fix: increase HTTP backend connect timeout from 5s to 30s and make configurable

[lpcox]

Problem

When an HTTP MCP backend is slow to initialize (e.g. cold-starting LiteLLM proxy taking 21s), the gateway's hardcoded 5-second per-transport connect timeout causes all three SDK transport attempts to fail. The tool registration silently drops the backend's tools and the agent starts without them.

From the issue logs:

[2026-04-13T16:37:36Z] [DEBUG] GetOrLaunch called for server: opslevel
[2026-04-13T16:37:57Z] [INFO]  Successfully registered tools from opslevel (took 21.213743169s)

The 21s is consumed by the transport fallback chain: streamable HTTP (5s timeout) → SSE (5s timeout) → plain JSON (~11s). If the backend only supports streamable HTTP, all three transports fail and tools are silently dropped.

Fix

1. Configurable per-transport connect timeout (default 30s → was 5s)

New connect_timeout field on server config:

TOML:

[servers.opslevel]
type = "http"
url = "http://opslevel-proxy:8080/mcp"
connect_timeout = 60  # seconds per transport attempt

JSON stdin:

{
  "mcpServers": {
    "opslevel": {
      "type": "http",
      "url": "http://opslevel-proxy:8080/mcp",
      "connect_timeout": 60
    }
  }
}

2. Better error visibility

• Failed backend tool registration promoted from WARN → ERROR
• Summary error log lists which backends failed and impact on agents:
  Tool registration incomplete: 1 of 3 backends failed: [opslevel] — agents will not see tools from these servers

3. Reconnect uses stored timeout

The connect timeout is stored on the Connection struct so reconnectSDKTransport reuses the same value instead of a hardcoded 10s.

Changes

File	Change
config_core.go	Add ConnectTimeout field, HTTPConnectTimeout() helper, DefaultConnectTimeout constant
config_stdin.go	Add ConnectTimeout to StdinServerConfig, map it during conversion
connection.go	Add connectTimeout to Connection, apply default of 30s when ≤ 0
http_transport.go	Pass configurable timeout through trySDKTransport and newHTTPConnection
launcher.go	Pass serverCfg.HTTPConnectTimeout() to NewHTTPConnection
tool_registry.go	Promote failed backends to ERROR, add summary log
Tests	Updated all NewHTTPConnection call sites with new parameter

Fixes #3718

[Copilot]

Pull request overview

This PR addresses missing HTTP MCP backend tools when backends are slow to initialize by increasing the per-transport HTTP connect timeout (now configurable per server, defaulting to 30s) and improving logging when tool registration fails.

Changes:

• Add per-server connect_timeout (seconds) to config (TOML + stdin JSON) and plumb it through HTTP connection creation / reconnect.
• Replace hardcoded SDK transport connect timeouts with a configurable timeout and persist it on the Connection.
• Improve tool registration failure visibility by logging per-backend failures as ERROR and emitting an ERROR summary.

Show a summary per file

File	Description
internal/config/config_core.go	Adds DefaultConnectTimeout, ServerConfig.ConnectTimeout, and HTTPConnectTimeout() helper.
internal/config/config_core_test.go	Adds tests for parsing and HTTPConnectTimeout() behavior.
internal/config/config_stdin.go	Adds connect_timeout to stdin JSON server config and maps it into ServerConfig.
internal/launcher/launcher.go	Passes per-server connect timeout into mcp.NewHTTPConnection.
internal/mcp/connection.go	Adds connectTimeout to Connection, applies default, and reuses it on SDK reconnect.
internal/mcp/http_transport.go	Threads connect timeout through SDK transport connection attempts and stores it on the connection.
internal/server/tool_registry.go	Promotes backend tool registration failures to ERROR and adds an ERROR summary log.
internal/mcp/http_transport_test.go	Updates NewHTTPConnection callsites for the new signature.
internal/mcp/http_connection_test.go	Updates NewHTTPConnection callsites for the new signature.
internal/mcp/http_error_propagation_test.go	Updates NewHTTPConnection callsites for the new signature.
internal/mcp/connection_test.go	Updates NewHTTPConnection/newHTTPConnection callsites for the new signature.
internal/mcp/connection_stderr_test.go	Updates NewHTTPConnection callsites for the new signature.
internal/mcp/connection_arguments_test.go	Updates NewHTTPConnection callsites for the new signature.
test/integration/http_error_test.go	Updates integration tests for new NewHTTPConnection signature and adds a short connect timeout in the intermittent-failure test.

Copilot's findings

Tip

Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Comments suppressed due to low confidence (1)

internal/config/config_stdin.go:129

• UnmarshalJSON tracks unknown per-server fields in AdditionalProperties via a hardcoded knownFields list. Since connect_timeout isn’t in that list, it will be treated as an additional/custom property, which can interfere with custom schema validation and makes debugging harder. Add connect_timeout to knownFields so it’s recognized as a first-class field.

	// ConnectTimeout is the per-transport timeout (in seconds) for connecting to HTTP backends.
	// Only applies to HTTP server types. Default: 30 seconds.
	ConnectTimeout *int `json:"connect_timeout,omitempty"`

• Files reviewed: 14/14 changed files
• Comments generated: 3