turbo

Enable TurboVM bytecode acceleration for normalization. Requires liblognorm 2.1.0+ built with --enable-turbo.

This parameter applies to mmnormalize: Log Message Normalization Module.

Name:

turbo

Scope:

action

Type:

boolean

Default:

off

Required?:

no

Introduced:

8.2602.0

Description

When enabled, mmnormalize compiles the rulebase into bytecode at startup and uses the TurboVM engine for SIMD-accelerated normalization. Each rsyslog worker thread gets an independent TurboVM context, ensuring thread-safe operation without mutexes on the hot path.

Normalized results are stored as opaque snapshots that support zero-JSON field resolution: template access reads fields directly from the snapshot without building a json-c object tree.

If bytecode compilation fails (e.g., unsupported parser types or inline data exceeding 60 bytes), mmnormalize falls back to standard ln_normalize() automatically. On HUP signal, per-worker TurboVM contexts are rebuilt.

Requirements:

  • liblognorm >= 2.1.0 with --enable-turbo

  • The lognorm-features.h header must define LOGNORM_TURBO_SUPPORTED

  • rsyslog detects this automatically at ./configure time

Action usage

action(type="mmnormalize"
       rulebase="/etc/rsyslog.d/rules.rb"
       turbo="on")

YAML usage

actions:
  - type: mmnormalize
    rulebase: /etc/rsyslog.d/rules.rb
    turbo: on

See also

See also mmnormalize: Log Message Normalization Module.

Support: rsyslog Assistant | GitHub Discussions | GitHub Issues: rsyslog source project

Contributing: Source & docs: rsyslog source project

© 2008–2026 Rainer Gerhards and others. Licensed under the Apache License 2.0.