diff --git a/CHANGES/1696.bugfix b/CHANGES/1696.bugfix
new file mode 100644
index 000000000..d514781b7
--- /dev/null
+++ b/CHANGES/1696.bugfix
@@ -0,0 +1 @@
+Fixed the long accept header limit exceed during sync.
diff --git a/pulp_container/app/downloaders.py b/pulp_container/app/downloaders.py
index fb361198e..55d81158f 100644
--- a/pulp_container/app/downloaders.py
+++ b/pulp_container/app/downloaders.py
@@ -19,6 +19,11 @@
     ["status_code", "path", "artifact_attributes", "url", "headers"],
 )
 
+# Usually webservers ignore accept headers for blob endpoint
+# However, some webservers when serving blobs need to have explicitly specified octet-stream
+# in the accept headers
+V2_ACCEPT_HEADERS_OCTET = {"Accept": V2_ACCEPT_HEADERS["Accept"] + ",application/octet-stream"}
+
 
 class RegistryAuthHttpDownloader(HttpDownloader):
     """
@@ -57,11 +62,7 @@ async def _run(self, handle_401=True, extra_data=None):
         # these accept headers are going to be sent with every request to ensure downloader
         # can download manifests, namely in the repair core task
         # FIXME this can be rolledback after https://github.com/pulp/pulp_container/issues/1288
-        headers = V2_ACCEPT_HEADERS
-        # Usually webservers ignore accept headers for blob endpoint
-        # However, some webservers when serving blobs need to have explicitly specified octet-stream
-        # in the accept headers
-        headers["Accept"] = headers["Accept"] + ",application/octet-stream"
+        headers = V2_ACCEPT_HEADERS_OCTET
         repo_name = None
         if extra_data is not None:
             headers = extra_data.get("headers", headers)