Vector v0.14.0 release notes

The Vector team is pleased to announce version 0.14.0!

This release includes a new exec source that can be used to execute arbitrary programs and use their output to generate events in Vector along with a number of enhancements.

Check out the changelog for more details.

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.


24 enhancements

graphql api Expose events_in & events_out metrics 6888 kubernetes_logs source Allow configuration of max_line_bytes 7226 cli Update top to use more accurate events_in_total and events_out_total metrics 7286 kafka source kafka sink Emit internal metrics provided by rdkafka 5874 remap Wdd whitespace option to parse_key_value function 7292 datadog sinks Generic site support 7281 http sink Add support for end-to-end acknowledgements 7265 remap Add parse_klog function 7349 datadog_logs sink Add acknowledgements support 7354 datadog source http source Add acknowledgement support 7325 helm platform Allowing setting command for vector-agent chart 7107 heroku_logs source http source Add end-to-end acknowledgements 7431 sinks Add acknowledgements support to more HTTP logs sinks 7411 helm platform Allow for global imagePullSecrets 7500 datadog_logs sink Use per-API key partitions for datadog logs sink 7394 sources Allow for opting out of acknowledgements 7531 buffers Improve disk buffer compaction performance 7264 remap Add match_any function 7414 vector sink Add acknowledgement support 7625 vector source Add acknowledgement support to v2 protocol 7618 aws_s3 sink Add acknowledgements support 7536 datadog_metrics sink Add acknowledgements support 7535 helm platform additional labels for Prometheus Operator 7474 aws_kinesis_firehose source Add record compression support to aws_kinesis_firehose source 7700

6 new features

new source New exec source 6876 remap transform Allow emitting multiple events by assigning . to an array 7267 config Initial [provider] implementation - http 7166 remap Add encode_logfmt function 6985 vector sink vector source Add gRPC-based implementation 7197 remap Add redact function to VRL 7250

25 bug fixes

remap Escape quotes and backslashes when printing VRL strings 6739 graphql api --no-default-features --features api build failure 7284 splunk_hec source Default host to remote address 7293 cli Enable comma separated lists to be used in configuration environment variables that take a list 7316 observability Correct all *_nanoseconds metrics to _seconds 7373 buffers Work around a LevelDB mechanism mismatch 7398 debian platform Add debhelper include tags to Debian scripts 7363 external docs Document real behaviour of the timestamp field in Syslog source 7443 clickhouse sink Use correct batch status 7452 kubernetes_logs source Expose data_dir configuration 7253 kubernetes platform Fix binfmt exclusion in Helm chart 7332 aws_s3 source Properly handle urlencoded S3 object keys 7470 external docs Fix configure command in installation instructions 7498 codecs Filter object and array fields with only_fields 7468 performance Avoid high memory usage from disk buffers at start-up due to scanning the entire buffer 7522 kubernetes_logs source Kubernetes stream errors should be recoverable 7484 kubernetes_logs source Replace incorrect invocation error with WatchStreamFailed 7534 external docs Fix relevant_when for socket.path config 7587 dev Optimistically add ubuntu to docker group 7583 log_to_metric transform Drop events where field is null 7487 cli include error details in vector vrl output 7581 buffers Avoid compacting small disk buffers 7614 aws_cloudwatch_logs sink Fix healthcheck to allow for log group creation at runtime 7601 buffers Improve disk buffer performance by batching deletes 7615 remap Accept quoted lists in parse_aws_alb_log function 7682

What’s next

End to end acknowledgements

We’ve heard from a number of users that they’d like improved delivery guarantees for events flowing through Vector. We are working on a feature to allow, for components that are able to support it, to only acknowledging data flowing into source components after that data has been sent by any associated sinks. For example, this would avoid acknowledging messages in Kafka until the data in those messages has been sent via all associated sinks.

This release includes the groundwork for this feature including support in a number of source and sink components that support acknowledgements, but it has not yet been fully documented. We expect to officially release this with 0.15.0.

Kubernetes aggregator role

We are hard at work at expanding the ability to run Vector as an aggregator in Kubernetes. This will allow you to build end-to-end observability pipelines in Kubernetes with Vector. Distributing processing on the edge, centralizing it with an aggregator, or both. If you are interested in beta testing, please join our chat and let us know.

We do expect this to be released in the next few months.

Download Version 0.14.0

Linux (deb)
Linux (rpm)
Windows (MSI)