From 71712df3202bb6a31d20756cbcaf972b8bfac09e Mon Sep 17 00:00:00 2001
From: Evan Lohn <evan@danswer.ai>
Date: Fri, 30 May 2025 15:13:38 -0400
Subject: [PATCH] jira daylight savings handling (#4797)

---
 backend/onyx/connectors/confluence/connector.py   | 15 ++++++++-------
 .../cross_connector_utils/miscellaneous_utils.py  |  4 ++++
 backend/onyx/connectors/onyx_jira/connector.py    | 15 +++++++++++++++
 3 files changed, 27 insertions(+), 7 deletions(-)

diff --git a/backend/onyx/connectors/confluence/connector.py b/backend/onyx/connectors/confluence/connector.py
index 208c3f5ae523..503cab542a65 100644
--- a/backend/onyx/connectors/confluence/connector.py
+++ b/backend/onyx/connectors/confluence/connector.py
@@ -21,6 +21,9 @@ from onyx.connectors.confluence.utils import datetime_from_string
 from onyx.connectors.confluence.utils import process_attachment
 from onyx.connectors.confluence.utils import update_param_in_path
 from onyx.connectors.confluence.utils import validate_attachment_filetype
+from onyx.connectors.cross_connector_utils.miscellaneous_utils import (
+    is_atlassian_date_error,
+)
 from onyx.connectors.exceptions import ConnectorValidationError
 from onyx.connectors.exceptions import CredentialExpiredError
 from onyx.connectors.exceptions import InsufficientPermissionsError
@@ -76,10 +79,6 @@ ONE_DAY = ONE_HOUR * 24
 MAX_CACHED_IDS = 100
 
 
-def _should_propagate_error(e: Exception) -> bool:
-    return "field 'updated' is invalid" in str(e)
-
-
 class ConfluenceCheckpoint(ConnectorCheckpoint):
 
     next_page_url: str | None
@@ -367,7 +366,7 @@ class ConfluenceConnector(
             )
         except Exception as e:
             logger.error(f"Error converting page {page.get('id', 'unknown')}: {e}")
-            if _should_propagate_error(e):
+            if is_atlassian_date_error(e):  # propagate error to be caught and retried
                 raise
             return ConnectorFailure(
                 failed_document=DocumentFailure(
@@ -446,7 +445,9 @@ class ConfluenceConnector(
                     f"Failed to extract/summarize attachment {attachment['title']}",
                     exc_info=e,
                 )
-                if _should_propagate_error(e):
+                if is_atlassian_date_error(
+                    e
+                ):  # propagate error to be caught and retried
                     raise
                 return ConnectorFailure(
                     failed_document=DocumentFailure(
@@ -536,7 +537,7 @@ class ConfluenceConnector(
         try:
             return self._fetch_document_batches(checkpoint, start, end)
         except Exception as e:
-            if _should_propagate_error(e) and start is not None:
+            if is_atlassian_date_error(e) and start is not None:
                 logger.warning(
                     "Confluence says we provided an invalid 'updated' field. This may indicate"
                     "a real issue, but can also appear during edge cases like daylight"
diff --git a/backend/onyx/connectors/cross_connector_utils/miscellaneous_utils.py b/backend/onyx/connectors/cross_connector_utils/miscellaneous_utils.py
index d32c4847466c..6eb82ea5f733 100644
--- a/backend/onyx/connectors/cross_connector_utils/miscellaneous_utils.py
+++ b/backend/onyx/connectors/cross_connector_utils/miscellaneous_utils.py
@@ -86,3 +86,7 @@ def get_oauth_callback_uri(base_domain: str, connector_id: str) -> str:
         # Used for development
         base_domain = CONNECTOR_LOCALHOST_OVERRIDE
     return f"{base_domain.strip('/')}/connector/oauth/callback/{connector_id}"
+
+
+def is_atlassian_date_error(e: Exception) -> bool:
+    return "field 'updated' is invalid" in str(e)
diff --git a/backend/onyx/connectors/onyx_jira/connector.py b/backend/onyx/connectors/onyx_jira/connector.py
index 3e29d3acd4f0..16bb271463e6 100644
--- a/backend/onyx/connectors/onyx_jira/connector.py
+++ b/backend/onyx/connectors/onyx_jira/connector.py
@@ -12,6 +12,9 @@ from onyx.configs.app_configs import INDEX_BATCH_SIZE
 from onyx.configs.app_configs import JIRA_CONNECTOR_LABELS_TO_SKIP
 from onyx.configs.app_configs import JIRA_CONNECTOR_MAX_TICKET_SIZE
 from onyx.configs.constants import DocumentSource
+from onyx.connectors.cross_connector_utils.miscellaneous_utils import (
+    is_atlassian_date_error,
+)
 from onyx.connectors.cross_connector_utils.miscellaneous_utils import time_str_to_utc
 from onyx.connectors.exceptions import ConnectorValidationError
 from onyx.connectors.exceptions import CredentialExpiredError
@@ -40,6 +43,8 @@ from onyx.utils.logger import setup_logger
 
 logger = setup_logger()
 
+ONE_HOUR = 3600
+
 JIRA_API_VERSION = os.environ.get("JIRA_API_VERSION") or "2"
 _JIRA_SLIM_PAGE_SIZE = 500
 _JIRA_FULL_PAGE_SIZE = 50
@@ -240,7 +245,17 @@ class JiraConnector(CheckpointedConnector[JiraConnectorCheckpoint], SlimConnecto
         checkpoint: JiraConnectorCheckpoint,
     ) -> CheckpointOutput[JiraConnectorCheckpoint]:
         jql = self._get_jql_query(start, end)
+        try:
+            return self._load_from_checkpoint(jql, checkpoint)
+        except Exception as e:
+            if is_atlassian_date_error(e):
+                jql = self._get_jql_query(start - ONE_HOUR, end)
+                return self._load_from_checkpoint(jql, checkpoint)
+            raise e
 
+    def _load_from_checkpoint(
+        self, jql: str, checkpoint: JiraConnectorCheckpoint
+    ) -> CheckpointOutput[JiraConnectorCheckpoint]:
         # Get the current offset from checkpoint or start at 0
         starting_offset = checkpoint.offset or 0
         current_offset = starting_offset