diff --git a/tests/rptest/clients/rpk.py b/tests/rptest/clients/rpk.py
index 19937826703a..1b54591d004d 100644
--- a/tests/rptest/clients/rpk.py
+++ b/tests/rptest/clients/rpk.py
@@ -406,7 +406,6 @@ def int_or_none(value):
             self._redpanda.logger.error(f"Missing columns: {missing_columns}")
             raise RpkException(f"Missing columns: {missing_columns}")
 
-        partitions = []
         for row in table.rows:
             obj = dict()
             obj["LAST-STABLE-OFFSET"] = "-"
@@ -440,9 +439,7 @@ def int_or_none(value):
                                      start_offset=obj["LOG-START-OFFSET"])
 
             if initialized or tolerant:
-                partitions.append(partition)
-
-        return iter(partitions)
+                yield partition
 
     def describe_topic_configs(self, topic):
         cmd = ['describe', topic, '-c']
diff --git a/tests/rptest/tests/topic_delete_test.py b/tests/rptest/tests/topic_delete_test.py
index 98b97a1351c0..066d97a37c1b 100644
--- a/tests/rptest/tests/topic_delete_test.py
+++ b/tests/rptest/tests/topic_delete_test.py
@@ -472,7 +472,7 @@ def _populate_topic(self,
 
         # Wait for everything to be uploaded: this avoids tests potentially trying
         # to delete topics mid-uploads, which can leave orphan segments.
-        quiesce_uploads(self.redpanda, topic_name, timeout_sec=60)
+        quiesce_uploads(self.redpanda, [topic_name], timeout_sec=60)
 
     @skip_debug_mode  # Rely on timely uploads during leader transfers
     @cluster(num_nodes=3,
diff --git a/tests/rptest/utils/si_utils.py b/tests/rptest/utils/si_utils.py
index 9dc199abd91f..4e3c7560ca5e 100644
--- a/tests/rptest/utils/si_utils.py
+++ b/tests/rptest/utils/si_utils.py
@@ -466,7 +466,10 @@ def remote_has_reached_hwm(ntp: NTP, hwm: int):
     rpk = RpkTool(redpanda)
     for topic_name in topic_names:
         described = rpk.describe_topic(topic_name)
+        p_count = 0
         for p in described:
+            p_count += 1
+
             ntp = NTP(ns='kafka', topic=topic_name, partition=p.id)
             hwm = p.high_watermark
 
@@ -479,6 +482,11 @@ def remote_has_reached_hwm(ntp: NTP, hwm: int):
                                 backoff_sec=1)
             redpanda.logger.debug(f"Partition {ntp} ready (reached HWM {hwm})")
 
+        if p_count == 0:
+            # We expect to be called on a topic where `rpk topic describe` returns
+            # some data, otherwise we can't check that against cloud storage content
+            raise RuntimeError(f"Found 0 partitions for topic '{topic_name}'")
+
 
 @dataclass(order=True)
 class SpillMeta: