Skip to main content

Fluent Bit to LynxDB

If you already run Fluent Bit​

Keep the es output and change only the host and port.

 [OUTPUT]
Name es
Match *
- Host elasticsearch
- Port 9200
+ Host lynxdb
+ Port 3100
Suppress_Type_Name On

Full annotated config​

# Fluent Bit configuration for LynxDB.
[INPUT]
Name tail
Path /var/log/*.log
Tag app
Read_From_Head On

[OUTPUT]
Name es
Match *
Host lynxdb
Port 3100
Logstash_Format Off
Index fluent-bit
Suppress_Type_Name On
Generate_ID On
Compress gzip
# LynxDB echoes generated IDs in bulk responses; storage stays append-only.

Where the events go​

The es output sends NDJSON bulk requests. LynxDB stores the events in the index named by Index (here fluent-bit) and accepts gzip-compressed bodies. With Logstash_Format On, the index name carries the date, such as logstash-2026.02.17; lynxdb indexes shows the names.

lynxdb query 'source=fluent-bit | stats count() by host'

Each record is stored as sent. The members of the kubernetes object that Fluent Bit's Kubernetes filter adds are fields named by their path:

lynxdb query 'source=fluent-bit | stats count() by kubernetes.namespace_name, kubernetes.pod_name'

Troubleshooting​

Use Suppress_Type_Name On with current Fluent Bit versions. If you see retry loops, check the Fluent Bit logs for HTTP 4xx or 5xx responses; a 400 with invalid_index_name_exception means the index name has characters LynxDB does not accept (bulk index names use lowercase letters, digits, ., _ and -, and do not start with _, -, + or .).

Supported behavior​

Supported: Name es, bulk requests, gzip compression, generated IDs echoed in responses, and ingestion without template setup.

Not supported: Elasticsearch template installation, update and delete actions, and Elasticsearch query APIs.