tests: added test validating consuming from empty log

Added test that validates if a consumer is able to continue consuming a
log that has been completely removed by delete retention.

Signed-off-by: Michal Maslanka <michal@redpanda.com>

tests/rptest/tests/log_segment_ms_test.py

@@ -6,6 +6,8 @@
 from rptest.clients.rpk import RpkTool
 from rptest.clients.types import TopicSpec
 from rptest.services.cluster import cluster
+from rptest.services.kafka_cli_consumer import KafkaCliConsumer
+
 from rptest.services.verifiable_consumer import VerifiableConsumer
 from rptest.services.verifiable_producer import VerifiableProducer
 from rptest.tests.redpanda_test import RedpandaTest
@@ -268,3 +270,93 @@ def test_segment_rolling_with_retention(self):
                    timeout_sec=120,
                    err_msg=f"Failed to consume messages")
         consumer.stop()
+
+    @cluster(num_nodes=4)
+    def test_segment_rolling_with_retention_consumer(self):
+        self.redpanda.set_cluster_config({
+            "log_segment_ms": None,
+            "log_segment_ms_min": 10000
+        })
+        topic = TopicSpec(segment_bytes=(1024 * 1024),
+                          replication_factor=1,
+                          partition_count=1)
+
+        self.client().create_topic(topic)
+
+        producer = VerifiableProducer(context=self.test_context,
+                                      num_nodes=1,
+                                      redpanda=self.redpanda,
+                                      topic=topic.name,
+                                      throughput=10000)
+
+        producer.start()
+        wait_until(
+            lambda: self._total_segments_count(topic) >= 5,
+            timeout_sec=120,
+            err_msg=
+            "producer failed to produce enough messages to create 5 segments")
+        # stop producer
+        producer.stop()
+        producer.clean()
+        producer.free()
+        del producer
+        start_count = self._total_segments_count(topic)
+        self.client().alter_topic_config(topic.name, "segment.ms", "15000")
+
+        # wait for the segment.ms policy to roll the segment
+        wait_until(lambda: self._total_segments_count(topic) > start_count,
+                   timeout_sec=60,
+                   err_msg=f"failed waiting for the segment to roll")
+        group_id = "test-group"
+
+        consumer = VerifiableConsumer(context=self.test_context,
+                                      num_nodes=1,
+                                      redpanda=self.redpanda,
+                                      topic=topic.name,
+                                      group_id=group_id)
+        consumer.start()
+
+        self.client().alter_topic_config(topic.name,
+                                         "retention.local.target.ms", "10000")
+        self.client().alter_topic_config(topic.name, "retention.ms", "10000")
+
+        # wait for retention policy to trigger
+        wait_until(lambda: self._total_segments_count(topic) <= 1,
+                   timeout_sec=60,
+                   err_msg=f"failed waiting for retention policy")
+        consumer.stop()
+
+        producer = VerifiableProducer(context=self.test_context,
+                                      num_nodes=1,
+                                      redpanda=self.redpanda,
+                                      topic=topic.name,
+                                      throughput=10000)
+        producer.start()
+        wait_until(
+            lambda: self._total_segments_count(topic) >= 2,
+            timeout_sec=120,
+            err_msg=
+            f"producer failed to produce enough messages to create 5 segments")
+        producer.stop()
+
+        rpk = RpkTool(self.redpanda)
+
+        def no_lag_present():
+            group = rpk.group_describe(group_id)
+            self.logger.info(f"group: {group}")
+            if len(group.partitions) != topic.partition_count:
+                return False
+
+            return all([p.lag == 0 for p in group.partitions])
+
+        consumer_2 = KafkaCliConsumer(context=self.test_context,
+                                      redpanda=self.redpanda,
+                                      topic=topic.name,
+                                      group=group_id)
+        consumer_2.start()
+        # # Wait for any messages to be consumed,
+        # # (if there is an issue in the offsets handling it will result in consumer being stuck)
+        wait_until(no_lag_present,
+                   timeout_sec=120,
+                   err_msg=f"Failed to consume all messages from {topic.name}")
+        consumer_2.stop()