feat: add startupProbe and replace readiness probe with liveness probe

jina/checker.py

@@ -31,15 +31,15 @@ def __init__(self, args: 'argparse.Namespace'):
                 ) as tc:
                     if args.target == 'executor':
                         hostname, port, protocol, _ = parse_host_scheme(args.host)
-                        r = WorkerRuntime.is_ready(f'{hostname}:{port}')
+                        r = WorkerRuntime.is_ready(ctrl_address=f'{hostname}:{port}')
                     elif args.target == 'gateway':
                         hostname, port, protocol, _ = parse_host_scheme(args.host)
                         r = GatewayRuntime.is_ready(
                             f'{hostname}:{port}',
-                            protocol=GatewayProtocolType.from_string(protocol),
+                            protocol=GatewayProtocolType.from_string(protocol)
                         )
                     elif args.target == 'flow':
-                        r = Client(host=args.host).is_flow_ready(timeout=args.timeout)
+                        r = Client(host=args.host).is_flow_ready(timeout=args.timeout / 1000)
                     if not r:
                         default_logger.warning(
                             'not responding, attempt (%d/%d) in 1s'

jina/clients/base/helper.py

@@ -155,7 +155,9 @@ async def send_dry_run(self, **kwargs):
         :param kwargs: keyword arguments to make sure compatible API with other clients
         :return: send get message
         """
-        return await self.session.get(url=self.url).__aenter__()
+        return await self.session.get(
+            url=self.url, timeout=kwargs.get('timeout', None)
+        ).__aenter__()
 
     async def recv_message(self):
         """Receive message for HTTP (sleep)

jina/resources/k8s/template/deployment-executor.yml

@@ -50,15 +50,25 @@ spec:
             valueFrom:
               fieldRef:
                 fieldPath: metadata.name
-          readinessProbe:
+          startupProbe:
             exec:
               command:
                 - jina
                 - ping
                 - executor
                 - 127.0.0.1:{port}
             initialDelaySeconds: 5
-            periodSeconds: 20
+            periodSeconds: 5
+            timeoutSeconds: 10
+          livenessProbe:
+            exec:
+              command:
+                - jina
+                - ping
+                - executor
+                - 127.0.0.1:{port}
+            initialDelaySeconds: 30
+            periodSeconds: 5
             timeoutSeconds: 10
           lifecycle:
             preStop:

jina/resources/k8s/template/deployment-gateway.yml

@@ -49,15 +49,25 @@ spec:
               valueFrom:
                 fieldRef:
                   fieldPath: metadata.name
-          readinessProbe:
+          startupProbe:
             exec:
               command:
                 - jina
                 - ping
                 - gateway
                 - {protocol}://127.0.0.1:{port}
             initialDelaySeconds: 5
-            periodSeconds: 20
+            periodSeconds: 5
+            timeoutSeconds: 10
+          livenessProbe:
+            exec:
+              command:
+                - jina
+                - ping
+                - gateway
+                - {protocol}://127.0.0.1:{port}
+            initialDelaySeconds: 30
+            periodSeconds: 5
             timeoutSeconds: 10
           lifecycle:
             preStop:

jina/resources/k8s/template/deployment-uses-after.yml

@@ -50,12 +50,26 @@ spec:
           valueFrom:
             fieldRef:
               fieldPath: metadata.name
-
-        readinessProbe:
-          tcpSocket:
-            port: {port}
+        startupProbe:
+          exec:
+            command:
+              - jina
+              - ping
+              - executor
+              - 127.0.0.1:{port}
           initialDelaySeconds: 5
-          periodSeconds: 10
+          periodSeconds: 5
+          timeoutSeconds: 10
+        livenessProbe:
+          exec:
+            command:
+              - jina
+              - ping
+              - executor
+              - 127.0.0.1:{port}
+          initialDelaySeconds: 30
+          periodSeconds: 5
+          timeoutSeconds: 10
         lifecycle:
           preStop:
             exec:
@@ -85,16 +99,25 @@ spec:
           valueFrom:
             fieldRef:
               fieldPath: metadata.name
-
-        readinessProbe:
+        startupProbe:
           exec:
             command:
               - jina
               - ping
               - executor
               - 127.0.0.1:{port_uses_after}
           initialDelaySeconds: 5
-          periodSeconds: 20
+          periodSeconds: 5
+          timeoutSeconds: 10
+        livenessProbe:
+          exec:
+            command:
+              - jina
+              - ping
+              - executor
+              - 127.0.0.1:{port_uses_after}
+          initialDelaySeconds: 30
+          periodSeconds: 5
           timeoutSeconds: 10
         lifecycle:
           preStop:

jina/resources/k8s/template/deployment-uses-before-after.yml

@@ -50,12 +50,26 @@ spec:
           valueFrom:
             fieldRef:
               fieldPath: metadata.name
-
-        readinessProbe:
-          tcpSocket:
-            port: {port}
+        startupProbe:
+          exec:
+            command:
+              - jina
+              - ping
+              - executor
+              - 127.0.0.1:{port_uses}
           initialDelaySeconds: 5
-          periodSeconds: 10
+          periodSeconds: 5
+          timeoutSeconds: 10
+        livenessProbe:
+          exec:
+            command:
+              - jina
+              - ping
+              - executor
+              - 127.0.0.1:{port}
+          initialDelaySeconds: 30
+          periodSeconds: 5
+          timeoutSeconds: 10
         lifecycle:
           preStop:
             exec:
@@ -85,16 +99,25 @@ spec:
           valueFrom:
             fieldRef:
               fieldPath: metadata.name
-
-        readinessProbe:
+        startupProbe:
           exec:
             command:
               - jina
               - ping
               - executor
               - 127.0.0.1:{port_uses_before}
           initialDelaySeconds: 5
-          periodSeconds: 20
+          periodSeconds: 5
+          timeoutSeconds: 10
+        livenessProbe:
+          exec:
+            command:
+              - jina
+              - ping
+              - executor
+              - 127.0.0.1:{port_uses_before}
+          initialDelaySeconds: 30
+          periodSeconds: 5
           timeoutSeconds: 10
         lifecycle:
           preStop:
@@ -125,15 +148,25 @@ spec:
           valueFrom:
             fieldRef:
               fieldPath: metadata.name
-        readinessProbe:
+        startupProbe:
           exec:
             command:
               - jina
               - ping
               - executor
               - 127.0.0.1:{port_uses_after}
           initialDelaySeconds: 5
-          periodSeconds: 20
+          periodSeconds: 5
+          timeoutSeconds: 10
+        livenessProbe:
+          exec:
+            command:
+              - jina
+              - ping
+              - executor
+              - 127.0.0.1:{port_uses_after}
+          initialDelaySeconds: 30
+          periodSeconds: 5
           timeoutSeconds: 10
         lifecycle:
           preStop:

jina/resources/k8s/template/deployment-uses-before.yml

@@ -50,11 +50,26 @@ spec:
           valueFrom:
             fieldRef:
               fieldPath: metadata.name
-        readinessProbe:
-          tcpSocket:
-            port: {port}
+        startupProbe:
+          exec:
+            command:
+              - jina
+              - ping
+              - executor
+              - 127.0.0.1:{port}
           initialDelaySeconds: 5
-          periodSeconds: 10
+          periodSeconds: 5
+          timeoutSeconds: 10
+        livenessProbe:
+          exec:
+            command:
+              - jina
+              - ping
+              - executor
+              - 127.0.0.1:{port}
+          initialDelaySeconds: 30
+          periodSeconds: 5
+          timeoutSeconds: 10
         lifecycle:
           preStop:
             exec:
@@ -84,16 +99,25 @@ spec:
           valueFrom:
             fieldRef:
               fieldPath: metadata.name
-
-        readinessProbe:
+        startupProbe:
           exec:
             command:
               - jina
               - ping
               - executor
               - 127.0.0.1:{port_uses_before}
           initialDelaySeconds: 5
-          periodSeconds: 20
+          periodSeconds: 5
+          timeoutSeconds: 10
+        livenessProbe:
+          exec:
+            command:
+              - jina
+              - ping
+              - executor
+              - 127.0.0.1:{port_uses_before}
+          initialDelaySeconds: 30
+          periodSeconds: 5
           timeoutSeconds: 10
         lifecycle:
           preStop:

jina/resources/k8s/template/statefulset-executor.yml

@@ -50,15 +50,25 @@ spec:
               valueFrom:
                 fieldRef:
                   fieldPath: metadata.name
-          readinessProbe:
+          startupProbe:
             exec:
               command:
                 - jina
                 - ping
                 - executor
                 - 127.0.0.1:{port}
             initialDelaySeconds: 5
-            periodSeconds: 20
+            periodSeconds: 5
+            timeoutSeconds: 10
+          livenessProbe:
+            exec:
+              command:
+                - jina
+                - ping
+                - executor
+                - 127.0.0.1:{port}
+            initialDelaySeconds: 30
+            periodSeconds: 5
             timeoutSeconds: 10
           lifecycle:
             preStop:

jina/serve/networking.py

@@ -975,6 +975,8 @@ async def _handle_aiorpcerror(
             error.code() != grpc.StatusCode.UNAVAILABLE
             and error.code() != grpc.StatusCode.CANCELLED
             and error.code() != grpc.StatusCode.DEADLINE_EXCEEDED
+            and error.code() != grpc.StatusCode.UNKNOWN
+            and error.code() != grpc.StatusCode.INTERNAL
         ):
             return error
         elif (

jina/serve/runtimes/asyncio.py

@@ -161,34 +161,25 @@ async def async_run_forever(self):
     # Static methods used by the Pod to communicate with the `Runtime` in the separate process
 
     @staticmethod
-    def activate(**kwargs):
-        """
-        Activate the runtime, does not apply to these runtimes
-
-        :param kwargs: extra keyword arguments
-        """
-        # does not apply to this types of runtimes
-        pass
-
-    @staticmethod
-    def is_ready(ctrl_address: str, **kwargs) -> bool:
+    def is_ready(ctrl_address: str, timeout: float = 1.0, **kwargs) -> bool:
         """
         Check if status is ready.
 
         :param ctrl_address: the address where the control request needs to be sent
+        :param timeout: timeout of the health check in seconds
         :param kwargs: extra keyword arguments
 
         :return: True if status is ready else False.
         """
-
         try:
             from grpc_health.v1 import health_pb2, health_pb2_grpc
 
             response = GrpcConnectionPool.send_health_check_sync(
-                ctrl_address, timeout=1.0
+                ctrl_address, timeout=timeout
+            )
+            return (
+                response.status == health_pb2.HealthCheckResponse.ServingStatus.SERVING
             )
-            # TODO: Get the proper value of the ServingStatus SERVING KEY
-            return response.status == 1
         except RpcError:
             return False