From a21104d9c2562e07c6eabb7acb17b0d355b40cf0 Mon Sep 17 00:00:00 2001
From: hagen-danswer <hagen@danswer.ai>
Date: Wed, 20 Nov 2024 16:00:14 -0800
Subject: [PATCH] decreased confluence retry times and added more logging
 (#3184)

* decreased confluence retry times and added more logging

* added check on connector startup

* no retries!

* fr no retries
---
 .../connectors/confluence/connector.py        | 17 +++++++++++++++
 .../connectors/confluence/onyx_confluence.py  |  6 +++++-
 .../danswer/connectors/confluence/utils.py    |  2 +-
 .../confluence/group_sync.py                  | 21 +++++++++++++++++--
 4 files changed, 42 insertions(+), 4 deletions(-)

diff --git a/backend/danswer/connectors/confluence/connector.py b/backend/danswer/connectors/confluence/connector.py
index 5c58cf194fa..8d614c163c7 100644
--- a/backend/danswer/connectors/confluence/connector.py
+++ b/backend/danswer/connectors/confluence/connector.py
@@ -3,6 +3,8 @@
 from typing import Any
 from urllib.parse import quote
 
+from atlassian import Confluence  # type: ignore
+
 from danswer.configs.app_configs import CONFLUENCE_CONNECTOR_LABELS_TO_SKIP
 from danswer.configs.app_configs import CONTINUE_ON_CONNECTOR_FAILURE
 from danswer.configs.app_configs import INDEX_BATCH_SIZE
@@ -116,6 +118,21 @@ def load_credentials(self, credentials: dict[str, Any]) -> dict[str, Any] | None
             is_cloud=self.is_cloud,
             wiki_base=self.wiki_base,
         )
+
+        client_without_retries = Confluence(
+            api_version="cloud" if self.is_cloud else "latest",
+            url=self.wiki_base.rstrip("/"),
+            username=credentials["confluence_username"] if self.is_cloud else None,
+            password=credentials["confluence_access_token"] if self.is_cloud else None,
+            token=credentials["confluence_access_token"] if not self.is_cloud else None,
+        )
+        spaces = client_without_retries.get_all_spaces(limit=1)
+        if not spaces:
+            raise RuntimeError(
+                f"No spaces found at {self.wiki_base}! "
+                "Check your credentials and wiki_base and make sure "
+                "is_cloud is set correctly."
+            )
         return None
 
     def _get_comment_string_for_page_id(self, page_id: str) -> str:
diff --git a/backend/danswer/connectors/confluence/onyx_confluence.py b/backend/danswer/connectors/confluence/onyx_confluence.py
index c6a4d3857b1..4820429ba0f 100644
--- a/backend/danswer/connectors/confluence/onyx_confluence.py
+++ b/backend/danswer/connectors/confluence/onyx_confluence.py
@@ -84,7 +84,7 @@ def handle_confluence_rate_limit(confluence_call: F) -> F:
     def wrapped_call(*args: list[Any], **kwargs: Any) -> Any:
         MAX_RETRIES = 5
 
-        TIMEOUT = 3600
+        TIMEOUT = 600
         timeout_at = time.monotonic() + TIMEOUT
 
         for attempt in range(MAX_RETRIES):
@@ -99,6 +99,10 @@ def wrapped_call(*args: list[Any], **kwargs: Any) -> Any:
                 return confluence_call(*args, **kwargs)
             except HTTPError as e:
                 delay_until = _handle_http_error(e, attempt)
+                logger.warning(
+                    f"HTTPError in confluence call. "
+                    f"Retrying in {delay_until} seconds..."
+                )
                 while time.monotonic() < delay_until:
                     # in the future, check a signal here to exit
                     time.sleep(1)
diff --git a/backend/danswer/connectors/confluence/utils.py b/backend/danswer/connectors/confluence/utils.py
index 9deff36b269..cb5253f4c14 100644
--- a/backend/danswer/connectors/confluence/utils.py
+++ b/backend/danswer/connectors/confluence/utils.py
@@ -283,6 +283,6 @@ def build_confluence_client(
         password=credentials_json["confluence_access_token"] if is_cloud else None,
         token=credentials_json["confluence_access_token"] if not is_cloud else None,
         backoff_and_retry=True,
-        max_backoff_retries=60,
+        max_backoff_retries=10,
         max_backoff_seconds=60,
     )
diff --git a/backend/ee/danswer/external_permissions/confluence/group_sync.py b/backend/ee/danswer/external_permissions/confluence/group_sync.py
index fab2bde4ea0..17140b33f71 100644
--- a/backend/ee/danswer/external_permissions/confluence/group_sync.py
+++ b/backend/ee/danswer/external_permissions/confluence/group_sync.py
@@ -1,3 +1,5 @@
+from atlassian import Confluence  # type: ignore
+
 from danswer.connectors.confluence.onyx_confluence import OnyxConfluence
 from danswer.connectors.confluence.utils import build_confluence_client
 from danswer.connectors.confluence.utils import get_user_email_from_username__server
@@ -32,11 +34,26 @@ def _get_group_members_email_paginated(
 def confluence_group_sync(
     cc_pair: ConnectorCredentialPair,
 ) -> list[ExternalUserGroup]:
+    credentials = cc_pair.credential.credential_json
     is_cloud = cc_pair.connector.connector_specific_config.get("is_cloud", False)
+    wiki_base = cc_pair.connector.connector_specific_config["wiki_base"]
+
+    # test connection with direct client, no retries
+    confluence_client = Confluence(
+        api_version="cloud" if is_cloud else "latest",
+        url=wiki_base.rstrip("/"),
+        username=credentials["confluence_username"] if is_cloud else None,
+        password=credentials["confluence_access_token"] if is_cloud else None,
+        token=credentials["confluence_access_token"] if not is_cloud else None,
+    )
+    spaces = confluence_client.get_all_spaces(limit=1)
+    if not spaces:
+        raise RuntimeError(f"No spaces found at {wiki_base}!")
+
     confluence_client = build_confluence_client(
-        credentials_json=cc_pair.credential.credential_json,
+        credentials_json=credentials,
         is_cloud=is_cloud,
-        wiki_base=cc_pair.connector.connector_specific_config["wiki_base"],
+        wiki_base=wiki_base,
     )
 
     # Get all group names