Dash0 acquires Polar Signals

Query Telemetry

Download OTLP spans as a file

post/api/spans/download

Returns the spans matching the given query as a downloadable file in the requested format (txt, ndjson, yaml or csv).

A single export is capped at 5000 spans. A limit above that cap is rejected rather than silently clamped: the response is a client error naming the cap. When the query matches more than 5000 spans, the export carries the first 5000 in result-set order. Use the Dash0 CLI (dash0 spans query) to export larger volumes.

The spans are the same ones POST /api/spans returns for the same query, except that adaptive sampling is always disabled for exports: an exported file is never a sample, whatever the query and whatever the data volume. This request accordingly takes no sampling, and one sent anyway is ignored.

The response is streamed as pages are retrieved and therefore carries no Content-Length. Because the response headers are committed before the first span is written, a failure partway through cannot be reported as a status code — it terminates the transfer instead. Clients MUST confirm the body was received completely before treating the download as a successful export.

bash
Sample request
1234567891011121314151617
curl --request POST 'https://api.eu-west-1.aws.dash0.com/api/spans/download' \
  --header "Authorization: Bearer ${DASH0_AUTH_TOKEN}" \
  --header 'Content-Type: application/json' \
  --data-raw '{
  "timeRange": {
    "from": "now-10m",
    "to": "now"
  },
  "format": "txt",
  "filter": [
    {
      "key": "service.name",
      "operator": "is",
      "value": "shop"
    }
  ]
}'

Request body required

References SpansDownloadRequest

Request for the SpansDownload API. Accepts the same query as GetSpans (filter, timeRange, dataset, ordering) and returns the matching spans serialized as a downloadable file rather than as a JSON page.

The response is streamed, so the export is not paginated: a single request returns up to limit spans, capped at 5000. Adaptive sampling is always disabled for exports, so an exported file is never a sample. Unlike GetSpans this request therefore takes no sampling, and one sent anyway is ignored: the sampling path can substitute a duration-percentile filter and quietly change which spans the file contains, so exports never take it.

FilterCriteria
TimeReferenceRangerequired

A range of time between two time references.

Dataset

Optional dataset to query across. Defaults to whatever is configured to be the default dataset for the organization.

OrderingCriteria

The ordering criteria for the exported spans. Spans are written to the file in this order. Supported ordering key: otel.span.start_time (span start timestamp, defaults to descending), optionally followed by otel.trace.id then otel.span.id ascending, which is the tiebreak span retrieval appends to keep its cursor stable.

SpanExportFormatrequired

The serialization format of the exported file.

  • txt: One rendered span line per record, carrying a fixed field set of start timestamp, service, span name, duration, kind, status code, trace ID and span ID. Ignores columns.
  • ndjson: One OTLP JSON payload per line, each carrying a single span inside the same resourceSpans envelope GetSpans returns, so each line is independently parseable by any OTLP JSON consumer.
  • yaml: The same per-span OTLP payloads as ndjson, as a multi-document YAML stream with documents separated by ---.
  • csv: A header row plus one row per span, flattened to the requested columns.
"txt""ndjson""yaml""csv"
integer

The maximum number of spans to export. Defaults to, and is capped at, 5000. A larger value is rejected rather than silently clamped. Use the Dash0 CLI (dash0 spans query) to export volumes beyond this cap.

string[]

Attribute keys to emit as columns, in order, for the csv format — resource, scope, or span attributes (e.g. service.name, otel.span.duration). Ignored by the other formats, which have a fixed field set.

Responses

The exported spans. The body's actual content type matches the requested format (text/plain for txt, application/x-ndjson for ndjson, application/yaml for yaml, text/csv for csv).

stringformat "binary"

Response headers

Content-Type
stringexample "text/csv"
Content-Disposition
stringexample "attachment; filename=\"spans-2026-07-27T10-15-00Z.csv\""