Yandex StoreDoc settings

When creating or editing a Yandex StoreDoc cluster, you can specify settings that pertain to Yandex StoreDoc.

The label next to the setting name helps determine which interface is used to set the value of this setting: the management console, CLI, API, or API.

Depending on the selected interface, the same setting will be represented differently. For example, Max incoming connections in the management console is the same as:

• max_incoming_connections in gRPC or the CLI
• maxIncomingConnections in the REST API

In a sharded cluster, some settings are only available to certain host roles.

How settings depend on the host classHow settings depend on the host class

The values of some Yandex StoreDoc settings can be automatically adjusted when you change the host class:

• If the values were not specified or are not suitable for the new class, the default settings for this class will be applied.
• If the settings you specified manually are suitable for the new class, they will not change.

The settings that depend on the host class are:

• Net → Max incoming connections.
• Engine config → Cache size gb.

Cluster-level DBMS settingsCluster-level DBMS settings

• Net All interfaces

  Networking settings.

  • Max incoming connections

    Maximum number of inbound connections.

    The minimum value is 10. The maximum value depends on the selected host class and is equal to the host's RAM volume in MB, with a maximum of 16384. The default value is 1024.

    For more information, see Connection limits.

  • Compression → Compressors

    Lists the compression methods that a host with the MONGOD or MONGOS role can use to compress network messages. The order of the listing is important.

    The disabled value disables compression. Default value: snappy,zstd,zlib.

• Operation profiling All interfaces

  Note

  These settings are not available for hosts with the MONGOS role in a sharded cluster.

  DBMS profiler settings. The profiler collects query data and then uses it to determine a query optimization strategy.

  • Mode

    DBMS profiler mode:

    • off: Profiling disabled.
    • slowOp (default): Only collects information about slow operations (that take longer than the threshold value specified in the Slow op threshold setting).
    • all: Collects information about all queries in progress.

  • Slow op sample rate

    The number of slow operations to profile or log. If enabled, this parameter affects diagnostic logging and the profiler.

    Minimum value: 0; maximum value: 1; default: 1.

  • Slow op threshold

    Sets the operation execution time (in milliseconds). If exceeded, the operation is considered slow.

    Minimum value: 0; maximum value: 36000000 (10 hours); default: 300.

• Set parameter

  • Enable flow control All interfaces

    Decides whether or not to control the rate of primary host's write operations. Enabling this parameter ensures that the "majority commited" metric for replica hosts will not exceed the pre-configured value of 10 seconds.

    Default value: false (flow control disabled).

  • Min snapshot history window in seconds All interfaces

    Note

    This setting is available for hosts with the MONGOD role only.

    Time, in seconds, for snapshot history to be retained.

    Possible values: from zero up. Default value: 60. Increasing this setting value increases disk usage.

• Storage All interfaces

  Note

  This setting is unavailable for hosts with the MONGOS role in a sharded cluster.

  Hosts with the MONGOCFG role only have the Engine config → Cache size gb setting available.

  • Journal → Commit interval

    The interval in milliseconds between the saves of Yandex StoreDoc journal data to the disk.

    The minimum value is 1; the maximum value is 500; default: is 300.

  • Wired tiger

    Settings for the storage engine:

    • Collection config → Block compressor

      Data compression option. The value set here only applies to new collections. You can also redefine it for individual collections and indexes.

      • none: Compression disabled.
      • snappy: Use the snappy library for data compression. If you select this setting, data compression/decompression will be, in most cases, faster than if using the zlib library; however, the resulting file size will be 20% to 100% larger.
      • zlib: Use the zlib library for data compression. This method is slower than snappy but, in most cases, it achieves better data compression.

    • Engine config → Cache size gb

      Maximum size of the internal cache used for data storage (in GB). This setting does not affect the amount of RAM used to create an index.

      Minimum value: 0.25 (256 MB). The maximum and default values depend on the selected host class and are calculated as follows:

      • Maximum value: 0.9 × <host_RAM_amount>.
      • Default value: 0.5 × <host_RAM_amount>.

      For example, for s1.medium (8 vCPU, 32 GB) class hosts, the maximum value is 28.8 and the default value is 16.

    • Index config → Prefix compression

      Enables or disables prefix compression for indexes. Changing the parameter only affects new indexes, not the existing ones.

      Default value: true (prefix compression enabled).