fix: make McpConnectionManager tolerant of MCPs that fail to start

This commit is contained in:
Michael Bolin
2025-05-08 23:16:16 -07:00
parent b940adae8e
commit 009403b02b
2 changed files with 51 additions and 15 deletions

View File

@@ -561,15 +561,36 @@ async fn submission_loop(
let writable_roots = Mutex::new(get_writable_roots(&cwd));
let mcp_connection_manager =
let (mcp_connection_manager, failed_clients) =
match McpConnectionManager::new(config.mcp_servers.clone()).await {
Ok(mgr) => mgr,
Ok((mgr, failures)) => (mgr, failures),
Err(e) => {
error!("Failed to create MCP connection manager: {e:#}");
McpConnectionManager::default()
(McpConnectionManager::default(), Default::default())
}
};
// Surface individual client start-up failures to the user.
if !failed_clients.is_empty() {
for (server_name, err) in failed_clients {
// Log the failure for debugging.
error!("MCP client for '{server_name}' failed to start: {err:#}");
// Emit an error event so the front-end can inform the user.
let event = Event {
id: sub.id.clone(),
msg: EventMsg::Error {
message: format!(
"Failed to start MCP server '{server_name}': {err}"
),
},
};
// Ignore send failures (agent might have died already).
let _ = tx_event.send(event).await;
}
}
// Attempt to create a RolloutRecorder *before* moving the
// `instructions` value into the Session struct.
let rollout_recorder = match RolloutRecorder::new(instructions.clone()).await {

View File

@@ -29,6 +29,10 @@ const MCP_TOOL_NAME_DELIMITER: &str = "__OAI_CODEX_MCP__";
/// Timeout for the `tools/list` request.
const LIST_TOOLS_TIMEOUT: Duration = Duration::from_secs(10);
/// Map that holds a startup error for every MCP server that could **not** be
/// spawned successfully.
pub type ClientStartErrors = HashMap<String, anyhow::Error>;
fn fully_qualified_tool_name(server: &str, tool: &str) -> String {
format!("{server}{MCP_TOOL_NAME_DELIMITER}{tool}")
}
@@ -60,40 +64,51 @@ impl McpConnectionManager {
/// * `mcp_servers` Map loaded from the user configuration where *keys*
/// are human-readable server identifiers and *values* are the spawn
/// instructions.
pub async fn new(mcp_servers: HashMap<String, McpServerConfig>) -> Result<Self> {
///
/// The function no longer errors out when *individual* MCP servers fail
/// to start. Instead, it returns a tuple `(Self, ClientStartErrors)` where
/// the map stores the error for every server that failed to spawn.
/// Call-sites are expected to inspect the map and surface the failures to
/// the user (e.g. via `EventMsg::Error`).
pub async fn new(
mcp_servers: HashMap<String, McpServerConfig>,
) -> Result<(Self, ClientStartErrors)> {
// Early exit if no servers are configured.
if mcp_servers.is_empty() {
return Ok(Self::default());
return Ok((Self::default(), ClientStartErrors::default()));
}
// Spin up all servers concurrently.
// Launch all configured servers concurrently.
let mut join_set = JoinSet::new();
// Spawn tasks to launch each server.
for (server_name, cfg) in mcp_servers {
// TODO: Verify server name: require `^[a-zA-Z0-9_-]+$`?
join_set.spawn(async move {
let McpServerConfig { command, args, env } = cfg;
let client_res = McpClient::new_stdio_client(command, args, env).await;
(server_name, client_res)
});
}
let mut clients: HashMap<String, std::sync::Arc<McpClient>> =
HashMap::with_capacity(join_set.len());
let mut errors: ClientStartErrors = HashMap::new();
while let Some(res) = join_set.join_next().await {
let (server_name, client_res) = res?;
let (server_name, client_res) = res?; // JoinError propagation
let client = client_res
.with_context(|| format!("failed to spawn MCP server `{server_name}`"))?;
clients.insert(server_name, std::sync::Arc::new(client));
match client_res {
Ok(client) => {
clients.insert(server_name, std::sync::Arc::new(client));
}
Err(e) => {
errors.insert(server_name, e.into());
}
}
}
let tools = list_all_tools(&clients).await?;
Ok(Self { clients, tools })
Ok((Self { clients, tools }, errors))
}
/// Returns a single map that contains **all** tools. Each key is the