Kafka使用者都将assigned_partition设置为0，但在运行时分配不同的分区

Question

我不确定我是否做错了什么--但如果有人能帮助澄清这一点。

场景：

null

Listening on Topic : TOPIC_2PART
Status :True
ClientId : consumer_3_1
***************
  CONFIG 
***************
bootstrap_servers -> $node1_port
client_id -> consumer_3_1
group_id -> CGRP_TOPIC_2PART
value_deserializer -> <function CreateConsumer.consume.<locals>.<lambda> at 0x0000020175217DC8>
partition_assignment_strategy -> (<class 'kafka.coordinator.assignors.range.RangePartitionAssignor'>, <class 'kafka.coordinator.assignors.roundrobin.RoundRobinPartitionAssignor'>)
metric_group_prefix -> consumer
***************
  METRICS 
***************         
    consumer-fetch-manager-metrics
    **assigned-partitions->0.0**

Listening on Topic : TOPIC_2PART
Status :True
ClientId : consumer_3
***************
  CONFIG 
***************
bootstrap_servers -> $node1_port
client_id -> consumer_3
group_id -> CGRP_TOPIC_2PART
value_deserializer -> <function CreateConsumer.consume.<locals>.<lambda> at 0x00000118DC956DC8>
partition_assignment_strategy -> (<class 'kafka.coordinator.assignors.range.RangePartitionAssignor'>, <class 'kafka.coordinator.assignors.roundrobin.RoundRobinPartitionAssignor'>)
selector -> <class 'selectors.SelectSelector'>
***************
  METRICS 
***************
            kafka-metrics-count
            consumer-fetch-manager-metrics
            consumer-coordinator-metrics
    assigned-partitions->0.0

ConsumerRecord(topic='TOPIC_2PART', **partition=0**, offset=14, timestamp=1616346804116, timestamp_type=0, key=None, value={xxxx}, headers=[], checksum=None, serialized_key_size=-1, serialized_value_size=99, serialized_header_size=-1)

ConsumerRecord(topic='TOPIC_2PART', **partition=1**, offset=4, timestamp=1616346723367, timestamp_type=0, key=None, value={}, headers=[], checksum=None, serialized_key_size=-1, serialized_value_size=83, serialized_header_size=-1)
ConsumerRecord(topic='TOPIC_2PART', **partition=1**, offset=5, timestamp=1616346743556, timestamp_type=0, key=None, value={}, headers=[], checksum=None, serialized_key_size=-1, serialized_value_size=94, serialized_header_size=-1)

Answer 1

assigned-partitions指标是分配给每个使用者的分区数，而不是分区ID。对于两个使用者和两个分区，一旦订阅，每个使用者和分区的值都应该设置为1。所以问题是--订阅什么时候发生？似乎您粘贴的信息是在订阅前提取的。