Export
Save the current page or every matching document as JSON, NDJSON or CSV — optionally with the index definition.
You can export documents from:
- an index, data stream or alias page in the Explorer (Export documents… / Export…);
- the results of the Index view’s Query tab and its Documents tab;
- a search response in the Query workspace;
- the results of an aggregation pipeline (Export results — one row per bucket, as CSV or NDJSON).
Options
What to export
- This page — the documents you’re looking at.
- All matching documents — every document matching the current query; the dialog shows how many there are. They are streamed from the cluster straight to disk with the scroll API, 1,000 at a time, so even millions of documents don’t fill your memory.
Format
| Format | Best for |
|---|---|
| NDJSON (default) | one document per line — ready for _bulk |
| JSON | a single array |
| CSV | spreadsheets; nested fields are flattened to columns like city.name |
Options
- Include _id and _index (on by default).
- Limit — stop after N documents.
- Also save mapping, settings & aliases — writes
<file>.definition.jsonnext to the export with everything needed to recreate the index: the mapping, the portable settings (shards, replicas, refresh interval, analysis, sorting, codec…) and the aliases. Available for whole-index exports from the index page and the Documents tab.
Only query, sort, _source and runtime_mappings are taken from your request when exporting all matching documents.
While it runs
A progress bar shows N / total documents written. Cancel export stops it and keeps what was written so far. Files are named like listings-v7-2026-10-02.ndjson (-page is added for page exports).
CSV columns come from the first 1,000 documents. If later documents have fields the first ones didn’t, use NDJSON or JSON to keep everything.