diff --git a/coco/client.py b/coco/client.py
index cb5e62b..e0cd3bb 100644
--- a/coco/client.py
+++ b/coco/client.py
@@ -46,7 +46,7 @@
 import yaml
 from aiohttp import ClientSession, ContentTypeError
 
-from . import metric, result
+from . import result
 from .config import DEFAULT_PORT, load_config
 from .endpoint import VALUE_TYPE
 
@@ -325,7 +325,6 @@ def client_send_request(
     # Extract useful information from the click context
     host = ctx.obj["host"]
     port = ctx.obj["port"]
-    metrics_port = ctx.obj["coco_config"]["metrics_port"]
 
     url = f"http://{host}:{port}/{path}"
 
@@ -346,11 +345,17 @@ def client_send_request(
 
     async def print_queue_size(metric_request_count):
         try:
-            q_size = await metric.get("coco_queue_length_total", metrics_port, host)
-        except RuntimeError as err:
-            if not isinstance(err, asyncio.CancelledError):
-                print(f"Couldn't get queue fill level from cocod: {err}")
+            async with ClientSession() as session:
+                async with session.get(f"http://{host}:{port}/qlen") as resp:
+                    result = await resp.text()
+                    q_size = int(result)
+        except asyncio.CancelledError:
+            # If cancelled, just return
             return
+        except (RuntimeError, ContentTypeError, KeyError) as err:
+            print(f"Couldn't get queue fill level from cocod: {err}")
+            return
+
         print(
             f"\rThere are {int(q_size)} requests in the queue.",
             sep=" ",
diff --git a/coco/core.py b/coco/core.py
index 84e0a10..4e93320 100644
--- a/coco/core.py
+++ b/coco/core.py
@@ -117,7 +117,6 @@ def __init__(
             "reset-state",
             "save-state",
             "load-state",
-            "get-coco-config",
             "wait",
         }
         all_endpoints |= all_local_endpoints
@@ -297,7 +296,10 @@ async def stop_slack_log(*_):
         self.sanic_app.register_listener(start_slack_log, "before_server_start")
         self.sanic_app.register_listener(stop_slack_log, "after_server_stop")
 
+        # non-endpoint routes.  These take precedence over an identically named
+        # endpoint
         self.sanic_app.add_route(self._get_config, "/config", methods=["GET"])
+        self.sanic_app.add_route(self._return_qlen, "/qlen", methods=["GET"])
 
         self.sanic_app.add_route(
             self.external_endpoint, "/<endpoint>", methods=["GET", "POST"]
@@ -483,6 +485,11 @@ async def _get_config(self, _):
         """
         return response.json(self.config)
 
+    async def _return_qlen(self, _):
+        from sanic import text
+
+        return text(str(self.redis_sync.llen("queue")))
+
     def _check_endpoint_links(self):
         def check(e):
             if e:
diff --git a/tests/coco_runner.py b/tests/coco_runner.py
index b1d7a7d..4d2f0de 100644
--- a/tests/coco_runner.py
+++ b/tests/coco_runner.py
@@ -35,6 +35,7 @@ def testN(coco_runner):
 import aiohttp
 import fakeredis
 import pytest
+import redis
 import yaml
 
 from coco import config
@@ -327,6 +328,22 @@ def start_daemon(self, *args):
                 # even after it successfully fetches the port.
                 sleep(0.2)
 
+    def redis_conn(self):
+        """Return a connection to the redis server.
+
+        Each returns a new 'redis.Redis' connection.  If the fakeredis server
+        isn't already running, it is started.
+        """
+
+        # Can't be called after stop()
+        if self._daemon_result:
+            raise RuntimeError("called after daemon stop.")
+
+        # Start the redis server, if necessary
+        redis_port = self._start_redis()
+
+        return redis.Redis(port=redis_port)
+
     def call_endpoint(
         self,
         endpoint: str,
@@ -363,6 +380,7 @@ def call_endpoint(
         # Can't be called after stop()
         if self._daemon_result:
             raise RuntimeError("called after daemon stop.")
+
         self.start_daemon()
 
         # Build the URL
diff --git a/tests/test_qlen.py b/tests/test_qlen.py
new file mode 100644
index 0000000..372e650
--- /dev/null
+++ b/tests/test_qlen.py
@@ -0,0 +1,61 @@
+"""Test monitoring the coco queue length."""
+
+import time
+
+
+def waiting_callback(path, data):
+    """Reply only after waiting."""
+
+    # Wait...
+    time.sleep(1)
+    return data
+
+
+def test_qlen(coco_runner):
+    """Test the /qlen route."""
+
+    # The "clean" targets don't wait.
+    # The "waiting" targets do.
+    coco_runner.add_targets("clean", 1)
+    coco_runner.add_targets("waiting", 1, callback=waiting_callback)
+
+    coco_runner.add_endpoint("waiting", {"group": "waiting"})
+    coco_runner.add_endpoint("clean", {"group": "clean"})
+    coco_runner.start_daemon()
+
+    # Check for clean daemon start
+    assert coco_runner.port is not None
+
+    # Queue starts empty
+    response, text = coco_runner.call_endpoint("qlen")
+    assert response.status == 200
+    assert int(text) == 0
+
+    # Connect to fakeredis
+    redis = coco_runner.redis_conn()
+
+    # Directly add some stuff to the queue
+    NUM_REQ = 7
+    for _ in range(NUM_REQ):
+        now = time.perf_counter()
+        tag = f"0-{now}"
+        redis.hset(
+            tag,
+            mapping={
+                "method": "GET",
+                "endpoint": "waiting",
+                "request": "",
+                "params": "",
+                "received": now,
+            },
+        )
+        redis.rpush("queue", tag)
+
+    # Check queue
+    response, text = coco_runner.call_endpoint("qlen")
+    assert response.status == 200
+    assert int(text) == NUM_REQ - 1  # The qworker popped one
+
+    # Can the client see this number?
+    result = coco_runner.client("clean")
+    assert "requests in the queue" in result.stdout