--- title: Fluentd description: Adapter that delivers logs to a Fluentd daemon. location: https://istio.io/docs/reference/config/policy-and-telemetry/adapters/fluentd.html layout: protoc-gen-docs generator: protoc-gen-docs supported_templates: logentry aliases: - /docs/reference/config/adapters/fluentd.html number_of_entries: 1 ---

The fluentd adapter is designed to deliver Istio log entries to a listening fluentd daemon.

This adapter supports the logentry template.

Params

Configuration parameters for the fluentd adapter.

This adapter accepts logentry instance. It then routes those logentries to a listening fluentd daemon with minimal transformation. Fluentd uses a “tag” for all logs. The “Name” of the logentry is used as the “tag”, unless the logentry already has a variable “tag”.

Field Type Description Required
address string

Address of listening fluentd daemon. Example: fluentd-server:24224 Default value is localhost:24224

No
integerDuration bool

Convert attributes of type duration to integers of unit ms in logs. Default behaviour is a string representation including unit.

No
instanceBufferSize int64

Controls the size of the queue of log entries awaiting processing by the data push goroutine. This impacts the overall size of the backlog of log entries that the handler will accept. If the handler falls behind and the number of outstanding instances exceeds this limit, the handler will begin to drop log entries. Defaults to 1024.

No
maxBatchSizeBytes int64

Controls the number of data bytes that are encoded and buffered by the handler before they are pushed to the fluentd backend. This impacts the size and frequency of requests sent to the fluentd backend. Once the buffer exceeds this limit, the handler will attempt to push data to the backend. Defaults to 8,388,608 (8 MiB).

No
pushIntervalDuration Duration

Duration interval for pushing batched data to the fluentd backend. At least once every duration, the handler will attempt to push data. Default to 1m.

No
pushTimeoutDuration Duration

Timeout duration for pushing batched data to the fluentd backend. If a request takes longer than the configured timeout, the request will be cancelled and dropped. Default to 1m.

No