diff --git a/app/glean/pipeline.py b/app/glean/pipeline.py
index 2cb3184..d6a99a6 100644
--- a/app/glean/pipeline.py
+++ b/app/glean/pipeline.py
@@ -528,7 +528,7 @@ def glean_dir(
     Pass ``force=True`` to bypass fingerprint checks and re-glean all files
     regardless of whether they have changed since the last run.
     """
-    files = sorted(corpus_dir.glob("*.jsonl")) + sorted(corpus_dir.glob("*.log"))
+    files = sorted(corpus_dir.rglob("*.jsonl")) + sorted(corpus_dir.rglob("*.log"))
     return _glean_files(files, db_path, pattern_file, batch_size, force=force)
 
 
diff --git a/scripts/gen_corpus.py b/scripts/gen_corpus.py
index fef20e9..bdc4056 100644
--- a/scripts/gen_corpus.py
+++ b/scripts/gen_corpus.py
@@ -284,7 +284,7 @@ def gen_docker(path: Path, start: datetime, end: datetime, rng: random.Random, e
             service  = rng.choice(_DOCKER_SERVICES)
             msg      = _pick_msg(_DOCKER_MESSAGES, severity, rng)
             entry = {
-                "SOURCE":  service,
+                "SOURCE":  f"docker:{service}",
                 "MESSAGE": msg,
             }
             fh.write(json.dumps(entry) + "\n")