Merge branch 'develop' of https://github.com/CenterForOpenScience/osf.io into refactor-notifications

* 'develop' of https://github.com/CenterForOpenScience/osf.io:
  Bump version no. Add CHANGELOG
  move CROSSREF_UNAVAILABLE_DELAY to settings.py
  handle and 5xx status code from crossref
  handle 500 error from crossref

Author: Johnetordoff

CHANGELOG

@@ -2,6 +2,11 @@
 
 We follow the CalVer (https://calver.org/) versioning scheme: YY.MINOR.MICRO.
 
+25.11.1 (2025-07-03)
+====================
+
+- Hotfix to handle crossref's anticipated downtime
+
 25.11.0 (2025-06-30)
 ====================
 

osf/management/commands/sync_doi_metadata.py

@@ -7,6 +7,9 @@
 from django.core.management.base import BaseCommand
 from osf.models import GuidMetadataRecord, Identifier, Registration, Preprint
 from framework.celery_tasks import app
+from website.identifiers.clients.exceptions import CrossRefUnavailableError
+from website.settings import CROSSREF_UNAVAILABLE_DELAY
+
 
 logger = logging.getLogger(__name__)
 
@@ -21,6 +24,9 @@ def sync_identifier_doi(self, identifier_id):
         identifier.referent.request_identifier_update('doi')
         identifier.save()
         logger.info(f'Doi update for {identifier.value} complete')
+    except CrossRefUnavailableError as err:
+        logger.warning(f'CrossRef is unavailable during sync of {identifier.value} DOI. Error: {err.error}')
+        self.retry(countdown=CROSSREF_UNAVAILABLE_DELAY)
     except Exception as err:
         logger.warning(f'[{err.__class__.__name__}] Doi update for {identifier.value} failed because of error: {err}')
         self.retry()
@@ -75,6 +81,9 @@ def async_request_identifier_update(self, preprint_id):
     preprint = Preprint.load(preprint_id)
     try:
         preprint.request_identifier_update('doi', create=True)
+    except CrossRefUnavailableError as err:
+        logger.warning(f'CrossRef is unavailable during DOI update for preprint {preprint._id}. Error: {err.error}')
+        self.retry(countdown=CROSSREF_UNAVAILABLE_DELAY)
     except Exception as err:
         logger.warning(f'[{err.__class__.__name__}] Doi creation failed for the preprint with id {preprint._id} because of error: {err}')
         self.retry()

package.json

@@ -1,6 +1,6 @@
 {
   "name": "OSF",
-  "version": "25.11.0",
+  "version": "25.11.1",
   "description": "Facilitating Open Science",
   "repository": "https://github.com/CenterForOpenScience/osf.io",
   "author": "Center for Open Science",

website/identifiers/clients/crossref.py

@@ -7,7 +7,7 @@
 import requests
 from django.db.models import QuerySet
 
-from .exceptions import CrossRefRateLimitError
+from .exceptions import CrossRefRateLimitError, CrossRefUnavailableError
 from framework.auth.utils import impute_names
 from website.identifiers.utils import remove_control_characters
 from website.identifiers.clients.base import AbstractIdentifierClient
@@ -267,6 +267,9 @@ def create_identifier(self, preprint, category, include_relation=True):
             if response.status_code == 429:
                 raise CrossRefRateLimitError(response.text)
 
+            if response.status_code >= 500:
+                raise CrossRefUnavailableError(response.text)
+
             return {'doi': doi}
 
         raise NotImplementedError()

website/identifiers/clients/exceptions.py

@@ -16,3 +16,12 @@ def __init__(self, error):
 
     def __str__(self):
         return self.error
+
+
+class CrossRefUnavailableError(Exception):
+
+    def __init__(self, error):
+        self.error = error
+
+    def __str__(self):
+        return self.error

website/preprints/tasks.py

@@ -4,6 +4,8 @@
 from framework.exceptions import HTTPError
 from framework.celery_tasks import app as celery_app
 from framework.postcommit_tasks.handlers import enqueue_postcommit_task, get_task_from_postcommit_queue
+from website.identifiers.clients.exceptions import CrossRefUnavailableError
+from website.settings import CROSSREF_UNAVAILABLE_DELAY
 
 
 CROSSREF_FAIL_RETRY_DELAY = 12 * 60 * 60
@@ -68,6 +70,7 @@ def update_or_enqueue_on_preprint_updated(preprint_id, saved_fields=None):
 @celery_app.task(bind=True, acks_late=True, max_retries=5, default_retry_delay=CROSSREF_FAIL_RETRY_DELAY)
 def mint_doi_on_crossref_fail(self, preprint_id):
     from osf.models import Preprint
+
     preprint = Preprint.load(preprint_id)
     vg = preprint.versioned_guids.first()
     existing_versions_without_minted_doi = Preprint.objects.filter(
@@ -84,5 +87,9 @@ def mint_doi_on_crossref_fail(self, preprint_id):
         self.retry()
     else:
         crossref_client = preprint.get_doi_client()
-        if crossref_client:
-            crossref_client.create_identifier(preprint, category='doi', include_relation=False)
+        try:
+            if crossref_client:
+                crossref_client.create_identifier(preprint, category='doi', include_relation=False)
+        except CrossRefUnavailableError as err:
+            logger.warning(f'CrossRef is unavailable during minting DOI on fail for preprint {preprint_id}. Error: {err.error}')
+            self.retry(countdown=CROSSREF_UNAVAILABLE_DELAY)

website/settings/defaults.py

@@ -342,6 +342,7 @@ def parent_dir(path):
 CROSSREF_PASSWORD = None
 CROSSREF_URL = None  # Location to POST crossref data. In production, change this to the production CrossRef API endpoint
 CROSSREF_DEPOSITOR_EMAIL = 'None'  # This email will receive confirmation/error messages from CrossRef on submission
+CROSSREF_UNAVAILABLE_DELAY = 24 * 60 * 60
 
 ECSARXIV_CROSSREF_USERNAME = None
 ECSARXIV_CROSSREF_PASSWORD = None