DM-43852: Get xsrf token from each redirect

CHANGELOG.md

@@ -2,7 +2,22 @@
 
 <!-- scriv-insert-here -->
 
+<a id='changelog-0.12.0'></a>
+
+## 0.12.0 (2024-05-15)
+
+### New features
+
+- Create Gafaelfawr service tokens instead of user tokens for authenticated calls to JupyterHub and JupyterLab. Gafaelfawr is standardizing on the new service token type for all service-to-service authentication.
+
+- Reduced the frequency of keep alive tasks for the Noteburst workers to once every 15 minutes, from once every 5 minutes. This is intended to clean up the logging output.
+
+### Bug fixes
+
+- Correctly extract cookies from the middle of the redirect chain caused by initial authentication to a Nublado lab. This fixes failures seen with labs containing JupyterHub 4.1.3.
+
 <a id='changelog-0.11.0'></a>
+
 ## 0.11.0 (2024-04-24)
 
 ### Bug fixes

src/noteburst/jupyterclient/jupyterlab.py

@@ -14,7 +14,7 @@
 
 import httpx
 import websockets
-from httpx import Cookies, Timeout
+from httpx import Cookies, Response, Timeout
 from pydantic import BaseModel, Field
 from structlog import BoundLogger
 from websockets.client import WebSocketClientProtocol
@@ -463,6 +463,23 @@ def url_for_websocket(self, path: str) -> str:
         http_url = self.url_for(path)
         return urlparse(http_url)._replace(scheme="wss").geturl()
 
+    def _extract_xsrf(self, response: Response) -> str | None:
+        """Extract the XSRF token from the cookies in a response.
+
+        Parameters
+        ----------
+        response
+            Response from a Jupyter server.
+
+        Returns
+        -------
+        str or None
+            Extracted XSRF value or `None` if none was present.
+        """
+        cookies = Cookies()
+        cookies.extract_cookies(response)
+        return cookies.get("_xsrf")
+
     async def log_into_hub(self) -> None:
         """Log into JupyterHub or raise a JupyterError."""
         self.logger.debug("Logging into JupyterHub")
@@ -472,25 +489,42 @@ async def log_into_hub(self) -> None:
         # to set cookies.
         if r.status_code >= 400:
             raise JupyterError.from_response(self.user.username, r)
-        cookies = Cookies()
-        cookies.extract_cookies(r)
-        xsrf = cookies.get("_xsrf")
+        xsrf = self._extract_xsrf(r)
         if xsrf:
             self._hub_xsrf = xsrf
 
     async def log_into_lab(self) -> None:
         """Log into JupyterLab or raise a JupyterError."""
         self.logger.debug("Logging into JupyterLab")
+        url = self.url_for(f"user/{self.user.username}/lab")
+        # Setting ``Sec-Fetch-Mode`` is not currently required, but it
+        # suppresses an annoying error message in the lab logs.
+        headers = {"Sec-Fetch-Mode": "navigate"}
         r = await self.http_client.get(
-            self.url_for(f"user/{self.user.username}/lab")
+            url, headers=headers, follow_redirects=False
         )
-        if r.status_code != 200:
-            raise JupyterError.from_response(self.user.username, r)
-        cookies = Cookies()
-        cookies.extract_cookies(r)
-        xsrf = cookies.get("_xsrf")
-        if xsrf:
+        while r.is_redirect:
+            xsrf = self._extract_xsrf(r)
+            if xsrf and xsrf != self._hub_xsrf:
+                self._lab_xsrf = xsrf
+            next_url = urljoin(url, r.headers["Location"])
+            r = await self.http_client.get(
+                next_url, headers=headers, follow_redirects=False
+            )
+        r.raise_for_status()
+        xsrf = self._extract_xsrf(r)
+        if xsrf and xsrf != self._hub_xsrf:
             self._lab_xsrf = xsrf
+        if not self._lab_xsrf:
+            raise JupyterError(
+                reason="No XSRF token found for JupyterLab",
+                url=url,
+                username=self.user.username,
+                status=r.status_code,
+                method="GET",
+                body=r.text,
+            )
+        self.logger.debug("Logged into JupyterLab with XSRF token")
 
     async def spawn_lab(self) -> JupyterImage:
         """Spawn a JupyterLab pod."""
@@ -629,6 +663,7 @@ async def open_lab_session(
         Send and receive messages from JupyterLab using the ``websocket``
         property on `JupyterLabSession`.
         """
+        self.logger.debug("Opening JupyterLab session")
         session_url = self.url_for(f"user/{self.user.username}/api/sessions")
         session_type = "notebook" if notebook_name else "console"
         body = {
@@ -640,6 +675,8 @@ async def open_lab_session(
         headers = {}
         if self._lab_xsrf:
             headers["X-XSRFToken"] = self._lab_xsrf
+        else:
+            self.logger.warning("No XSRF token found for JupyterLab.")
         r = await self.http_client.post(
             session_url, json=body, headers=headers
         )
@@ -726,6 +763,11 @@ async def execute_notebook(
         headers = {}
         if self._lab_xsrf:
             headers["X-XSRFToken"] = self._lab_xsrf
+        elif self._hub_xsrf:
+            self.logger.warning(
+                "No XSRF token found for JupyterLab, using hub token."
+            )
+            headers["X-XSRFToken"] = self._hub_xsrf
         try:
             # The timeout is designed to catch issues connecting to JupyterLab
             # but to wait as long as possible for the notebook itself

src/noteburst/jupyterclient/user.py

@@ -98,7 +98,6 @@ async def create(
             "username": username,
             "name": "Noteburst",
             "token_type": "service",
-            "token_name": f"noteburst {float(time.time())!s}",
             "scopes": scopes,
             "expires": int(time.time() + lifetime),
         }

src/noteburst/worker/main.py

@@ -208,7 +208,7 @@ async def shutdown(ctx: dict[Any, Any]) -> None:  # noqa: PLR0912
 elif config.worker_keepalive == WorkerKeepAliveSetting.normal:
     f = cron(
         keep_alive,
-        minute={0, 5, 10, 15, 20, 25, 30, 35, 40, 45, 50, 55},
+        minute={0, 15, 30, 45},
         unique=False,
     )
     cron_jobs.append(f)

tests/support/gafaelfawr.py

@@ -56,7 +56,6 @@ def handler(request: httpx.Request) -> httpx.Response:
         assert request_json == {
             "username": ANY,
             "token_type": "service",
-            "token_name": ANY,
             "scopes": ["exec:notebook"],
             "expires": ANY,
             "name": "Noteburst",
@@ -69,7 +68,6 @@ def handler(request: httpx.Request) -> httpx.Response:
             assert request_json["uid"] == uid
         if gid:
             assert request_json["gid"] == gid
-        assert request_json["token_name"].startswith("noteburst ")
         assert request_json["expires"] > time.time()
         response = {"token": make_gafaelfawr_token(request_json["username"])}
         return httpx.Response(200, json=response, request=request)