Geo-replication is the replication of persistently stored message data across multiple clusters of a Pulsar instance.
How geo-replication works
下图说明了 Pulsar 在不同集群之间跨地域复制的过程：
In this diagram, whenever P1, P2, and P3 producers publish messages to the T1 topic on Cluster-A, Cluster-B, and Cluster-C clusters respectively, those messages are instantly replicated across clusters. Once the messages are replicated, C1 and C2 consumers can consume those messages from their respective clusters.
Without geo-replication, C1 and C2 consumers are not able to consume messages that P3 producer publishes.
跨地域复制和 Pulsar 属性
You must enable geo-replication on a per-tenant basis in Pulsar. You can enable geo-replication between clusters only when a tenant is created that allows access to both clusters.
Although geo-replication must be enabled between two clusters, actually geo-replication is managed at the namespace level. You must complete the following tasks to enable geo-replication for a namespace:
Any message published on any topic in that namespace is replicated to all clusters in the specified set.
When messages are produced on a Pulsar topic, messages are first persisted in the local cluster, and then forwarded asynchronously to the remote clusters.
In normal cases, when connectivity issues are none, messages are replicated immediately, at the same time as they are dispatched to local consumers. Typically, the network round-trip time (RTT) between the remote regions defines end-to-end delivery latency.
While producers and consumers can publish to and consume from any cluster in a Pulsar instance, subscriptions are local to the clusters in which the subscriptions are created and cannot be transferred between clusters. If you do need to transfer a subscription, you need to create a new subscription in the desired cluster.
In the aforementioned example, the T1 topic is replicated among three clusters, Cluster-A, Cluster-B, and Cluster-C.
这三个集群中的任何一个集群生成的所有消息都交付给其他集群中的所有订阅。 In this case, C1 and C2 consumers receive all messages that P1, P2, and P3 producers publish. 指令仍然在每个生产者的basis上有所保障。
Grant permissions to properties
To replicate to a cluster, the tenant needs permission to use that cluster. You can grant permission to the tenant when you create the tenant or grant later.
Specify all the intended clusters when you create a tenant:
bin/pulsar-admin tenants create my-tenant \ --admin-roles my-admin-role \ --allowed-clusters us-west,us-east,us-cent
bin/pulsar-admin namespaces create my-tenant/my-namespace
Initially, the namespace is not assigned to any cluster. You can assign the namespace to clusters using the
bin/pulsar-admin namespaces set-clusters my-tenant/my-namespace \ --clusters us-west,us-east,us-cent
You can change the replication clusters for a namespace at any time, without disruption to ongoing traffic. 一旦配置发生更改，复制通道将立即在所有集群中设置或停止。
Use topics with geo-replication
Once you create a geo-replication namespace, any topics that producers or consumers create within that namespace is replicated across clusters. Typically, each application uses the
serviceUrl for the local cluster.
By default, messages are replicated to all clusters configured for the namespace. You can restrict replication selectively by specifying a replication list for a message, and then that message is replicated only to the subset in the replication list.
List<String> restrictReplicationTo = Arrays.asList( "us-west", "us-east" ); Producer producer = client.newProducer() .topic("some-topic") .create(); producer.newMessage() .value("my-payload".getBytes()) .setReplicationClusters(restrictReplicationTo) .send();
bin/pulsar-admin persistent stats persistent://my-tenant/my-namespace/my-topic
Delete a geo-replication topic
Given that geo-replication topics exist in multiple regions, directly deleting a geo-replication topic is not possible. Instead, you should rely on automatic topic garbage collection.
In Pulsar, a topic is automatically deleted when the topic meets the following three conditions:
- no producers or consumers are connected to it;
- no subscriptions to it;
- no more messages are kept for retention. For geo-replication topics, each region uses a fault-tolerant mechanism to decide when deleting the topic locally is safe.
要删除跨域复制主题，请关闭该主题上的所有生产者和消费者，并删除每个复制集群中的所有本地订阅。 When Pulsar determines that no valid subscription for the topic remains across the system, it will garbage collect the topic.
Pulsar supports replicated subscriptions, so you can keep subscription state in sync, within a sub-second timeframe, in the context of a topic that is being asynchronously replicated across multiple geographical regions.
In case of failover, a consumer can restart consuming from the failure point in a different cluster.
Enable replicated subscription
Replicated subscription is disabled by default. You can enable replicated subscription when creating a consumer.
Consumer<String> consumer = client.newConsumer(Schema.STRING) .topic("my-topic") .subscriptionName("my-subscription") .replicateSubscriptionState(true) .subscribe();
- It is easy to implement the logic.
- You can choose to enable or disable replicated subscription.
- When you enable it, the overhead is low, and it is easy to configure.
- When you disable it, the overhead is zero.
When you enable replicated subscription, you're creating a consistent distributed snapshot to establish an association between message ids from different clusters. The snapshots are taken periodically. The default value is
1 second. It means that a consumer failing over to a different cluster can potentially receive 1 second of duplicates. You can also configure the frequency of the snapshot in the