Vector v0.44.0 release notes

The Vector team is pleased to announce version 0.44.0!

Be sure to check out the upgrade guide for breaking changes in this release.

This release contains numerous enhancements and fixes.

Upgrading Vector

When upgrading, we recommend stepping through minor versions as these can each contain breaking changes while Vector is pre-1.0. These breaking changes are noted in their respective upgrade guides.

Highlights

0.44 Upgrade Guide

January 13, 2025

Pavlos Rontidis (pront)

type: breaking change

Known issues

The aws_s3 source in this version produces many internal logs. These logs will be downgraded to debug in the next release. Until then, you can suppress them with VECTOR_LOG=info,vector::sources::aws_s3=warn.

Vector Changelog

1 new features

VRL was updated to v0.21.0. This includes the following changes:
Breaking Changes & Upgrade Guide
- to_unix_timestamp, to_float, and uuid_v7 can now return an error if the supplied timestamp is unrepresentable as a nanosecond timestamp. Previously the function calls would panic. (https://github.com/vectordotdev/vrl/pull/979)
New Features
- Added new crc function to calculate CRC (Cyclic Redundancy Check) checksum
- Add parse_cbor function (https://github.com/vectordotdev/vrl/pull/1152)
- Added new zip function to iterate over an array of arrays and produce a new arrays containing an item from each one. (https://github.com/vectordotdev/vrl/pull/1158)
- Add new decode_charset, encode_charset functions to decode and encode strings between different charsets. (https://github.com/vectordotdev/vrl/pull/1162)
- Added new object_from_array function to create an object from an array of value pairs such as what zip can produce. (https://github.com/vectordotdev/vrl/pull/1164)
- Added support for multi-unit duration strings (e.g., 1h2s, 2m3s) in the parse_duration function. (https://github.com/vectordotdev/vrl/pull/1197)
- Added new parse_bytes function to parse given bytes string such as 1MiB or 1TB either in binary or decimal base. (https://github.com/vectordotdev/vrl/pull/1198)
- Add main log format for parse_nginx_log. (https://github.com/vectordotdev/vrl/pull/1202)
- Added support for optional timezone argument in the parse_timestamp function. (https://github.com/vectordotdev/vrl/pull/1207)
Fixes
- Fix a panic in float subtraction that produces NaN values. (https://github.com/vectordotdev/vrl/pull/1186)

11 enhancements

aws_s3 source now logs when S3 objects are fetched. If ACKs are enabled, it also logs on delivery.
Thanks to fdamstra for contributing this change!
The file sink now supports any input event type that the configured encoding supports. It previously only supported log events.
Thanks to nionata for contributing this change!
The NATS sink can now parse comma separated urls.
Thanks to whatcouldbepizza for contributing this change!
The log_to_metric transformer tag key are now template-able which enables tags expansion.
Thanks to titaneric for contributing this change!
Add VRL function parse_dnstap that can parse dnstap data and produce output in the same format as dnstap source.
Thanks to esensar, Quad9DNS for contributing this change!
Adds a force_path_style option to the aws_s3 sink that allows users to configure virtual host style addressing. The value defaults to true to maintain existing behavior.
Thanks to sam6258 for contributing this change!
The socket sink now supports unix_datagram as a valid mode. This feature is only available on Linux.
Thanks to jpovixwm for contributing this change!
Add active and inactive metrics for anon and file memory to the cgroup collector. These additional metrics allow you to better understand the existing cgroup memory metrics.
Thanks to nionata for contributing this change!
Add an option to Elasticsearch sink to set a fallback index if the provided template in the bulk.index field cannot be resolved
Thanks to ArunPiduguDD for contributing this change!
The Alpine base image used for Vector -alpine and -distroless-static images was updated to 3.21.
A new GLACIER_IR option was added to storage_class for aws_s3 sink.
Thanks to MikeHsu0618 for contributing this change!

9 bug fixes

The kubernetes_logs source now sets a user-agent header when querying k8s apiserver.
Thanks to ganelo for contributing this change!
Fix the HTTP config provider to correctly parse TOML provided by the given HTTP endpoint.
Thanks to PriceHiller for contributing this change!
Allow the skip_unknown_fields setting to be optional, thereby allowing use of the defaults provided by the ClickHouse server. Setting it to true will permit skipping unknown fields and false will make ClickHouse strict on what fields it accepts.
Thanks to PriceHiller for contributing this change!
Retry Kafka messages that error with a policy violation so messages are not lost.
Thanks to PriceHiller for contributing this change!
Changes the fingerprint for file sources to use uncompressed file content as a source of truth when fingerprinting lines and checking ignored_header_bytes. Previously this was using the compressed bytes. For now, only gzip compression is supported.
Thanks to roykim98 for contributing this change!
The filter transform now generates a more accurate config when generated via vector generate by using a comparison rather than an assignment.
Thanks to abcdam for contributing this change!
The gcp_pubsub source no longer has a 4MB message size limit.
Thanks to sbalmos for contributing this change!
Fix opentelemetry sink input resolution. The sink is now using the underlying protocol to determine what inputs are accepted.
Thanks to pront for contributing this change!
The sample transform now correctly uses the configured sample_rate_key instead of always using "sample_rate".
Thanks to dekelpilli for contributing this change!

Vector v0.44.0 release notes

Highlights

Known issues

Vector Changelog

1 new features

Breaking Changes & Upgrade Guide

New Features

Fixes

11 enhancements

9 bug fixes

Download Version 0.44.0