cache::cached_response("tree.get_tree", &inner, || {
@@ -68,6 +87,7 @@ impl tree_service_server::TreeService for GitksService {
} else {
gb.get_tree(inner).map_err(into_status)?
};
m.record("ok");
Ok(tonic::Response::new(resp))
}
@@ -75,7 +95,9 @@ impl tree_service_server::TreeService for GitksService {
&self,
request: tonic::Request<GetBlobRequest>,
) -> Result<tonic::Response<Blob>, tonic::Status> {
let m = crate::metrics::RequestMetrics::new("gitks.TreeService/GetBlob");
let inner = request.into_inner();
let _rate = self.acquire_rate_limit(inner.repository.as_ref()).await?;
let repo = self.repo_label(inner.repository.as_ref());
let path = inner.path.clone();
let span = tracing::info_span!("tree.get_blob", %repo, %path);
@@ -86,11 +108,16 @@ impl tree_service_server::TreeService for GitksService {
if let Some(mut client) =
remote_tree_client(self, inner.repository.as_ref(), false).await?
{
m.record("ok");
return client.get_blob(inner).await;
}
crate::metrics::record_rpc_error(&m, &err);
return Err(err);
}
Err(err) => {
crate::metrics::record_rpc_error(&m, &err);
return Err(err);
}
Err(err) => return Err(err),
};
let resp = if cache::selector_is_oid(&inner.revision) {
cache::cached_response("tree.get_blob", &inner, || {
@@ -99,6 +126,7 @@ impl tree_service_server::TreeService for GitksService {
} else {
gb.get_blob(inner).map_err(into_status)?
};
m.record("ok");
Ok(tonic::Response::new(resp))
}
@@ -106,7 +134,9 @@ impl tree_service_server::TreeService for GitksService {
&self,
request: tonic::Request<GetRawBlobRequest>,
) -> Result<tonic::Response<Self::GetRawBlobStream>, tonic::Status> {
let m = crate::metrics::RequestMetrics::new("gitks.TreeService/GetRawBlob");
let inner = request.into_inner();
let _rate = self.acquire_rate_limit(inner.repository.as_ref()).await?;
let repo = self.repo_label(inner.repository.as_ref());
let span = tracing::info_span!("tree.get_raw_blob", %repo);
let _enter = span.enter();
@@ -116,13 +146,18 @@ impl tree_service_server::TreeService for GitksService {
if let Some(mut client) =
remote_tree_client(self, inner.repository.as_ref(), false).await?
{
m.record("ok");
let resp = client.get_raw_blob(inner).await?;
let stream = super::bridge_server_stream(resp.into_inner());
return Ok(tonic::Response::new(stream));
}
crate::metrics::record_rpc_error(&m, &err);
return Err(err);
}
Err(err) => {
crate::metrics::record_rpc_error(&m, &err);
return Err(err);
}
Err(err) => return Err(err),
};
let items = if inner.oid.is_some() {
cache::cached_vec_response("tree.get_raw_blob", &inner, || {
@@ -135,6 +170,7 @@ impl tree_service_server::TreeService for GitksService {
} else {
gb.get_raw_blob(inner).map_err(into_status)?
};
m.record("ok");
Ok(tonic::Response::new(into_stream(items)))
}
@@ -142,7 +178,9 @@ impl tree_service_server::TreeService for GitksService {
&self,
request: tonic::Request<GetFileMetadataRequest>,
) -> Result<tonic::Response<FileMetadata>, tonic::Status> {
let m = crate::metrics::RequestMetrics::new("gitks.TreeService/GetFileMetadata");
let inner = request.into_inner();
let _rate = self.acquire_rate_limit(inner.repository.as_ref()).await?;
let repo = self.repo_label(inner.repository.as_ref());
let span = tracing::info_span!("tree.get_file_metadata", %repo);
let _enter = span.enter();
@@ -152,11 +190,16 @@ impl tree_service_server::TreeService for GitksService {
if let Some(mut client) =
remote_tree_client(self, inner.repository.as_ref(), false).await?
{
m.record("ok");
return client.get_file_metadata(inner).await;
}
crate::metrics::record_rpc_error(&m, &err);
return Err(err);
}
Err(err) => {
crate::metrics::record_rpc_error(&m, &err);
return Err(err);
}
Err(err) => return Err(err),
};
let resp = if cache::selector_is_oid(&inner.revision) {
cache::cached_response("tree.get_file_metadata", &inner, || {
@@ -165,6 +208,7 @@ impl tree_service_server::TreeService for GitksService {
} else {
gb.get_file_metadata(inner).map_err(into_status)?
};
m.record("ok");
Ok(tonic::Response::new(resp))
}
@@ -172,7 +216,9 @@ impl tree_service_server::TreeService for GitksService {
&self,
request: tonic::Request<FindFilesRequest>,
) -> Result<tonic::Response<FindFilesResponse>, tonic::Status> {
let m = crate::metrics::RequestMetrics::new("gitks.TreeService/FindFiles");
let inner = request.into_inner();
let _rate = self.acquire_rate_limit(inner.repository.as_ref()).await?;
let repo = self.repo_label(inner.repository.as_ref());
let span = tracing::info_span!("tree.find_files", %repo);
let _enter = span.enter();
@@ -182,11 +228,16 @@ impl tree_service_server::TreeService for GitksService {
if let Some(mut client) =
remote_tree_client(self, inner.repository.as_ref(), false).await?
{
m.record("ok");
return client.find_files(inner).await;
}
crate::metrics::record_rpc_error(&m, &err);
return Err(err);
}
Err(err) => {
crate::metrics::record_rpc_error(&m, &err);
return Err(err);
}
Err(err) => return Err(err),
};
let resp = if cache::selector_is_oid(&inner.revision) {
cache::cached_response("tree.find_files", &inner, || {
@@ -196,6 +247,7 @@ impl tree_service_server::TreeService for GitksService {
gb.find_files(inner).map_err(into_status)?
};
tracing::info!(%repo, count = resp.files.len(), "find_files done");
m.record("ok");
Ok(tonic::Response::new(resp))
}
}