tencent cloud

TencentDB for MongoDB

Node Overview

Download
フォーカスモード
フォントサイズ
最終更新日: 2026-05-09 16:02:05
TencentDB for MongoDB provides two deployment architectures: replica set and sharded cluster, designed to meet different business requirements such as high availability, read/write splitting, and horizontal scaling.
Replica Set Architecture: Each replica set instance contains a primary node, one or more secondary nodes, and a hidden node.
Sharded Cluster Architecture: A sharded cluster is a horizontal scaling solution built on replica sets. The cluster is composed of multiple shards, each of which is a replica set itself. Data is distributed across these different shards based on the shard key.
Node Role
Description
Remarks
Primary node
The primary node handles all data read and write requests.
Each replica set instance can have only one primary node.
Secondary node

Secondary nodes replicate data from the primary node by periodically polling its oplog (operation log) to maintain data consistency. If the primary node fails, an automatic election promotes a secondary node to primary, ensuring high availability.
When a client connects to a secondary node, it can only read data, not write data.
You can add more secondary nodes as needed. For details, see Add Secondary Node.
A secondary node can be promoted to a primary node. For specific operations, see Promote Secondary Node to Primary Node.
Hidden node

By default, one of the secondary nodes in each newly created instance is designated as the hidden node, serving as a hidden member that backs up data. When a secondary node fails, the hidden node can be promoted to replace the failed secondary node, ensuring high availability.
A replica set instance can have only one hidden node.
A secondary node that has been configured as the hidden node cannot be deleted.
A hidden node has its priority set to 0, making it ineligible for election as the primary. However, it still participates in the election voting process.
Read-only node

When you enable the read-only routing feature, the system designates one or more secondary nodes as read-only nodes. Read-only nodes are primarily designed for scenarios with massive read requests. They synchronize data from the primary node or secondary nodes through the oplog (operation log). The system automatically routes read requests to read-only nodes to reduce the load on the primary node.
A read-only node does not participate in primary election and cannot be elected as the primary node.
A replica set instance can have multiple read-only nodes. For details, see Add Read-Only Node.

ヘルプとサポート

この記事はお役に立ちましたか?

フィードバック