Circuit breaker changes using pybreaker

pyproject.toml

@@ -26,6 +26,7 @@ pyarrow = [
     { version = ">=18.0.0", python = ">=3.13", optional=true }
 ]
 pyjwt = "^2.0.0"
+pybreaker = "^1.0.0"
 requests-kerberos = {version = "^0.15.0", optional = true}
 
 

src/databricks/sql/auth/common.py

@@ -51,6 +51,7 @@ def __init__(
         pool_connections: Optional[int] = None,
         pool_maxsize: Optional[int] = None,
         user_agent: Optional[str] = None,
+        telemetry_circuit_breaker_enabled: Optional[bool] = None,
     ):
         self.hostname = hostname
         self.access_token = access_token
@@ -83,6 +84,7 @@ def __init__(
         self.pool_connections = pool_connections or 10
         self.pool_maxsize = pool_maxsize or 20
         self.user_agent = user_agent
+        self.telemetry_circuit_breaker_enabled = bool(telemetry_circuit_breaker_enabled)
 
 
 def get_effective_azure_login_app_id(hostname) -> str:

src/databricks/sql/common/unified_http_client.py

@@ -264,7 +264,31 @@ def request_context(
             yield response
         except MaxRetryError as e:
             logger.error("HTTP request failed after retries: %s", e)
-            raise RequestError(f"HTTP request failed: {e}")
+
+            # Try to extract HTTP status code from the MaxRetryError
+            http_code = None
+            if (
+                hasattr(e, "reason")
+                and e.reason is not None
+                and hasattr(e.reason, "response")
+                and e.reason.response is not None
+            ):
+                # The reason may contain a response object with status
+                http_code = getattr(e.reason.response, "status", None)
+            elif (
+                hasattr(e, "response")
+                and e.response is not None
+                and hasattr(e.response, "status")
+            ):
+                # Or the error itself may have a response
+                http_code = e.response.status
+
+            context = {}
+            if http_code is not None:
+                context["http-code"] = http_code
+                logger.error("HTTP request failed with status code: %d", http_code)
+
+            raise RequestError(f"HTTP request failed: {e}", context=context)
         except Exception as e:
             logger.error("HTTP request error: %s", e)
             raise RequestError(f"HTTP request error: {e}")

src/databricks/sql/exc.py

@@ -126,3 +126,8 @@ class SessionAlreadyClosedError(RequestError):
 
 class CursorAlreadyClosedError(RequestError):
     """Thrown if CancelOperation receives a code 404. ThriftBackend should gracefully proceed as this is expected."""
+
+
+class TelemetryRateLimitError(Exception):
+    """Raised when telemetry endpoint returns 429 or 503, indicating rate limiting or service unavailable.
+    This exception is used exclusively by the circuit breaker to track telemetry rate limiting events."""

src/databricks/sql/telemetry/circuit_breaker_manager.py

@@ -0,0 +1,181 @@
+"""
+Circuit breaker implementation for telemetry requests.
+
+This module provides circuit breaker functionality to prevent telemetry failures
+from impacting the main SQL operations. It uses pybreaker library to implement
+the circuit breaker pattern with configurable thresholds and timeouts.
+"""
+
+import logging
+import threading
+from typing import Dict, Optional, Any
+from dataclasses import dataclass
+
+import pybreaker
+from pybreaker import CircuitBreaker, CircuitBreakerError, CircuitBreakerListener
+
+from databricks.sql.exc import TelemetryRateLimitError
+
+logger = logging.getLogger(__name__)
+
+# Circuit Breaker Configuration Constants
+DEFAULT_MINIMUM_CALLS = 20
+DEFAULT_RESET_TIMEOUT = 30
+DEFAULT_NAME = "telemetry-circuit-breaker"
+
+# Circuit Breaker State Constants (used in logging)
+CIRCUIT_BREAKER_STATE_OPEN = "open"
+CIRCUIT_BREAKER_STATE_CLOSED = "closed"
+CIRCUIT_BREAKER_STATE_HALF_OPEN = "half-open"
+
+# Logging Message Constants
+LOG_CIRCUIT_BREAKER_STATE_CHANGED = "Circuit breaker state changed from %s to %s for %s"
+LOG_CIRCUIT_BREAKER_OPENED = (
+    "Circuit breaker opened for %s - telemetry requests will be blocked"
+)
+LOG_CIRCUIT_BREAKER_CLOSED = (
+    "Circuit breaker closed for %s - telemetry requests will be allowed"
+)
+LOG_CIRCUIT_BREAKER_HALF_OPEN = (
+    "Circuit breaker half-open for %s - testing telemetry requests"
+)
+
+
+class CircuitBreakerStateListener(CircuitBreakerListener):
+    """Listener for circuit breaker state changes."""
+
+    def before_call(self, cb: CircuitBreaker, func, *args, **kwargs) -> None:
+        """Called before the circuit breaker calls a function."""
+        pass
+
+    def failure(self, cb: CircuitBreaker, exc: BaseException) -> None:
+        """Called when a function called by the circuit breaker fails."""
+        pass
+
+    def success(self, cb: CircuitBreaker) -> None:
+        """Called when a function called by the circuit breaker succeeds."""
+        pass
+
+    def state_change(self, cb: CircuitBreaker, old_state, new_state) -> None:
+        """Called when the circuit breaker state changes."""
+        old_state_name = old_state.name if old_state else "None"
+        new_state_name = new_state.name if new_state else "None"
+
+        logger.info(
+            LOG_CIRCUIT_BREAKER_STATE_CHANGED, old_state_name, new_state_name, cb.name
+        )
+
+        if new_state_name == CIRCUIT_BREAKER_STATE_OPEN:
+            logger.warning(LOG_CIRCUIT_BREAKER_OPENED, cb.name)
+        elif new_state_name == CIRCUIT_BREAKER_STATE_CLOSED:
+            logger.info(LOG_CIRCUIT_BREAKER_CLOSED, cb.name)
+        elif new_state_name == CIRCUIT_BREAKER_STATE_HALF_OPEN:
+            logger.info(LOG_CIRCUIT_BREAKER_HALF_OPEN, cb.name)
+
+
+@dataclass(frozen=True)
+class CircuitBreakerConfig:
+    """Configuration for circuit breaker behavior.
+
+    This class is immutable to prevent modification of circuit breaker settings.
+    All configuration values are set to constants defined at the module level.
+    """
+
+    # Minimum number of calls before circuit can open
+    minimum_calls: int = DEFAULT_MINIMUM_CALLS
+
+    # Time to wait before trying to close circuit (in seconds)
+    reset_timeout: int = DEFAULT_RESET_TIMEOUT
+
+    # Name for the circuit breaker (for logging)
+    name: str = DEFAULT_NAME
+
+
+class CircuitBreakerManager:
+    """
+    Manages circuit breaker instances for telemetry requests.
+
+    This class provides a singleton pattern to manage circuit breaker instances
+    per host, ensuring that telemetry failures don't impact main SQL operations.
+    """
+
+    _instances: Dict[str, CircuitBreaker] = {}
+    _lock = threading.RLock()
+    _config: Optional[CircuitBreakerConfig] = None
+
+    @classmethod
+    def initialize(cls, config: CircuitBreakerConfig) -> None:
+        """
+        Initialize the circuit breaker manager with configuration.
+
+        Args:
+            config: Circuit breaker configuration
+        """
+        with cls._lock:
+            cls._config = config
+            logger.debug("CircuitBreakerManager initialized with config: %s", config)
+
+    @classmethod
+    def get_circuit_breaker(cls, host: str) -> CircuitBreaker:
+        """
+        Get or create a circuit breaker instance for the specified host.
+
+        Args:
+            host: The hostname for which to get the circuit breaker
+
+        Returns:
+            CircuitBreaker instance for the host
+        """
+        if not cls._config:
+            # Return a no-op circuit breaker if not initialized
+            return cls._create_noop_circuit_breaker()
+
+        with cls._lock:
+            if host not in cls._instances:
+                cls._instances[host] = cls._create_circuit_breaker(host)
+                logger.debug("Created circuit breaker for host: %s", host)
+
+            return cls._instances[host]
+
+    @classmethod
+    def _create_circuit_breaker(cls, host: str) -> CircuitBreaker:
+        """
+        Create a new circuit breaker instance for the specified host.
+
+        Args:
+            host: The hostname for the circuit breaker
+
+        Returns:
+            New CircuitBreaker instance
+        """
+        config = cls._config
+        if config is None:
+            raise RuntimeError("CircuitBreakerManager not initialized")
+
+        # Create circuit breaker with configuration
+        breaker = CircuitBreaker(
+            fail_max=config.minimum_calls,  # Number of failures before circuit opens
+            reset_timeout=config.reset_timeout,
+            name=f"{config.name}-{host}",
+        )
+
+        # Add state change listeners for logging
+        breaker.add_listener(CircuitBreakerStateListener())
+
+        return breaker
+
+    @classmethod
+    def _create_noop_circuit_breaker(cls) -> CircuitBreaker:
+        """
+        Create a no-op circuit breaker that always allows calls.
+
+        Returns:
+            CircuitBreaker that never opens
+        """
+        # Create a circuit breaker with very high thresholds so it never opens
+        breaker = CircuitBreaker(
+            fail_max=1000000,  # Very high threshold
+            reset_timeout=1,  # Short reset time
+            name="noop-circuit-breaker",
+        )
+        return breaker

src/databricks/sql/telemetry/telemetry_client.py

@@ -41,6 +41,11 @@
 from databricks.sql.common.feature_flag import FeatureFlagsContextFactory
 from databricks.sql.common.unified_http_client import UnifiedHttpClient
 from databricks.sql.common.http import HttpMethod
+from databricks.sql.telemetry.telemetry_push_client import (
+    ITelemetryPushClient,
+    TelemetryPushClient,
+    CircuitBreakerTelemetryPushClient,
+)
 
 if TYPE_CHECKING:
     from databricks.sql.client import Connection
@@ -166,21 +171,21 @@ class TelemetryClient(BaseTelemetryClient):
 
     def __init__(
         self,
-        telemetry_enabled,
-        session_id_hex,
+        telemetry_enabled: bool,
+        session_id_hex: str,
         auth_provider,
-        host_url,
+        host_url: str,
         executor,
-        batch_size,
+        batch_size: int,
         client_context,
-    ):
+    ) -> None:
         logger.debug("Initializing TelemetryClient for connection: %s", session_id_hex)
         self._telemetry_enabled = telemetry_enabled
         self._batch_size = batch_size
         self._session_id_hex = session_id_hex
         self._auth_provider = auth_provider
         self._user_agent = None
-        self._events_batch = []
+        self._events_batch: list = []
         self._lock = threading.RLock()
         self._driver_connection_params = None
         self._host_url = host_url
@@ -189,6 +194,19 @@ def __init__(
         # Create own HTTP client from client context
         self._http_client = UnifiedHttpClient(client_context)
 
+        # Create telemetry push client based on circuit breaker enabled flag
+        if client_context.telemetry_circuit_breaker_enabled:
+            # Create circuit breaker telemetry push client with fixed configuration
+            self._telemetry_push_client: ITelemetryPushClient = (
+                CircuitBreakerTelemetryPushClient(
+                    TelemetryPushClient(self._http_client),
+                    host_url,
+                )
+            )
+        else:
+            # Circuit breaker disabled - use direct telemetry push client
+            self._telemetry_push_client = TelemetryPushClient(self._http_client)
+
     def _export_event(self, event):
         """Add an event to the batch queue and flush if batch is full"""
         logger.debug("Exporting event for connection %s", self._session_id_hex)
@@ -254,7 +272,7 @@ def _send_telemetry(self, events):
     def _send_with_unified_client(self, url, data, headers, timeout=900):
         """Helper method to send telemetry using the unified HTTP client."""
         try:
-            response = self._http_client.request(
+            response = self._telemetry_push_client.request(
                 HttpMethod.POST, url, body=data, headers=headers, timeout=timeout
             )
             return response