Elasticsearch Compatibility Stubs
LynxDB implements enough Elasticsearch-compatible surface area for log shippers to start, health check, and send bulk data. These endpoints are not a general Elasticsearch API.
Ingest routes
The ingest routes take documents while ingest.es_compat.enabled is on (the default); otherwise they answer 503 with "type": "unavailable".
| Route | Behavior |
|---|---|
POST /_bulk | Bulk ingest into the index each action names |
POST /{index}/_bulk | Same, with {index} for actions that name no _index |
POST /_data_stream/{name}/_bulk | Same, with the data stream {name} for actions that name no _index |
POST /api/v1/es/_bulk | Prefixed bulk alias |
POST /api/v1/es/_data_stream/{name}/_bulk | Prefixed data stream bulk alias |
POST /api/v1/es/{index}/_doc | Single-document ingest into {index} |
A bulk body is NDJSON with application/x-ndjson or application/json as its content type, optionally gzip-compressed. Each document line is appended, as it is, to the index named by the action's _index, else by the {index} or data stream of the URL, else to main:
printf '{"index":{"_index":"web"}}\n{"@timestamp":"2026-02-10T08:00:00Z","message":"hello","host":{"name":"h1"}}\n' |
curl -s -X POST localhost:3100/_bulk -H 'Content-Type: application/x-ndjson' --data-binary @-
curl -s localhost:3100/api/v1/query -d '{"q":"source=web"}'
{"data":{"type":"events","events":[{"@timestamp":"2026-02-10T08:00:00Z","_time":"2026-02-10T08:00:00Z","host.name":"h1","message":"hello"}],"total":1,"has_more":false},"meta":{"took_ms":0.096,"query_id":"qry_a0579f645c0c50f0","rows":1}}
The document keeps its fields; nested objects are read as paths such as host.name. _time comes from @timestamp or another field of storage.time_fields, else from the time of arrival.
indexandcreateactions are accepted.updateanddeleteactions are refused per item with400 action_request_validation_exception.- An index name holds lowercase letters, digits,
.,_and-, does not start with-,_,+or., and is at most 255 bytes long; any other name is refused per item with400 invalid_index_name_exception. - A data line that is not a JSON object is refused per item with
400 mapper_parsing_exception. - The server makes the accepted documents durable before it answers, because shippers count a
200as stored. _idvalues are echoed in the response, or generated when the action has none. LynxDB is append-only: they are not used for deduplication or updates.
Probe and setup stubs
| Route | Response shape |
|---|---|
GET / | Elasticsearch-like cluster handshake |
HEAD / | Empty 200 OK handshake |
GET /_xpack | Basic license feature summary |
GET /_xpack/license | Active basic license |
GET /_license | Active basic license |
GET /_cat/templates | Empty JSON array |
GET /_cat/indices | Empty JSON array |
GET /_cluster/health | Green single-node health response |
GET /_index_template/{name} | Empty index template list |
PUT /_index_template/{name} | {"acknowledged": true} |
GET /_ilm/policy/{name} | Empty 404 policy response |
GET /_ingest/pipeline/{name} | Empty 404 pipeline response |
GET /_nodes/{path} | Minimal node HTTP info |
GET /_alias | Empty aliases object |
GET /_data_stream/{name} | Empty data stream list |
PUT /_data_stream/{name} | {"acknowledged": true} |
GET /_search | Empty hits response |
POST /_security/user/_authenticate | Authenticated user stub |
HEAD /{index} | Empty 200 OK index probe for shipper health checks |
Every unprefixed probe above has a matching /api/v1/es/... alias. Drop-in shipper configs should use the unprefixed routes.
Under the /api/v1/es/ prefix, PUT /api/v1/es/_ilm/policy/{name}, PUT /api/v1/es/_ingest/pipeline/{name} and PUT /api/v1/es/{index} answer 501 with {"error":"unsupported Elasticsearch management endpoint","status":501} and log a warning with the method and path.
Explicit non-goals
These stubs do not store Elasticsearch templates, ILM policies, ingest pipelines, aliases, users, roles, or data streams. They only keep setup probes from blocking log ingest.