From d2dc4b18f1f7322b283e599d3cfc5531662e78d5 Mon Sep 17 00:00:00 2001
From: Simon Schrottner <simon.schrottner@dynatrace.com>
Date: Thu, 28 Nov 2024 19:15:40 +0100
Subject: [PATCH 1/9] feat(flagd-rpc): add caching with tests

Signed-off-by: Simon Schrottner <simon.schrottner@dynatrace.com>
---
 .../contrib/provider/flagd/resolvers/grpc.py  |  14 ++
 .../tests/e2e/config.feature                  | 189 ++++++++++++++++++
 .../tests/e2e/rpc_cache.feature               |  44 ++++
 3 files changed, 247 insertions(+)
 create mode 100644 providers/openfeature-provider-flagd/tests/e2e/config.feature
 create mode 100644 providers/openfeature-provider-flagd/tests/e2e/rpc_cache.feature

diff --git a/providers/openfeature-provider-flagd/src/openfeature/contrib/provider/flagd/resolvers/grpc.py b/providers/openfeature-provider-flagd/src/openfeature/contrib/provider/flagd/resolvers/grpc.py
index 8dfda518..4699fb53 100644
--- a/providers/openfeature-provider-flagd/src/openfeature/contrib/provider/flagd/resolvers/grpc.py
+++ b/providers/openfeature-provider-flagd/src/openfeature/contrib/provider/flagd/resolvers/grpc.py
@@ -65,6 +65,12 @@ def __init__(
         self.deadline = config.deadline_ms * 0.001
         self.connected = False
 
+        self._cache: typing.Optional[BaseCacheImpl] = (
+            LRUCache(maxsize=self.config.max_cache_size)
+            if self.config.cache_type == CacheType.LRU
+            else None
+        )
+
     def _create_stub(
         self,
     ) -> typing.Tuple[evaluation_pb2_grpc.ServiceStub, grpc.Channel]:
@@ -80,6 +86,14 @@ def _create_stub(
 
     def initialize(self, evaluation_context: EvaluationContext) -> None:
         self.connect()
+        self.retry_backoff_seconds = 0.1
+        self.connected = False
+
+        self._cache = (
+            LRUCache(maxsize=self.config.max_cache_size)
+            if self.config.cache_type == CacheType.LRU
+            else None
+        )
 
     def shutdown(self) -> None:
         self.active = False
diff --git a/providers/openfeature-provider-flagd/tests/e2e/config.feature b/providers/openfeature-provider-flagd/tests/e2e/config.feature
new file mode 100644
index 00000000..f2bd715e
--- /dev/null
+++ b/providers/openfeature-provider-flagd/tests/e2e/config.feature
@@ -0,0 +1,189 @@
+Feature: Configuration Test
+
+  @rpc @in-process
+  Scenario Outline: Default Config
+    When we initialize a config
+    Then the option "<option>" of type "<type>" should have the value "<default>"
+    Examples: Basic
+      | option       | type         | default   |
+      | resolverType | ResolverType | rpc       |
+      | host         | String       | localhost |
+      | port         | Integer      | 8013      |
+      | tls          | Boolean      | false     |
+      | deadline     | Integer      | 500       |
+    @customCert
+    Examples: Certificates
+      | option   | type   | default |
+      | certPath | String | null    |
+    @unixsocket
+    Examples: Unixsocket
+      | option     | type   | default |
+      | socketPath | String | null    |
+    @events
+    Examples: Events
+      | option           | type    | default |
+      | streamDeadlineMs | Integer | 600000  |
+      | keepAlive        | Long    | 0       |
+      | retryBackoffMs   | Integer | 1000    |
+    @sync
+    Examples: Sync
+      | option           | type    | default |
+      | streamDeadlineMs | Integer | 600000  |
+      | keepAlive        | Long    | 0       |
+      | retryBackoffMs   | Integer | 1000    |
+      | selector         | String  | null    |
+    @caching
+    Examples: caching
+      | option       | type      | default |
+      | cacheType    | CacheType | lru     |
+      | maxCacheSize | Integer   | 1000    |
+    @offline
+    Examples: offline
+      | option                | type   | default |
+      | offlineFlagSourcePath | String | null    |
+
+  @rpc
+  Scenario Outline: Default Config RPC
+    When we initialize a config for "rpc"
+    Then the option "<option>" of type "<type>" should have the value "<default>"
+    Examples:
+      | option | type    | default |
+      | port   | Integer | 8013    |
+
+  @in-process
+  Scenario Outline: Default Config In-Process
+    When we initialize a config for "in-process"
+    Then the option "<option>" of type "<type>" should have the value "<default>"
+    Examples:
+      | option | type    | default |
+      | port   | Integer | 8015    |
+
+  Scenario Outline: Dedicated Config
+    When we have an option "<option>" of type "<type>" with value "<value>"
+    And we initialize a config
+    Then the option "<option>" of type "<type>" should have the value "<value>"
+    Examples:
+      | option       | type         | value      |
+      | resolverType | ResolverType | in-process |
+      | host         | String       | local      |
+      | tls          | Boolean      | True       |
+      | port         | Integer      | 1234       |
+      | deadline     | Integer      | 123        |
+    @customCert
+    Examples:
+      | option   | type   | value |
+      | certPath | String | path  |
+    @unixsocket
+    Examples:
+      | option     | type   | value |
+      | socketPath | String | path  |
+    @events
+    Examples:
+      | option           | type    | value  |
+      | streamDeadlineMs | Integer | 500000 |
+      | keepAlive        | Long    | 5      |
+      | retryBackoffMs   | Integer | 5000   |
+    @sync
+    Examples:
+      | option           | type    | value    |
+      | streamDeadlineMs | Integer | 500000   |
+      | keepAlive        | Long    | 5        |
+      | retryBackoffMs   | Integer | 5000     |
+      | selector         | String  | selector |
+    @caching
+    Examples:
+      | option       | type      | value    |
+      | cacheType    | CacheType | disabled |
+      | maxCacheSize | Integer   | 1236     |
+    @offline
+    Examples:
+      | option                | type   | value |
+      | offlineFlagSourcePath | String | path  |
+
+  Scenario Outline: Dedicated Config via Env_var
+    When we have an environment variable "<env>" with value "<value>"
+    And we initialize a config
+    Then the option "<option>" of type "<type>" should have the value "<value>"
+    Examples:
+      | option       | env               | type         | value      |
+      | resolverType | FLAGD_RESOLVER    | ResolverType | in-process |
+      | resolverType | FLAGD_RESOLVER    | ResolverType | IN-PROCESS |
+      | resolverType | FLAGD_RESOLVER    | ResolverType | rpc        |
+      | resolverType | FLAGD_RESOLVER    | ResolverType | RPC        |
+      | host         | FLAGD_HOST        | String       | local      |
+      | tls          | FLAGD_TLS         | Boolean      | True       |
+      | port         | FLAGD_PORT        | Integer      | 1234       |
+      | deadline     | FLAGD_DEADLINE_MS | Integer      | 123        |
+    @customCert
+    Examples:
+      | option   | env                    | type   | value |
+      | certPath | FLAGD_SERVER_CERT_PATH | String | path  |
+    @unixsocket
+    Examples:
+      | option     | env               | type   | value |
+      | socketPath | FLAGD_SOCKET_PATH | String | path  |
+    @events
+    Examples:
+      | option           | env                      | type    | value  |
+      | streamDeadlineMs | FLAGD_STREAM_DEADLINE_MS | Integer | 500000 |
+      | keepAlive        | FLAGD_KEEP_ALIVE_TIME_MS | Long    | 5      |
+      | retryBackoffMs   | FLAGD_RETRY_BACKOFF_MS   | Integer | 5000   |
+    @sync
+    Examples:
+      | option           | env                      | type    | value    |
+      | streamDeadlineMs | FLAGD_STREAM_DEADLINE_MS | Integer | 500000   |
+      | keepAlive        | FLAGD_KEEP_ALIVE_TIME_MS | Long    | 5        |
+      | retryBackoffMs   | FLAGD_RETRY_BACKOFF_MS   | Integer | 5000     |
+      | selector         | FLAGD_SOURCE_SELECTOR    | String  | selector |
+    @caching
+    Examples:
+      | option       | env                  | type      | value    |
+      | cacheType    | FLAGD_CACHE          | CacheType | disabled |
+      | maxCacheSize | FLAGD_MAX_CACHE_SIZE | Integer   | 1236     |
+    @offline
+    Examples:
+      | option                | env                            | type   | value |
+      | offlineFlagSourcePath | FLAGD_OFFLINE_FLAG_SOURCE_PATH | String | path  |
+
+  Scenario Outline: Dedicated Config via Env_var and set
+    When we have an environment variable "<env>" with value "<env-value>"
+    And we have an option "<option>" of type "<type>" with value "<value>"
+    And we initialize a config
+    Then the option "<option>" of type "<type>" should have the value "<value>"
+    Examples:
+      | option       | env               | type         | value      | env-value |
+      | resolverType | FLAGD_RESOLVER    | ResolverType | in-process | rpc       |
+      | host         | FLAGD_HOST        | String       | local      | l         |
+      | tls          | FLAGD_TLS         | Boolean      | True       | False     |
+      | port         | FLAGD_PORT        | Integer      | 1234       | 3456      |
+      | deadline     | FLAGD_DEADLINE_MS | Integer      | 123        | 345       |
+    @customCert
+    Examples:
+      | option   | env                    | type   | value | env-value |
+      | certPath | FLAGD_SERVER_CERT_PATH | String | path  | rpc       |
+    @unixsocket
+    Examples:
+      | option     | env               | type   | value | env-value |
+      | socketPath | FLAGD_SOCKET_PATH | String | path  | rpc       |
+    @events
+    Examples:
+      | option           | env                      | type    | value  | env-value |
+      | streamDeadlineMs | FLAGD_STREAM_DEADLINE_MS | Integer | 500000 | 400       |
+      | keepAlive        | FLAGD_KEEP_ALIVE_TIME_MS | Long    | 5      | 4         |
+      | retryBackoffMs   | FLAGD_RETRY_BACKOFF_MS   | Integer | 5000   | 400       |
+    @sync
+    Examples:
+      | option           | env                      | type    | value  | env-value |
+      | streamDeadlineMs | FLAGD_STREAM_DEADLINE_MS | Integer | 500000 | 400       |
+      | keepAlive        | FLAGD_KEEP_ALIVE_TIME_MS | Long    | 5      | 4         |
+      | retryBackoffMs   | FLAGD_RETRY_BACKOFF_MS   | Integer | 5000   | 400       |
+      | selector | FLAGD_SOURCE_SELECTOR | String | selector | sele      |
+    @caching
+    Examples:
+      | option       | env                  | type      | value    | env-value |
+      | cacheType    | FLAGD_CACHE          | CacheType | disabled | lru       |
+      | maxCacheSize | FLAGD_MAX_CACHE_SIZE | Integer   | 1236     | 2345      |
+    @offline
+    Examples:
+      | option                | env                            | type   | value | env-value |
+      | offlineFlagSourcePath | FLAGD_OFFLINE_FLAG_SOURCE_PATH | String | path  | lll       |
diff --git a/providers/openfeature-provider-flagd/tests/e2e/rpc_cache.feature b/providers/openfeature-provider-flagd/tests/e2e/rpc_cache.feature
new file mode 100644
index 00000000..86aaacba
--- /dev/null
+++ b/providers/openfeature-provider-flagd/tests/e2e/rpc_cache.feature
@@ -0,0 +1,44 @@
+Feature: Flag evaluation with Caching
+
+# This test suite contains scenarios to test the flag evaluation API.
+
+  Background:
+    Given a provider is registered
+
+  Scenario: Resolves boolean details with caching
+    When a boolean flag with key "boolean-flag" is evaluated with details and default value "false"
+    Then the resolved boolean details value should be "true", the variant should be "on", and the reason should be "STATIC"
+    Then the resolved boolean details value should be "true", the variant should be "on", and the reason should be "CACHED"
+
+  Scenario: Resolves string details with caching
+    When a string flag with key "string-flag" is evaluated with details and default value "bye"
+    Then the resolved string details value should be "hi", the variant should be "greeting", and the reason should be "STATIC"
+    Then the resolved string details value should be "hi", the variant should be "greeting", and the reason should be "CACHED"
+
+  Scenario: Resolves integer details with caching
+    When an integer flag with key "integer-flag" is evaluated with details and default value 1
+    Then the resolved integer details value should be 10, the variant should be "ten", and the reason should be "STATIC"
+    Then the resolved integer details value should be 10, the variant should be "ten", and the reason should be "CACHED"
+
+  Scenario: Resolves float details with caching
+    When a float flag with key "float-flag" is evaluated with details and default value 0.1
+    Then the resolved float details value should be 0.5, the variant should be "half", and the reason should be "STATIC"
+    Then the resolved float details value should be 0.5, the variant should be "half", and the reason should be "CACHED"
+
+  Scenario: Resolves object details with caching
+    When an object flag with key "object-flag" is evaluated with details and a null default value
+    Then the resolved object details value should be contain fields "showImages", "title", and "imagesPerPage", with values "true", "Check out these pics!" and 100, respectively
+    And the variant should be "template", and the reason should be "STATIC"
+    Then the resolved object details value should be contain fields "showImages", "title", and "imagesPerPage", with values "true", "Check out these pics!" and 100, respectively
+    And the variant should be "template", and the reason should be "CACHED"
+
+  Scenario: Flag change event with caching
+    When a string flag with key "changing-flag" is evaluated with details
+    When a PROVIDER_CONFIGURATION_CHANGED handler is added
+    And a flag with key "changing-flag" is modified
+    Then the returned reason should be "STATIC"
+    Then the returned reason should be "CACHED"
+    Then the PROVIDER_CONFIGURATION_CHANGED handler must run
+    And the event details must indicate "changing-flag" was altered
+    Then the returned reason should be "STATIC"
+    Then the returned reason should be "CACHED"
\ No newline at end of file

From faca6a063b459459d2f26573e4a4e47a02cff95c Mon Sep 17 00:00:00 2001
From: Simon Schrottner <simon.schrottner@dynatrace.com>
Date: Mon, 2 Dec 2024 18:34:11 +0100
Subject: [PATCH 2/9] fixup: using new test-harness

Signed-off-by: Simon Schrottner <simon.schrottner@dynatrace.com>
---
 .../tests/e2e/config.feature                  | 189 ------------------
 .../tests/e2e/rpc_cache.feature               |  44 ----
 2 files changed, 233 deletions(-)
 delete mode 100644 providers/openfeature-provider-flagd/tests/e2e/config.feature
 delete mode 100644 providers/openfeature-provider-flagd/tests/e2e/rpc_cache.feature

diff --git a/providers/openfeature-provider-flagd/tests/e2e/config.feature b/providers/openfeature-provider-flagd/tests/e2e/config.feature
deleted file mode 100644
index f2bd715e..00000000
--- a/providers/openfeature-provider-flagd/tests/e2e/config.feature
+++ /dev/null
@@ -1,189 +0,0 @@
-Feature: Configuration Test
-
-  @rpc @in-process
-  Scenario Outline: Default Config
-    When we initialize a config
-    Then the option "<option>" of type "<type>" should have the value "<default>"
-    Examples: Basic
-      | option       | type         | default   |
-      | resolverType | ResolverType | rpc       |
-      | host         | String       | localhost |
-      | port         | Integer      | 8013      |
-      | tls          | Boolean      | false     |
-      | deadline     | Integer      | 500       |
-    @customCert
-    Examples: Certificates
-      | option   | type   | default |
-      | certPath | String | null    |
-    @unixsocket
-    Examples: Unixsocket
-      | option     | type   | default |
-      | socketPath | String | null    |
-    @events
-    Examples: Events
-      | option           | type    | default |
-      | streamDeadlineMs | Integer | 600000  |
-      | keepAlive        | Long    | 0       |
-      | retryBackoffMs   | Integer | 1000    |
-    @sync
-    Examples: Sync
-      | option           | type    | default |
-      | streamDeadlineMs | Integer | 600000  |
-      | keepAlive        | Long    | 0       |
-      | retryBackoffMs   | Integer | 1000    |
-      | selector         | String  | null    |
-    @caching
-    Examples: caching
-      | option       | type      | default |
-      | cacheType    | CacheType | lru     |
-      | maxCacheSize | Integer   | 1000    |
-    @offline
-    Examples: offline
-      | option                | type   | default |
-      | offlineFlagSourcePath | String | null    |
-
-  @rpc
-  Scenario Outline: Default Config RPC
-    When we initialize a config for "rpc"
-    Then the option "<option>" of type "<type>" should have the value "<default>"
-    Examples:
-      | option | type    | default |
-      | port   | Integer | 8013    |
-
-  @in-process
-  Scenario Outline: Default Config In-Process
-    When we initialize a config for "in-process"
-    Then the option "<option>" of type "<type>" should have the value "<default>"
-    Examples:
-      | option | type    | default |
-      | port   | Integer | 8015    |
-
-  Scenario Outline: Dedicated Config
-    When we have an option "<option>" of type "<type>" with value "<value>"
-    And we initialize a config
-    Then the option "<option>" of type "<type>" should have the value "<value>"
-    Examples:
-      | option       | type         | value      |
-      | resolverType | ResolverType | in-process |
-      | host         | String       | local      |
-      | tls          | Boolean      | True       |
-      | port         | Integer      | 1234       |
-      | deadline     | Integer      | 123        |
-    @customCert
-    Examples:
-      | option   | type   | value |
-      | certPath | String | path  |
-    @unixsocket
-    Examples:
-      | option     | type   | value |
-      | socketPath | String | path  |
-    @events
-    Examples:
-      | option           | type    | value  |
-      | streamDeadlineMs | Integer | 500000 |
-      | keepAlive        | Long    | 5      |
-      | retryBackoffMs   | Integer | 5000   |
-    @sync
-    Examples:
-      | option           | type    | value    |
-      | streamDeadlineMs | Integer | 500000   |
-      | keepAlive        | Long    | 5        |
-      | retryBackoffMs   | Integer | 5000     |
-      | selector         | String  | selector |
-    @caching
-    Examples:
-      | option       | type      | value    |
-      | cacheType    | CacheType | disabled |
-      | maxCacheSize | Integer   | 1236     |
-    @offline
-    Examples:
-      | option                | type   | value |
-      | offlineFlagSourcePath | String | path  |
-
-  Scenario Outline: Dedicated Config via Env_var
-    When we have an environment variable "<env>" with value "<value>"
-    And we initialize a config
-    Then the option "<option>" of type "<type>" should have the value "<value>"
-    Examples:
-      | option       | env               | type         | value      |
-      | resolverType | FLAGD_RESOLVER    | ResolverType | in-process |
-      | resolverType | FLAGD_RESOLVER    | ResolverType | IN-PROCESS |
-      | resolverType | FLAGD_RESOLVER    | ResolverType | rpc        |
-      | resolverType | FLAGD_RESOLVER    | ResolverType | RPC        |
-      | host         | FLAGD_HOST        | String       | local      |
-      | tls          | FLAGD_TLS         | Boolean      | True       |
-      | port         | FLAGD_PORT        | Integer      | 1234       |
-      | deadline     | FLAGD_DEADLINE_MS | Integer      | 123        |
-    @customCert
-    Examples:
-      | option   | env                    | type   | value |
-      | certPath | FLAGD_SERVER_CERT_PATH | String | path  |
-    @unixsocket
-    Examples:
-      | option     | env               | type   | value |
-      | socketPath | FLAGD_SOCKET_PATH | String | path  |
-    @events
-    Examples:
-      | option           | env                      | type    | value  |
-      | streamDeadlineMs | FLAGD_STREAM_DEADLINE_MS | Integer | 500000 |
-      | keepAlive        | FLAGD_KEEP_ALIVE_TIME_MS | Long    | 5      |
-      | retryBackoffMs   | FLAGD_RETRY_BACKOFF_MS   | Integer | 5000   |
-    @sync
-    Examples:
-      | option           | env                      | type    | value    |
-      | streamDeadlineMs | FLAGD_STREAM_DEADLINE_MS | Integer | 500000   |
-      | keepAlive        | FLAGD_KEEP_ALIVE_TIME_MS | Long    | 5        |
-      | retryBackoffMs   | FLAGD_RETRY_BACKOFF_MS   | Integer | 5000     |
-      | selector         | FLAGD_SOURCE_SELECTOR    | String  | selector |
-    @caching
-    Examples:
-      | option       | env                  | type      | value    |
-      | cacheType    | FLAGD_CACHE          | CacheType | disabled |
-      | maxCacheSize | FLAGD_MAX_CACHE_SIZE | Integer   | 1236     |
-    @offline
-    Examples:
-      | option                | env                            | type   | value |
-      | offlineFlagSourcePath | FLAGD_OFFLINE_FLAG_SOURCE_PATH | String | path  |
-
-  Scenario Outline: Dedicated Config via Env_var and set
-    When we have an environment variable "<env>" with value "<env-value>"
-    And we have an option "<option>" of type "<type>" with value "<value>"
-    And we initialize a config
-    Then the option "<option>" of type "<type>" should have the value "<value>"
-    Examples:
-      | option       | env               | type         | value      | env-value |
-      | resolverType | FLAGD_RESOLVER    | ResolverType | in-process | rpc       |
-      | host         | FLAGD_HOST        | String       | local      | l         |
-      | tls          | FLAGD_TLS         | Boolean      | True       | False     |
-      | port         | FLAGD_PORT        | Integer      | 1234       | 3456      |
-      | deadline     | FLAGD_DEADLINE_MS | Integer      | 123        | 345       |
-    @customCert
-    Examples:
-      | option   | env                    | type   | value | env-value |
-      | certPath | FLAGD_SERVER_CERT_PATH | String | path  | rpc       |
-    @unixsocket
-    Examples:
-      | option     | env               | type   | value | env-value |
-      | socketPath | FLAGD_SOCKET_PATH | String | path  | rpc       |
-    @events
-    Examples:
-      | option           | env                      | type    | value  | env-value |
-      | streamDeadlineMs | FLAGD_STREAM_DEADLINE_MS | Integer | 500000 | 400       |
-      | keepAlive        | FLAGD_KEEP_ALIVE_TIME_MS | Long    | 5      | 4         |
-      | retryBackoffMs   | FLAGD_RETRY_BACKOFF_MS   | Integer | 5000   | 400       |
-    @sync
-    Examples:
-      | option           | env                      | type    | value  | env-value |
-      | streamDeadlineMs | FLAGD_STREAM_DEADLINE_MS | Integer | 500000 | 400       |
-      | keepAlive        | FLAGD_KEEP_ALIVE_TIME_MS | Long    | 5      | 4         |
-      | retryBackoffMs   | FLAGD_RETRY_BACKOFF_MS   | Integer | 5000   | 400       |
-      | selector | FLAGD_SOURCE_SELECTOR | String | selector | sele      |
-    @caching
-    Examples:
-      | option       | env                  | type      | value    | env-value |
-      | cacheType    | FLAGD_CACHE          | CacheType | disabled | lru       |
-      | maxCacheSize | FLAGD_MAX_CACHE_SIZE | Integer   | 1236     | 2345      |
-    @offline
-    Examples:
-      | option                | env                            | type   | value | env-value |
-      | offlineFlagSourcePath | FLAGD_OFFLINE_FLAG_SOURCE_PATH | String | path  | lll       |
diff --git a/providers/openfeature-provider-flagd/tests/e2e/rpc_cache.feature b/providers/openfeature-provider-flagd/tests/e2e/rpc_cache.feature
deleted file mode 100644
index 86aaacba..00000000
--- a/providers/openfeature-provider-flagd/tests/e2e/rpc_cache.feature
+++ /dev/null
@@ -1,44 +0,0 @@
-Feature: Flag evaluation with Caching
-
-# This test suite contains scenarios to test the flag evaluation API.
-
-  Background:
-    Given a provider is registered
-
-  Scenario: Resolves boolean details with caching
-    When a boolean flag with key "boolean-flag" is evaluated with details and default value "false"
-    Then the resolved boolean details value should be "true", the variant should be "on", and the reason should be "STATIC"
-    Then the resolved boolean details value should be "true", the variant should be "on", and the reason should be "CACHED"
-
-  Scenario: Resolves string details with caching
-    When a string flag with key "string-flag" is evaluated with details and default value "bye"
-    Then the resolved string details value should be "hi", the variant should be "greeting", and the reason should be "STATIC"
-    Then the resolved string details value should be "hi", the variant should be "greeting", and the reason should be "CACHED"
-
-  Scenario: Resolves integer details with caching
-    When an integer flag with key "integer-flag" is evaluated with details and default value 1
-    Then the resolved integer details value should be 10, the variant should be "ten", and the reason should be "STATIC"
-    Then the resolved integer details value should be 10, the variant should be "ten", and the reason should be "CACHED"
-
-  Scenario: Resolves float details with caching
-    When a float flag with key "float-flag" is evaluated with details and default value 0.1
-    Then the resolved float details value should be 0.5, the variant should be "half", and the reason should be "STATIC"
-    Then the resolved float details value should be 0.5, the variant should be "half", and the reason should be "CACHED"
-
-  Scenario: Resolves object details with caching
-    When an object flag with key "object-flag" is evaluated with details and a null default value
-    Then the resolved object details value should be contain fields "showImages", "title", and "imagesPerPage", with values "true", "Check out these pics!" and 100, respectively
-    And the variant should be "template", and the reason should be "STATIC"
-    Then the resolved object details value should be contain fields "showImages", "title", and "imagesPerPage", with values "true", "Check out these pics!" and 100, respectively
-    And the variant should be "template", and the reason should be "CACHED"
-
-  Scenario: Flag change event with caching
-    When a string flag with key "changing-flag" is evaluated with details
-    When a PROVIDER_CONFIGURATION_CHANGED handler is added
-    And a flag with key "changing-flag" is modified
-    Then the returned reason should be "STATIC"
-    Then the returned reason should be "CACHED"
-    Then the PROVIDER_CONFIGURATION_CHANGED handler must run
-    And the event details must indicate "changing-flag" was altered
-    Then the returned reason should be "STATIC"
-    Then the returned reason should be "CACHED"
\ No newline at end of file

From 632537da78be338ce28eda9211093ea0d1f2f1e5 Mon Sep 17 00:00:00 2001
From: Simon Schrottner <simon.schrottner@dynatrace.com>
Date: Wed, 4 Dec 2024 17:07:51 +0100
Subject: [PATCH 3/9] fixup(flagd): remove merge conflict error as stated by
 warber

Signed-off-by: Simon Schrottner <simon.schrottner@dynatrace.com>
---
 .../contrib/provider/flagd/resolvers/grpc.py       | 14 --------------
 1 file changed, 14 deletions(-)

diff --git a/providers/openfeature-provider-flagd/src/openfeature/contrib/provider/flagd/resolvers/grpc.py b/providers/openfeature-provider-flagd/src/openfeature/contrib/provider/flagd/resolvers/grpc.py
index 4699fb53..8dfda518 100644
--- a/providers/openfeature-provider-flagd/src/openfeature/contrib/provider/flagd/resolvers/grpc.py
+++ b/providers/openfeature-provider-flagd/src/openfeature/contrib/provider/flagd/resolvers/grpc.py
@@ -65,12 +65,6 @@ def __init__(
         self.deadline = config.deadline_ms * 0.001
         self.connected = False
 
-        self._cache: typing.Optional[BaseCacheImpl] = (
-            LRUCache(maxsize=self.config.max_cache_size)
-            if self.config.cache_type == CacheType.LRU
-            else None
-        )
-
     def _create_stub(
         self,
     ) -> typing.Tuple[evaluation_pb2_grpc.ServiceStub, grpc.Channel]:
@@ -86,14 +80,6 @@ def _create_stub(
 
     def initialize(self, evaluation_context: EvaluationContext) -> None:
         self.connect()
-        self.retry_backoff_seconds = 0.1
-        self.connected = False
-
-        self._cache = (
-            LRUCache(maxsize=self.config.max_cache_size)
-            if self.config.cache_type == CacheType.LRU
-            else None
-        )
 
     def shutdown(self) -> None:
         self.active = False

From eb48bfae69bd044b3b2a7437f1cc3af423d54365 Mon Sep 17 00:00:00 2001
From: Simon Schrottner <simon.schrottner@dynatrace.com>
Date: Fri, 6 Dec 2024 09:19:18 +0100
Subject: [PATCH 4/9] feat(flagd): add graceful attempts

Signed-off-by: Simon Schrottner <simon.schrottner@dynatrace.com>
---
 .../tests/e2e/events.feature                  | 29 +++++++++++++++++++
 1 file changed, 29 insertions(+)
 create mode 100644 providers/openfeature-provider-flagd/tests/e2e/events.feature

diff --git a/providers/openfeature-provider-flagd/tests/e2e/events.feature b/providers/openfeature-provider-flagd/tests/e2e/events.feature
new file mode 100644
index 00000000..946eece5
--- /dev/null
+++ b/providers/openfeature-provider-flagd/tests/e2e/events.feature
@@ -0,0 +1,29 @@
+Feature: flagd providers
+
+  # This test suite contains scenarios to test flagd providers.
+  # It's associated with the flags configured in flags/changing-flag.json and flags/zero-flags.json.
+  # It should be used in conjunction with the suites supplied by the OpenFeature specification.
+
+  Background:
+    Given a flagd provider is set
+
+  # events
+  Scenario Outline: Provider events
+    When a <event> handler is added
+    Then the <event> handler must run
+    Examples:
+      | event                          |
+      | PROVIDER_ERROR                 |
+      | PROVIDER_STALE                 |
+      | PROVIDER_READY                 |
+
+  # events
+  Scenario: Provider events chain ready -> stale -> error -> ready
+    When a PROVIDER_READY handler is added
+    Then the PROVIDER_READY handler must run
+    When a PROVIDER_STALE handler is added
+    Then the PROVIDER_STALE handler must run
+    When a PROVIDER_ERROR handler is added
+    Then the PROVIDER_ERROR handler must run
+    When a PROVIDER_READY handler is added
+    Then the PROVIDER_READY handler must run

From 7aee3c89aeafef66beb6bf19f590aa7add311fc1 Mon Sep 17 00:00:00 2001
From: Simon Schrottner <simon.schrottner@dynatrace.com>
Date: Fri, 6 Dec 2024 09:19:39 +0100
Subject: [PATCH 5/9] feat(flagd): add graceful attempts

Signed-off-by: Simon Schrottner <simon.schrottner@dynatrace.com>
---
 .../tests/e2e/events.feature                  | 29 -------------------
 1 file changed, 29 deletions(-)
 delete mode 100644 providers/openfeature-provider-flagd/tests/e2e/events.feature

diff --git a/providers/openfeature-provider-flagd/tests/e2e/events.feature b/providers/openfeature-provider-flagd/tests/e2e/events.feature
deleted file mode 100644
index 946eece5..00000000
--- a/providers/openfeature-provider-flagd/tests/e2e/events.feature
+++ /dev/null
@@ -1,29 +0,0 @@
-Feature: flagd providers
-
-  # This test suite contains scenarios to test flagd providers.
-  # It's associated with the flags configured in flags/changing-flag.json and flags/zero-flags.json.
-  # It should be used in conjunction with the suites supplied by the OpenFeature specification.
-
-  Background:
-    Given a flagd provider is set
-
-  # events
-  Scenario Outline: Provider events
-    When a <event> handler is added
-    Then the <event> handler must run
-    Examples:
-      | event                          |
-      | PROVIDER_ERROR                 |
-      | PROVIDER_STALE                 |
-      | PROVIDER_READY                 |
-
-  # events
-  Scenario: Provider events chain ready -> stale -> error -> ready
-    When a PROVIDER_READY handler is added
-    Then the PROVIDER_READY handler must run
-    When a PROVIDER_STALE handler is added
-    Then the PROVIDER_STALE handler must run
-    When a PROVIDER_ERROR handler is added
-    Then the PROVIDER_ERROR handler must run
-    When a PROVIDER_READY handler is added
-    Then the PROVIDER_READY handler must run

From 86dd9fdcc912f5bc0c397e7478daeaed8d3e536c Mon Sep 17 00:00:00 2001
From: Simon Schrottner <simon.schrottner@dynatrace.com>
Date: Wed, 11 Dec 2024 15:23:52 +0100
Subject: [PATCH 6/9] feat: better reconnect gherkins

Signed-off-by: Simon Schrottner <simon.schrottner@dynatrace.com>
---
 .../contrib/provider/flagd/config.py          |  14 +-
 .../contrib/provider/flagd/provider.py        |   4 +-
 .../contrib/provider/flagd/resolvers/grpc.py  | 149 ++++++++++--------
 .../flagd/resolvers/process/file_watcher.py   |   4 +-
 .../tests/e2e/conftest.py                     |  21 ---
 .../tests/e2e/steps.py                        |  64 +++++++-
 .../tests/e2e/test_config.py                  |   9 +-
 .../tests/e2e/test_in_process_file.py         |   5 +
 8 files changed, 164 insertions(+), 106 deletions(-)

diff --git a/providers/openfeature-provider-flagd/src/openfeature/contrib/provider/flagd/config.py b/providers/openfeature-provider-flagd/src/openfeature/contrib/provider/flagd/config.py
index bcd4da85..f2eac5ec 100644
--- a/providers/openfeature-provider-flagd/src/openfeature/contrib/provider/flagd/config.py
+++ b/providers/openfeature-provider-flagd/src/openfeature/contrib/provider/flagd/config.py
@@ -26,7 +26,7 @@ class CacheType(Enum):
 DEFAULT_RESOLVER_TYPE = ResolverType.RPC
 DEFAULT_RETRY_BACKOFF = 1000
 DEFAULT_RETRY_BACKOFF_MAX = 120000
-DEFAULT_RETRY_GRACE_ATTEMPTS = 5
+DEFAULT_RETRY_GRACE_PERIOD = 5
 DEFAULT_STREAM_DEADLINE = 600000
 DEFAULT_TLS = False
 
@@ -41,7 +41,7 @@ class CacheType(Enum):
 ENV_VAR_RESOLVER_TYPE = "FLAGD_RESOLVER"
 ENV_VAR_RETRY_BACKOFF_MS = "FLAGD_RETRY_BACKOFF_MS"
 ENV_VAR_RETRY_BACKOFF_MAX_MS = "FLAGD_RETRY_BACKOFF_MAX_MS"
-ENV_VAR_RETRY_GRACE_ATTEMPTS = "FLAGD_RETRY_GRACE_ATTEMPTS"
+ENV_VAR_RETRY_GRACE_PERIOD = "FLAGD_RETRY_GRACE_PERIOD"
 ENV_VAR_STREAM_DEADLINE_MS = "FLAGD_STREAM_DEADLINE_MS"
 ENV_VAR_TLS = "FLAGD_TLS"
 
@@ -81,7 +81,7 @@ def __init__(  # noqa: PLR0913
         offline_poll_interval_ms: typing.Optional[int] = None,
         retry_backoff_ms: typing.Optional[int] = None,
         retry_backoff_max_ms: typing.Optional[int] = None,
-        retry_grace_attempts: typing.Optional[int] = None,
+        retry_grace_period: typing.Optional[int] = None,
         deadline_ms: typing.Optional[int] = None,
         stream_deadline_ms: typing.Optional[int] = None,
         keep_alive_time: typing.Optional[int] = None,
@@ -115,14 +115,14 @@ def __init__(  # noqa: PLR0913
             else retry_backoff_max_ms
         )
 
-        self.retry_grace_attempts: int = (
+        self.retry_grace_period: int = (
             int(
                 env_or_default(
-                    ENV_VAR_RETRY_GRACE_ATTEMPTS, DEFAULT_RETRY_GRACE_ATTEMPTS, cast=int
+                    ENV_VAR_RETRY_GRACE_PERIOD, DEFAULT_RETRY_GRACE_PERIOD, cast=int
                 )
             )
-            if retry_grace_attempts is None
-            else retry_grace_attempts
+            if retry_grace_period is None
+            else retry_grace_period
         )
 
         self.resolver = (
diff --git a/providers/openfeature-provider-flagd/src/openfeature/contrib/provider/flagd/provider.py b/providers/openfeature-provider-flagd/src/openfeature/contrib/provider/flagd/provider.py
index 07e148e1..dd8beeab 100644
--- a/providers/openfeature-provider-flagd/src/openfeature/contrib/provider/flagd/provider.py
+++ b/providers/openfeature-provider-flagd/src/openfeature/contrib/provider/flagd/provider.py
@@ -53,7 +53,7 @@ def __init__(  # noqa: PLR0913
         cache_type: typing.Optional[CacheType] = None,
         max_cache_size: typing.Optional[int] = None,
         retry_backoff_max_ms: typing.Optional[int] = None,
-        retry_grace_attempts: typing.Optional[int] = None,
+        retry_grace_period: typing.Optional[int] = None,
     ):
         """
         Create an instance of the FlagdProvider
@@ -84,7 +84,7 @@ def __init__(  # noqa: PLR0913
             deadline_ms=deadline,
             retry_backoff_ms=retry_backoff_ms,
             retry_backoff_max_ms=retry_backoff_max_ms,
-            retry_grace_attempts=retry_grace_attempts,
+            retry_grace_period=retry_grace_period,
             resolver=resolver_type,
             offline_flag_source_path=offline_flag_source_path,
             stream_deadline_ms=stream_deadline_ms,
diff --git a/providers/openfeature-provider-flagd/src/openfeature/contrib/provider/flagd/resolvers/grpc.py b/providers/openfeature-provider-flagd/src/openfeature/contrib/provider/flagd/resolvers/grpc.py
index 8dfda518..9502a72d 100644
--- a/providers/openfeature-provider-flagd/src/openfeature/contrib/provider/flagd/resolvers/grpc.py
+++ b/providers/openfeature-provider-flagd/src/openfeature/contrib/provider/flagd/resolvers/grpc.py
@@ -7,6 +7,7 @@
 from cachebox import BaseCacheImpl, LRUCache
 from google.protobuf.json_format import MessageToDict
 from google.protobuf.struct_pb2 import Struct
+from grpc import ChannelConnectivity
 
 from openfeature.evaluation_context import EvaluationContext
 from openfeature.event import ProviderEventDetails
@@ -47,6 +48,7 @@ def __init__(
             [ProviderEventDetails], None
         ],
     ):
+        self.active = False
         self.config = config
         self.emit_provider_ready = emit_provider_ready
         self.emit_provider_error = emit_provider_error
@@ -57,26 +59,30 @@ def __init__(
             if self.config.cache == CacheType.LRU
             else None
         )
-        self.stub, self.channel = self._create_stub()
-        self.retry_backoff_seconds = config.retry_backoff_ms * 0.001
-        self.retry_backoff_max_seconds = config.retry_backoff_max_ms * 0.001
-        self.retry_grace_attempts = config.retry_grace_attempts
+
+        self.retry_grace_period = config.retry_grace_period
         self.streamline_deadline_seconds = config.stream_deadline_ms * 0.001
         self.deadline = config.deadline_ms * 0.001
         self.connected = False
-
-    def _create_stub(
-        self,
-    ) -> typing.Tuple[evaluation_pb2_grpc.ServiceStub, grpc.Channel]:
-        config = self.config
         channel_factory = grpc.secure_channel if config.tls else grpc.insecure_channel
-        channel = channel_factory(
+
+        # Create the channel with the service config
+        options = [
+            ("grpc.keepalive_time_ms", config.keep_alive_time),
+            ("grpc.initial_reconnect_backoff_ms", config.retry_backoff_ms),
+            ("grpc.max_reconnect_backoff_ms", config.retry_backoff_max_ms),
+            ("grpc.min_reconnect_backoff_ms", config.deadline_ms),
+        ]
+        self.channel = channel_factory(
             f"{config.host}:{config.port}",
-            options=(("grpc.keepalive_time_ms", config.keep_alive_time),),
+            options=options,
         )
-        stub = evaluation_pb2_grpc.ServiceStub(channel)
+        self.stub = evaluation_pb2_grpc.ServiceStub(self.channel)
+
+        self.thread: typing.Optional[threading.Thread] = None
+        self.timer: typing.Optional[threading.Timer] = None
 
-        return stub, channel
+        self.start_time = time.time()
 
     def initialize(self, evaluation_context: EvaluationContext) -> None:
         self.connect()
@@ -89,11 +95,12 @@ def shutdown(self) -> None:
 
     def connect(self) -> None:
         self.active = True
-        self.thread = threading.Thread(
-            target=self.listen, daemon=True, name="FlagdGrpcServiceWorkerThread"
-        )
-        self.thread.start()
 
+        # Run monitoring in a separate thread
+        self.monitor_thread = threading.Thread(
+            target=self.monitor, daemon=True, name="FlagdGrpcServiceMonitorThread"
+        )
+        self.monitor_thread.start()
         ## block until ready or deadline reached
         timeout = self.deadline + time.time()
         while not self.connected and time.time() < timeout:
@@ -105,32 +112,72 @@ def connect(self) -> None:
                 "Blocking init finished before data synced. Consider increasing startup deadline to avoid inconsistent evaluations."
             )
 
+    def monitor(self) -> None:
+        def state_change_callback(new_state: ChannelConnectivity) -> None:
+            logger.debug(f"gRPC state change: {new_state}")
+            if new_state == ChannelConnectivity.READY:
+                if not self.thread or not self.thread.is_alive():
+                    self.thread = threading.Thread(
+                        target=self.listen,
+                        daemon=True,
+                        name="FlagdGrpcServiceWorkerThread",
+                    )
+                    self.thread.start()
+
+                if self.timer and self.timer.is_alive():
+                    logger.debug("gRPC error timer expired")
+                    self.timer.cancel()
+
+            elif new_state == ChannelConnectivity.TRANSIENT_FAILURE:
+                # this is the failed reonnect attempt so we are going into stale
+                self.emit_provider_stale(
+                    ProviderEventDetails(
+                        message="gRPC sync disconnected, reconnecting",
+                    )
+                )
+                self.start_time = time.time()
+                # adding a timer, so we can emit the error event after time
+                self.timer = threading.Timer(self.retry_grace_period, self.emit_error)
+
+                logger.debug("gRPC error timer started")
+                self.timer.start()
+                self.connected = False
+
+        self.channel.subscribe(state_change_callback, try_to_connect=True)
+
+    def emit_error(self) -> None:
+        logger.debug("gRPC error emitted")
+        if self.cache:
+            self.cache.clear()
+        self.emit_provider_error(
+            ProviderEventDetails(
+                message="gRPC sync disconnected, reconnecting",
+                error_code=ErrorCode.GENERAL,
+            )
+        )
+
     def listen(self) -> None:
-        retry_delay = self.retry_backoff_seconds
+        logger.debug("gRPC starting listener thread")
         call_args = (
             {"timeout": self.streamline_deadline_seconds}
             if self.streamline_deadline_seconds > 0
             else {}
         )
-        retry_counter = 0
-        while self.active:
-            request = evaluation_pb2.EventStreamRequest()
+        call_args["wait_for_ready"] = True
+        request = evaluation_pb2.EventStreamRequest()
 
+        # defining a never ending loop to recreate the stream
+        while self.active:
             try:
                 logger.debug("Setting up gRPC sync flags connection")
                 for message in self.stub.EventStream(request, **call_args):
                     if message.type == "provider_ready":
-                        if not self.connected:
-                            self.emit_provider_ready(
-                                ProviderEventDetails(
-                                    message="gRPC sync connection established"
-                                )
+                        self.connected = True
+                        self.emit_provider_ready(
+                            ProviderEventDetails(
+                                message="gRPC sync connection established"
                             )
-                            self.connected = True
-                            retry_counter = 0
-                            # reset retry delay after successsful read
-                            retry_delay = self.retry_backoff_seconds
-
+                        )
                     elif message.type == "configuration_change":
                         data = MessageToDict(message)["data"]
                         self.handle_changed_flags(data)
@@ -138,48 +185,14 @@ def listen(self) -> None:
                     if not self.active:
                         logger.info("Terminating gRPC sync thread")
                         return
-            except grpc.RpcError as e:
-                logger.error(f"SyncFlags stream error, {e.code()=} {e.details()=}")
-                # re-create the stub if there's a connection issue - otherwise reconnect does not work as expected
-                self.stub, self.channel = self._create_stub()
+            except grpc.RpcError as e:  # noqa: PERF203
+                # although it seems like this error log is not interesting, without it, the retry is not working as expected
+                logger.debug(f"SyncFlags stream error, {e.code()=} {e.details()=}")
             except ParseError:
                 logger.exception(
                     f"Could not parse flag data using flagd syntax: {message=}"
                 )
 
-            self.connected = False
-            self.on_connection_error(retry_counter, retry_delay)
-
-            retry_delay = self.handle_retry(retry_counter, retry_delay)
-
-            retry_counter = retry_counter + 1
-
-    def handle_retry(self, retry_counter: int, retry_delay: float) -> float:
-        if retry_counter == 0:
-            logger.info("gRPC sync disconnected, reconnecting immediately")
-        else:
-            logger.info(f"gRPC sync disconnected, reconnecting in {retry_delay}s")
-            time.sleep(retry_delay)
-            retry_delay = min(1.1 * retry_delay, self.retry_backoff_max_seconds)
-        return retry_delay
-
-    def on_connection_error(self, retry_counter: int, retry_delay: float) -> None:
-        if retry_counter == self.retry_grace_attempts:
-            if self.cache:
-                self.cache.clear()
-            self.emit_provider_error(
-                ProviderEventDetails(
-                    message=f"gRPC sync disconnected, reconnecting in {retry_delay}s",
-                    error_code=ErrorCode.GENERAL,
-                )
-            )
-        elif retry_counter == 1:
-            self.emit_provider_stale(
-                ProviderEventDetails(
-                    message=f"gRPC sync disconnected, reconnecting in {retry_delay}s",
-                )
-            )
-
     def handle_changed_flags(self, data: typing.Any) -> None:
         changed_flags = list(data["flags"].keys())
 
diff --git a/providers/openfeature-provider-flagd/src/openfeature/contrib/provider/flagd/resolvers/process/file_watcher.py b/providers/openfeature-provider-flagd/src/openfeature/contrib/provider/flagd/resolvers/process/file_watcher.py
index 0918981f..74835f94 100644
--- a/providers/openfeature-provider-flagd/src/openfeature/contrib/provider/flagd/resolvers/process/file_watcher.py
+++ b/providers/openfeature-provider-flagd/src/openfeature/contrib/provider/flagd/resolvers/process/file_watcher.py
@@ -31,17 +31,19 @@ def __init__(
         self.last_modified = 0.0
         self.flag_data: typing.Mapping[str, Flag] = {}
         self.load_data()
+        self.active = True
         self.thread = threading.Thread(target=self.refresh_file, daemon=True)
         self.thread.start()
 
     def shutdown(self) -> None:
+        self.active = False
         pass
 
     def get_flag(self, key: str) -> typing.Optional[Flag]:
         return self.flag_data.get(key)
 
     def refresh_file(self) -> None:
-        while True:
+        while self.active:
             time.sleep(self.poll_interval_seconds)
             logger.debug("checking for new flag store contents from file")
             last_modified = os.path.getmtime(self.file_path)
diff --git a/providers/openfeature-provider-flagd/tests/e2e/conftest.py b/providers/openfeature-provider-flagd/tests/e2e/conftest.py
index 077e6926..e80eb15b 100644
--- a/providers/openfeature-provider-flagd/tests/e2e/conftest.py
+++ b/providers/openfeature-provider-flagd/tests/e2e/conftest.py
@@ -1,8 +1,5 @@
 import typing
 
-import pytest
-from testcontainers.core.container import DockerContainer
-from tests.e2e.flagd_container import FlagdContainer
 from tests.e2e.steps import *  # noqa: F403
 
 JsonPrimitive = typing.Union[str, bool, float, int]
@@ -18,21 +15,3 @@ def pytest_collection_modifyitems(config):
     # this seems to not work with python 3.8
     if hasattr(config.option, "markexpr") and config.option.markexpr == "":
         config.option.markexpr = marker
-
-
-@pytest.fixture(autouse=True, scope="module")
-def setup(request, port, image):
-    container: DockerContainer = FlagdContainer(
-        image=image,
-        port=port,
-    )
-    # Setup code
-    c = container.start()
-
-    def fin():
-        c.stop()
-
-    # Teardown code
-    request.addfinalizer(fin)
-
-    return c.get_exposed_port(port)
diff --git a/providers/openfeature-provider-flagd/tests/e2e/steps.py b/providers/openfeature-provider-flagd/tests/e2e/steps.py
index c81c8871..a48ad2f3 100644
--- a/providers/openfeature-provider-flagd/tests/e2e/steps.py
+++ b/providers/openfeature-provider-flagd/tests/e2e/steps.py
@@ -1,10 +1,13 @@
 import logging
+import threading
 import time
 import typing
 
 import pytest
 from asserts import assert_equal, assert_in, assert_not_equal, assert_true
 from pytest_bdd import given, parsers, then, when
+from testcontainers.core.container import DockerContainer
+from tests.e2e.flagd_container import FlagdContainer
 from tests.e2e.parsers import to_bool, to_list
 
 from openfeature import api
@@ -26,9 +29,21 @@ def evaluation_context() -> EvaluationContext:
 
 @given("a flagd provider is set", target_fixture="client")
 @given("a provider is registered", target_fixture="client")
-def setup_provider(setup, resolver_type, client_name) -> OpenFeatureClient:
+def setup_provider(
+    container: FlagdContainer, resolver_type, client_name, port
+) -> OpenFeatureClient:
+    try:
+        container.get_exposed_port(port)
+    except:  # noqa: E722
+        container.start()
+
     api.set_provider(
-        FlagdProvider(resolver_type=resolver_type, port=setup, timeout=1),
+        FlagdProvider(
+            resolver_type=resolver_type,
+            port=int(container.get_exposed_port(port)),
+            timeout=1,
+            retry_grace_period=3,
+        ),
         client_name,
     )
     client = api.get_client(client_name)
@@ -517,6 +532,12 @@ def error_handles() -> list:
     return []
 
 
+@given(
+    parsers.cfparse(
+        "a {event_type:ProviderEvent} handler is added",
+        extra_types={"ProviderEvent": ProviderEvent},
+    ),
+)
 @when(
     parsers.cfparse(
         "a {event_type:ProviderEvent} handler is added",
@@ -591,7 +612,7 @@ def assert_handlers(
     )
 )
 def assert_handler_run(event_type: ProviderEvent, event_handles):
-    assert_handlers(event_handles, event_type, max_wait=30)
+    assert_handlers(event_handles, event_type, max_wait=20)
 
 
 @then(
@@ -602,7 +623,7 @@ def assert_handler_run(event_type: ProviderEvent, event_handles):
 )
 def assert_disconnect_handler(error_handles, event_type: ProviderEvent):
     # docker sync upstream restarts every 5s, waiting 2 cycles reduces test noise
-    assert_handlers(error_handles, event_type, max_wait=30)
+    assert_handlers(error_handles, event_type, max_wait=20)
 
 
 @when(
@@ -631,7 +652,10 @@ def assert_disconnect_error(
 def assert_flag_changed(event_handles, key):
     handle = None
     for h in event_handles:
-        if h["type"] == ProviderEvent.PROVIDER_CONFIGURATION_CHANGED:
+        if (
+            h["type"] == ProviderEvent.PROVIDER_CONFIGURATION_CHANGED
+            and key in h["event"].flags_changed
+        ):
             handle = h
             break
 
@@ -668,3 +692,33 @@ def flagd_init(client: OpenFeatureClient, event_handles, error_handles):
 @then("an error should be indicated within the configured deadline")
 def flagd_error(error_handles):
     assert_handlers(error_handles, ProviderEvent.PROVIDER_ERROR)
+
+
+@when(parsers.cfparse("the connection is lost for {seconds}s"))
+def flagd_restart(seconds, container):
+    def starting():
+        container.start()
+
+    container.stop()
+    threading.Timer(int(seconds), starting).start()
+
+
+@pytest.fixture(autouse=True, scope="module")
+def container(request, port, image):
+    container: DockerContainer = FlagdContainer(
+        image=image,
+        port=port,
+    )
+    # Setup code
+    container = container.start()
+
+    def fin():
+        try:
+            container.stop()
+        except:  # noqa: E722
+            logging.debug("container was not running anymore")
+
+    # Teardown code
+    request.addfinalizer(fin)
+
+    return container
diff --git a/providers/openfeature-provider-flagd/tests/e2e/test_config.py b/providers/openfeature-provider-flagd/tests/e2e/test_config.py
index dc413aee..238112d2 100644
--- a/providers/openfeature-provider-flagd/tests/e2e/test_config.py
+++ b/providers/openfeature-provider-flagd/tests/e2e/test_config.py
@@ -37,8 +37,13 @@ def convert_resolver_type(val: typing.Union[str, ResolverType]) -> ResolverType:
 }
 
 
-@pytest.fixture(autouse=True, scope="module")
-def setup(request):
+@pytest.fixture(autouse=True)
+def container():
+    pass
+
+
+@pytest.fixture(autouse=True)
+def setup_provider(request):
     pass
 
 
diff --git a/providers/openfeature-provider-flagd/tests/e2e/test_in_process_file.py b/providers/openfeature-provider-flagd/tests/e2e/test_in_process_file.py
index f73dc990..278bd1ea 100644
--- a/providers/openfeature-provider-flagd/tests/e2e/test_in_process_file.py
+++ b/providers/openfeature-provider-flagd/tests/e2e/test_in_process_file.py
@@ -62,6 +62,11 @@ def resolver_type() -> ResolverType:
     return ResolverType.IN_PROCESS
 
 
+@pytest.fixture(autouse=True)
+def container():
+    pass
+
+
 @pytest.fixture(autouse=True, scope="module")
 def setup(request, client_name, file_name, resolver_type):
     """nothing to boot"""

From e5133f69d617212cadb8dec98a2a2c4859c9e318 Mon Sep 17 00:00:00 2001
From: Simon Schrottner <simon.schrottner@dynatrace.com>
Date: Wed, 25 Dec 2024 21:57:34 +0100
Subject: [PATCH 7/9] fixup: unblock

Signed-off-by: Simon Schrottner <simon.schrottner@dynatrace.com>
---
 providers/openfeature-provider-flagd/openfeature/test-harness | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/providers/openfeature-provider-flagd/openfeature/test-harness b/providers/openfeature-provider-flagd/openfeature/test-harness
index e908fb7d..706a7e95 160000
--- a/providers/openfeature-provider-flagd/openfeature/test-harness
+++ b/providers/openfeature-provider-flagd/openfeature/test-harness
@@ -1 +1 @@
-Subproject commit e908fb7d19def6a4768ca90b02665075bbc1afbb
+Subproject commit 706a7e951bb72a145523b38fe83060becc34c4d7

From b91f73ec7fab24a8352b0ca142dbee24d6d6a396 Mon Sep 17 00:00:00 2001
From: Simon Schrottner <simon.schrottner@dynatrace.com>
Date: Wed, 25 Dec 2024 22:01:11 +0100
Subject: [PATCH 8/9] fixup: incuberating feedback from pr review

Signed-off-by: Simon Schrottner <simon.schrottner@dynatrace.com>
---
 .../contrib/provider/flagd/config.py          |  8 ++-
 .../contrib/provider/flagd/resolvers/grpc.py  | 56 +++++++++----------
 2 files changed, 33 insertions(+), 31 deletions(-)

diff --git a/providers/openfeature-provider-flagd/src/openfeature/contrib/provider/flagd/config.py b/providers/openfeature-provider-flagd/src/openfeature/contrib/provider/flagd/config.py
index f2eac5ec..1eb37463 100644
--- a/providers/openfeature-provider-flagd/src/openfeature/contrib/provider/flagd/config.py
+++ b/providers/openfeature-provider-flagd/src/openfeature/contrib/provider/flagd/config.py
@@ -26,7 +26,7 @@ class CacheType(Enum):
 DEFAULT_RESOLVER_TYPE = ResolverType.RPC
 DEFAULT_RETRY_BACKOFF = 1000
 DEFAULT_RETRY_BACKOFF_MAX = 120000
-DEFAULT_RETRY_GRACE_PERIOD = 5
+DEFAULT_RETRY_GRACE_PERIOD_SECONDS = 5
 DEFAULT_STREAM_DEADLINE = 600000
 DEFAULT_TLS = False
 
@@ -41,7 +41,7 @@ class CacheType(Enum):
 ENV_VAR_RESOLVER_TYPE = "FLAGD_RESOLVER"
 ENV_VAR_RETRY_BACKOFF_MS = "FLAGD_RETRY_BACKOFF_MS"
 ENV_VAR_RETRY_BACKOFF_MAX_MS = "FLAGD_RETRY_BACKOFF_MAX_MS"
-ENV_VAR_RETRY_GRACE_PERIOD = "FLAGD_RETRY_GRACE_PERIOD"
+ENV_VAR_RETRY_GRACE_PERIOD_SECONDS = "FLAGD_RETRY_GRACE_PERIOD"
 ENV_VAR_STREAM_DEADLINE_MS = "FLAGD_STREAM_DEADLINE_MS"
 ENV_VAR_TLS = "FLAGD_TLS"
 
@@ -118,7 +118,9 @@ def __init__(  # noqa: PLR0913
         self.retry_grace_period: int = (
             int(
                 env_or_default(
-                    ENV_VAR_RETRY_GRACE_PERIOD, DEFAULT_RETRY_GRACE_PERIOD, cast=int
+                    ENV_VAR_RETRY_GRACE_PERIOD_SECONDS,
+                    DEFAULT_RETRY_GRACE_PERIOD_SECONDS,
+                    cast=int,
                 )
             )
             if retry_grace_period is None
diff --git a/providers/openfeature-provider-flagd/src/openfeature/contrib/provider/flagd/resolvers/grpc.py b/providers/openfeature-provider-flagd/src/openfeature/contrib/provider/flagd/resolvers/grpc.py
index 9502a72d..5841912c 100644
--- a/providers/openfeature-provider-flagd/src/openfeature/contrib/provider/flagd/resolvers/grpc.py
+++ b/providers/openfeature-provider-flagd/src/openfeature/contrib/provider/flagd/resolvers/grpc.py
@@ -113,37 +113,37 @@ def connect(self) -> None:
             )
 
     def monitor(self) -> None:
-        def state_change_callback(new_state: ChannelConnectivity) -> None:
-            logger.debug(f"gRPC state change: {new_state}")
-            if new_state == ChannelConnectivity.READY:
-                if not self.thread or not self.thread.is_alive():
-                    self.thread = threading.Thread(
-                        target=self.listen,
-                        daemon=True,
-                        name="FlagdGrpcServiceWorkerThread",
-                    )
-                    self.thread.start()
-
-                if self.timer and self.timer.is_alive():
-                    logger.debug("gRPC error timer expired")
-                    self.timer.cancel()
-
-            elif new_state == ChannelConnectivity.TRANSIENT_FAILURE:
-                # this is the failed reonnect attempt so we are going into stale
-                self.emit_provider_stale(
-                    ProviderEventDetails(
-                        message="gRPC sync disconnected, reconnecting",
-                    )
+        self.channel.subscribe(self._state_change_callback, try_to_connect=True)
+
+    def _state_change_callback(self, new_state: ChannelConnectivity) -> None:
+        logger.debug(f"gRPC state change: {new_state}")
+        if new_state == ChannelConnectivity.READY:
+            if not self.thread or not self.thread.is_alive():
+                self.thread = threading.Thread(
+                    target=self.listen,
+                    daemon=True,
+                    name="FlagdGrpcServiceWorkerThread",
                 )
-                self.start_time = time.time()
-                # adding a timer, so we can emit the error event after time
-                self.timer = threading.Timer(self.retry_grace_period, self.emit_error)
+                self.thread.start()
 
-                logger.debug("gRPC error timer started")
-                self.timer.start()
-                self.connected = False
+            if self.timer and self.timer.is_alive():
+                logger.debug("gRPC error timer expired")
+                self.timer.cancel()
 
-        self.channel.subscribe(state_change_callback, try_to_connect=True)
+        elif new_state == ChannelConnectivity.TRANSIENT_FAILURE:
+            # this is the failed reconnect attempt so we are going into stale
+            self.emit_provider_stale(
+                ProviderEventDetails(
+                    message="gRPC sync disconnected, reconnecting",
+                )
+            )
+            self.start_time = time.time()
+            # adding a timer, so we can emit the error event after time
+            self.timer = threading.Timer(self.retry_grace_period, self.emit_error)
+
+            logger.debug("gRPC error timer started")
+            self.timer.start()
+            self.connected = False
 
     def emit_error(self) -> None:
         logger.debug("gRPC error emitted")

From fe1df5df62aa2db34069914b8a6967dcee125a6b Mon Sep 17 00:00:00 2001
From: Simon Schrottner <simon.schrottner@dynatrace.com>
Date: Wed, 25 Dec 2024 22:01:11 +0100
Subject: [PATCH 9/9] fixup: incuberating feedback from pr review

Signed-off-by: Simon Schrottner <simon.schrottner@dynatrace.com>
---
 providers/openfeature-provider-flagd/tests/e2e/steps.py | 9 +++------
 1 file changed, 3 insertions(+), 6 deletions(-)

diff --git a/providers/openfeature-provider-flagd/tests/e2e/steps.py b/providers/openfeature-provider-flagd/tests/e2e/steps.py
index a48ad2f3..b1325aff 100644
--- a/providers/openfeature-provider-flagd/tests/e2e/steps.py
+++ b/providers/openfeature-provider-flagd/tests/e2e/steps.py
@@ -612,7 +612,7 @@ def assert_handlers(
     )
 )
 def assert_handler_run(event_type: ProviderEvent, event_handles):
-    assert_handlers(event_handles, event_type, max_wait=20)
+    assert_handlers(event_handles, event_type, max_wait=30)
 
 
 @then(
@@ -623,7 +623,7 @@ def assert_handler_run(event_type: ProviderEvent, event_handles):
 )
 def assert_disconnect_handler(error_handles, event_type: ProviderEvent):
     # docker sync upstream restarts every 5s, waiting 2 cycles reduces test noise
-    assert_handlers(error_handles, event_type, max_wait=20)
+    assert_handlers(error_handles, event_type, max_wait=30)
 
 
 @when(
@@ -674,10 +674,7 @@ def wait_for(pred, poll_sec=2, timeout_sec=10):
 @given("flagd is unavailable", target_fixture="client")
 def flagd_unavailable(resolver_type):
     api.set_provider(
-        FlagdProvider(
-            resolver_type=resolver_type,
-            port=99999,
-        ),
+        FlagdProvider(resolver_type=resolver_type, port=99999, retry_grace_period=2),
         "unavailable",
     )
     return api.get_client("unavailable")