v5.1.3

KNOWLEDGE BASE

Release Notes v5.1.3

Fluent Bit is a Fast and Lightweight Telemetry Agent for Linux, BSD, macOS and Windows. We are proud to announce the availability of Fluent Bit v5.1.3.

For people upgrading from previous versions, please read the Upgrading Notes section of our documentation:

https://docs.fluentbit.io/manual/5.1/installation/upgrade_notes

Fluent Bit v5.1.3 is a maintenance release that strengthens ingestion, delivery, and shutdown reliability. It adds Kafka Protobuf serialization, a new ArvanCloud CloudLogs output, TLS certificate revocation checks, and new controls for cloud storage uploads, Syslog framing, and metric freshness.

The release contains 296 commits since v5.1.2, with improvements across the engine, networking, encoding, metrics and traces, cloud outputs, bundled libraries, packaging, and test coverage.

This release also reflects security work across the project in collaboration with independent security researchers and the Anthropic team. We thank them for their careful review and contributions to strengthening Fluent Bit.

What’s new ?

Core

• engine: bound concurrent output flushes and retries to keep pipelines responsive under heavy backpressure
• lifecycle: improve input initialization cleanup, connection handling, and shutdown of pending output requests
• storage: restore persisted trace chunks for replay, preserve normalized tags, and improve chunk handling under backpressure
• TLS: add tls.crl_file for certificate revocation checks on inputs and outputs, preserve it across certificate reloads, and honor client-certificate verification settings consistently
• HTTP server: improve HTTP/2 request handling and session cleanup, reduce idle-timeout log noise, and tighten trace API controls
• encoding: improve handling of nested JSON, MessagePack, Lua tables, record accessors, GELF strings, and Unicode conversions
• OpenTelemetry: improve trace status interoperability, preserve signed 64-bit log body integers, and reduce metadata lookup overhead during metrics JSON decoding
• configuration: resolve YAML include paths correctly on Windows
• stream processor: improve snapshot handling for large records and reliability of windowed queries
• packaging: improve Windows package integrity checks, refresh container build dependencies, and enable Kafka plugins in CentOS Stream 9 and 10 packages
• tests: expand coverage for backpressure, shutdown, trace replay, TLS revocation, ingestion validation, serialization, multiline processing, and concurrent cloud uploads

Libraries

• cmetrics: upgrade to v2.2.5, with improvements to metrics decoding, encoding, and validation
• ctraces: upgrade to v0.7.3, improving trace interoperability and decoding reliability
• cprofiles: upgrade to v0.2.2, improving profile encoding, decoding, and validation
• Monkey: upgrade to v1.8.10
• flb_libco: refresh the bundled coroutine library, including portability and runtime validation improvements

Plugins

Inputs

• OpenTelemetry
  • improve validation and cleanup of OTLP log and trace requests
• Elasticsearch
  • improve malformed bulk request responses and preserve retryable ingestion failures
• Forward
  • improve packed-record validation, tag fallback, and pause handling with listener workers
• TCP, Syslog, and MQTT
  • improve connection lifecycle handling and protocol validation
  • correct Syslog stream timestamps and framing, and TCP handling of non-record JSON values
• Node Exporter Metrics
  • expire unobserved systemd metric series so exported metrics reflect the current snapshot
• Podman Metrics
  • continue startup when the initial scrape fails and retry on the collection interval
  • improve handling of container metadata
• Kubernetes Events
  • add database indexes to improve event lookup performance with large stored histories
• Windows Exporter Metrics
  • correct TCP connection-state collection
• Tail
  • improve resource cleanup on Windows

Processors & Filters

• Multiline
  • coordinate ingestion with pause and shutdown, and improve partial-message and stream-group handling
• Kubernetes
  • improve pod matching and metadata value validation
• Lua, Modify, and Nest
  • improve initialization cleanup and record transformation reliability
• Content Modifier, SQL, and Sampling processors
  • improve conversion, extraction, key lookup, and span identifier validation
  • preserve static metric labels across processor chains

Outputs

• ArvanCloud CloudLogs
  • add a new arvancloud_cloudlogs output with API-key authentication and configurable log type and timestamp extraction
• Kafka
  • add format=protobuf for builds with the optional Protobuf encoder enabled
  • support Confluent-compatible Protobuf serialization, referenced schemas, message selection, and schema caching through Schema Registry
  • extend the Avro Schema Registry support already available in the 5.1 series
• Google Cloud Storage
  • add total_file_size to trigger uploads by buffered size, with a default of 100M; set it to 0 to disable size-triggered uploads
  • improve multi-worker uploads, ordered queue draining, backlog recovery, and net.* configuration support
• Amazon S3
  • improve upload and credential handling across multiple workers while retaining concurrency between tags
• Syslog
  • add RFC 6587 octet-counting framing for TCP and TLS with syslog_framing=octet_counting
  • add syslog_sd_preset as a fallback for RFC 5424 structured data
• OpenTelemetry
  • add optional cut_off_time filtering to exclude old metric samples
  • preserve failed log batch results for retry handling and improve trace export and shutdown reliability
• Elasticsearch and OpenSearch
  • improve partial bulk delivery accounting, health reporting, and response diagnostics
  • add opt-in drop_unrecoverable_records to discard records rejected by non-retryable bulk errors; disabled by default
• CloudWatch Logs
  • isolate stream caches per output worker
• Google Cloud Logging
  • isolate authentication connections per worker and improve token refresh cleanup
• Chronicle
  • correct ingestion record counts and log_key handling
• HTTP, Splunk, and other outputs
  • improve record-derived header validation, attribute handling, formatting, and resource cleanup

Join us

We want to hear from you. Our community keeps growing, and you can be part of it:

• Github: http://github.com/fluent/fluent-bit
• Slack: CNCF Slack, channel #fluent-bit (https://communityinviter.com/apps/cloud-native/cncf)
• Twitter: @fluentbit