Package org.elasticsearch.xpack.esql.datasources.spi

Interface FormatReaderStatus

All Superinterfaces:
NamedWriteable, ToXContent, ToXContentFragment, Writeable
public interface FormatReaderStatus extends NamedWriteable, ToXContentFragment
Typed per-format reader counter snapshot folded into the format_reader field of the external-source operator status. Each format module (parquet, orc, ndjson, csv) contributes one concrete implementation and registers it as a NamedWriteable from its data-source plugin, so the coordinating node can deserialize whichever format produced the profile.

Modeled as a NamedWriteable rather than a generic Map so the fields stay machine-readable across versions — the operator already scrapes readNanos() to roll format-reader time up into the operator status, and a Map on the wire is a one-way door (converting it to a typed shape after release is a wire-compatibility break). The esql_external_source_profile transport version that gates this payload is unreleased, so the typed shape ships from the first release that carries it.

The three accessors are the fields every format shares; format-specific counters live on each implementation and surface through its ToXContent.toXContent(org.elasticsearch.xcontent.XContentBuilder, org.elasticsearch.xcontent.ToXContent.Params) body.

  • Method Details

    • format

      String format()
      Format identifier as it appears in the profile, e.g. "parquet", "tsv".

    • rowsEmitted

      long rowsEmitted()
      Rows the reader emitted into the operator.

    • readNanos

      long readNanos()
      Cumulative producer-thread time spent inside the format reader (open, decode, decompress).