在 Atlas 中管理在线存档

使用 Atlas CLI 检索在线存档

要使用 Atlas CLI 列出集群的所有在线存档，请运行以下命令：

atlas clusters onlineArchives list [options]

如需返回使用 Atlas CLI 指定的集群的详细信息，请运行以下命令：

atlas clusters onlineArchives describe <archiveId> [options]

要进一步了解前面命令的语法和参数，请参阅 atlas clusters onlineArchives list 和 atlas clusters onlineArchives describe 的 Atlas CLI 文档。

使用 API 检索在线存档

To retrieve an online archive through the API, send a GET request to the onlineArchives endpoint with the unique ID of the online archive to retrieve. To learn more about the API syntax and options, see API.

使用 API 检索集群的所有在线存档

To retrieve all the online archives configured for a cluster using the API, send a GET request to the onlineArchives endpoint for the cluster. To learn more about the syntax and options, see API.

在用户界面中查看在线存档

查看 Online Archive 列表：

该页面显示集群的在线存档。对于每个在线存档，可以看到以下信息：

通过 Atlas CLI 编辑存档规则

要使用 Atlas CLI 为集群更新在线存档，请运行以下命令：

atlas clusters onlineArchives update <archiveId> [options]

To learn more about the command syntax and parameters, see the Atlas CLI documentation for atlas clusters onlineArchives update.

通过 API 编辑存档规则

To edit an online archive through the API, send a PATCH request to the onlineArchives endpoint with the unique ID of the online archive to update. To learn more about the API syntax and options, see API.

通过用户界面编辑存档规则

要编辑在线存档，请在 Atlas 用户界面中：

编辑云对象存储上的分区

You can't modify the partition fields or structure from the Atlas UI or API. However, you can manually migrate the data from the cloud object storage using $merge, use $merge to rehydrate data back to Atlas cluster, delete the online archive, and then create a new online archive for the collection with the desired partition fields and structure.

连接字符串

To run queries, you must first connect to your Online Archive. Your cluster connection string allows you to only query data in your Atlas cluster. To query your Online Archive, you must use one of the following:

• 连接到在线存档和集群 — 此只读连接字符串允许直接从实时集群读取数据。这会影响 IOPS 的可用资源以及在线存档中的可用资源。

• 连接到在线存档 — 此只读连接字符串只允许从在线存档读取数据，不会影响集群资源。

性能考虑因素

一般来说，查询存档数据比查询 Atlas 集群上的数据要慢得多。当您通过联合连接字符串查询集群和 Online Archive 中的数据时：

• 阻止查询（例如在返回结果之前消耗并处理发送至排序操作的所有输入文档的排序）具有与最慢存储（正在查询的存档）相关的性能特征。排序操作要求在返回结果之前对来源中的所有数据进行查询。

• 流媒体查询，例如查找，具有与被查询的最高性能存储 Atlas cluster 相关的性能特征。Atlas 在结果可用后立即返回结果，这意味着从存档返回结果比从 Atlas 集群返回结果需要更长的时间。

查询价格

对于联合查询和仅存档查询，您需要为以下项目支付费用。

通过 Atlas CLI 删除在线存档

要使用 Atlas CLI 为集群删除在线存档，请运行以下命令：

atlas clusters onlineArchives delete <archiveId> [options]

To learn more about the command syntax and parameters, see the Atlas CLI documentation for atlas clusters onlineArchives delete.

从用户界面删除在线存档

要删除在线存档，请在 Atlas 用户界面中：

通过 API 删除在线存档

To delete an online archive through the API, send a DELETE request to the onlineArchives endpoint with the unique ID of the online archive to delete. To learn more about the syntax and options, see API.

删除标记为归档的时间序列存储桶

When Online Archive archives a time series collection, it first sets the internal control.closed field to true on each bucket that it marks for archival. After this field is set, the time series collection doesn't allow writes to the bucket. Inserts that would target the bucket write to another bucket or open a new one. Updates and deletes through the time series collection ignore the bucket.

如果您在在线存档从集群中删除这些存储桶之前删除在线存档，则这些存储桶将保留在集群上。对时间序列集合的删除操作会返回这些存储桶所包含文档的 deletedCount，即 0。

To remove these buckets, run a delete operation directly against the buckets. To run this operation, your database user must have the readWriteAnyDatabase role. The method depends on your MongoDB version:

• 对于早于 8.2 的 MongoDB 版本，请在 system.buckets.<collection> 集合上运行删除操作。

• For MongoDB 8.2 and later, run the delete command on the time series collection with the rawData: true option.

const cutoff = ISODate("2026-03-01T00:00:00Z");
db.getSiblingDB("<database>")
  .getCollection("system.buckets.<collection>")
  .find({
    "control.closed": true,
    "control.max.<timeField>": { $lte: cutoff }
  })
  .itcount()

const cutoff = ISODate("2026-03-01T00:00:00Z");
db.getSiblingDB("<database>")
  .getCollection("system.buckets.<collection>")
  .deleteMany({
    "control.closed": true,
    "control.max.<timeField>": { $lte: cutoff }
  })

const cutoff = ISODate("2026-03-01T00:00:00Z");
db.getSiblingDB("<database>").runCommand({
  delete: "<collection>",
  deletes: [
    {
      q: {
        "control.closed": true,
        "control.max.<timeField>": { $lte: cutoff }
      },
      limit: 0
    }
  ],
  rawData: true
})