Skip to main content

Async search

Async searches provide the ability to run search requests in the background.

This ability is especially valuable when executing long-running queries that take longer to complete. Usually this queries are aggregations and histograms. Async searches can also be used to search documents using with_docs field in the StartAsyncSearch request, though primary use case is aggregations.

Read API docs for more info about the public API.

Async search data is persisted on disk for the specified retention time. Minimum retention is 5 minutes and maximum is 30 days. Retention is set by the retention field in the StartAsyncSearch request. Data is deleted after the retention period expires.

When data size exceeds the limits, read-only mode is enabled: new async searches are rejected, and unfinished searches are suspended until disk space is freed by deleting older searches by retention.

Configuration​

Configuration parameters are:

  • data_dir [string] - directory that contains async search data. By default, it is a subdirectory of config.storage.data_dir.
  • concurrency [int] - number of concurrent async search executions.
  • max_total_size [bytes] - maximum total size of all async searches data per store.
  • max_size_per_request [bytes] - maximum total size of a single async search data per store.

Configuration parameters are part of async_search object in the config file.