=== Beats version 6.0.0
==== Breaking changes
Affecting all Beats
- The log directory (`path.log`) for Windows services is now set to
`C:\ProgramData\[beatname]\logs`.
- The _all field is disabled in Elasticsearch 6.0.
- Fail if removed setting output.X.flush_interval is explicitly
configured.
- Rename the `/usr/bin/beatname.sh` script (e.g. `metricbeat.sh`) to
`/usr/bin/beatname`.
- Beat does not start if elasticsearch index pattern was modified but
not the template name and pattern.
- Fail if removed setting output.X.flush_interval is explicitly
configured.
- Rename `kubernetes` processor to `add_kubernetes_metadata`.
- Rename `.full.yml` config files to `*.reference.yml`.
- The `scripts/import_dashboards` is removed from packages. Use the
`setup` command instead.
- Change format of the saved kibana dashboards to have a single JSON
file for each dashboard
- Rename `configtest` command to `test config`.
- Remove setting `queue_size` and `bulk_queue_size`.
- Remove setting `dashboard.snapshot` and `dashboard.snapshot_url`.
- Beats can no longer be launched from Windows Explorer (GUI), command
line is required.
Filebeat
- Rename `input_type` field to `prospector.type`
- The `@metadata.type` field, added by the Logstash output, is now
hardcoded to `doc` and will be removed in future versions.
Metricbeat
- Change all `system.cpu..pct` metrics to be scaled by the number of
CPU cores.
- Remove filters setting from metricbeat modules.
- Added `type` field to filesystem metrics.
Heartbeat
- Renamed the heartbeat RPM/DEB name to `heartbeat-elastic`.
Packetbeat
- Remove not-working `runoptions.uid` and `runoptions.gid` options in
Packetbeat.
- Remove the already unsupported `pf_ring` sniffer option.
Auditbeat
- Changed file metricset config to make `file.paths` a list instead of
a dictionary.
==== Bugfixes
Affecting all Beats
- Fix data race accessing watched containers.
- Do not require template if index change and template disabled
- Fix missing ACK in redis output.
- Fix the `/usr/bin/beatname` script to accept `-d ""` as a parameter.
- Combine `fields.yml` properties when they are defined in different
sources.
- Keep Docker & Kubernetes pod metadata after container dies while
they are needed by processors.
- Fix `fields.yml` lookup when using `export template` with a custom
`path.config` param.
- Remove runner creation from every reload check
- Fix add_kubernetes_metadata matcher registry lookup.
- Register kubernetes `field_format` matcher and remove logger in
`Encode` API
- Fix go plugins not loaded when beat starts
- Add support for `initContainers` in `add_kubernetes_metadata`
processor.
- Eliminate deprecated _default_ mapping in 6.x
- Fix pod name indexer to use both namespace, pod name to frame index
key
- Don't stop with error loading the ES template if the ES output is
not enabled.
- Fix race condition in internal logging rotator.
- Normalize all times to UTC to ensure proper index naming.
- Fix issue with loading dashboards to ES 6.0 when .kibana index did
not already exist.
- Fix importing the dashboards when the limit for max open files is
too low.
- Fix configuration documentation for kubernetes processor
- Fix misspelling in `add_locale` configuration option for
abbreviation.
Filebeat
- Fix machine learning jobs setup for dynamic modules.
- Fix default paths for redis 4.0.1 logs on macOS
- Fix Filebeat not starting if command line and modules configs are
used together.
- Fix double `@timestamp` field when JSON decoding was used.
- Fix issue where the `fileset.module` could have the wrong value.
- Fix race condition on harvester stopping with reloading enabled.
- Fix recursive glob config parsing and resolution across restarts.
- Allow string characters in user agent patch version (NGINX and
Apache)
- Fix grok pattern in filebeat module system/auth without hostname.
Winlogbeat
- Removed validation of top-level config keys.
Metricbeat
- Use `beat.name` instead of `beat.hostname` in the Host Overview
dashboard.
- Fix the loading of 5.x dashboards.
- Fix a memory allocation issue where more memory was allocated than
needed in the windows-perfmon metricset.
- Don't start metricbeat if external modules config is wrong and
reload is disabled
- The MongoDB module now connects on each fetch, to avoid stopping the
whole Metricbeat instance if MongoDB is not up when starting.
- Fix kubernetes events module to be able to index time fields
properly.
- Fixed `cmd_set` and `cmd_get` being mixed in the Memcache module.
- Added missing mongodb configuration file to the `modules.d` folder.
- Fix wrong MySQL CRUD queries timelion visualization
- Add new metrics to CPU metricsset
- Fix issue affecting Windows services timing out at startup.
- Fix incorrect docker.diskio.total metric calculation.
- Vsphere module: used memory field corrected.
- Set correct format for percent fields in memory module.
- Fix a debug statement that said a module wrapper had stopped when it
hadn't.
- Use MemAvailable value from /proc/meminfo on Linux 3.14.
- Fix panic when events were dropped by filters.
- Add filtering to system filesystem metricset to remove relative
mountpoints like those from Linux network namespaces.
- Remove unnecessary print statement in schema apis.
- Fix type of field `haproxy.stat.check.health.last`.
Heartbeat
- Fix monitor.name being empty by default.
- Fix wrong event timestamps.
Packetbeat
- Fix missing length check in the PostgreSQL module.
- Fix panic in ACK handler if event is dropped on blocked queue
- Update flow timestamp on each packet being received.
- Enabled /proc/net/tcp6 scanning and fixed ip v6 parsing.
- Enable memcache filtering only if a port is specified in the config
file.
Auditbeat
- Fix `file.max_file_size` config option for the audit file metricset.
==== Added
Affecting all Beats
- Enable flush timeout by default.
- Add @metadata.version to events send to Logstash.
- Add setting to enable/disable the slow start in logstash output.
- Update init scripts to use the `test config` subcommand instead of
the deprecated `-configtest` flag.
- Get by default the credentials for connecting to Kibana from the
Elasticsearch output configuration.
- Added `cloud.id` and `cloud.auth` settings, for simplifying using
Beats with the Elastic Cloud.
- Add lz4 compression support to kafka output.
- Add newer kafka versions to kafka output.
- Configure the index name when loading the dashboards and the index
pattern.
- New cli subcommands interface.
- Allow source path matching in `add_docker_metadata` processor.
- Add support for analyzers and multifields in fields.yml.
- Add support for JSON logging.
- Add `test output` command, to test Elasticsearch and Logstash output
settings.
- Introduce configurable event queue settings: queue.mem.events,
queue.mem.flush.min_events and queue.mem.flush.timeout.
- Enable pipelining in Logstash output by default.
- Added 'result' field to Elasticsearch QueryResult struct for
compatibility with 6.x Index and Delete API responses.
- The sample dashboards are now included in the Beats packages.
- Add `pattern` option to be used in the fields.yml to specify the
pattern for a number field.
- Upgraded to Golang 1.8.3.
- Added the possibility to set Elasticsearch mapping template settings
from the Beat configuration file.
- Add a variable to the SysV init scripts to make it easier to change
the user.
- Add the option to write the generated Elasticsearch mapping template
into a file.
- Add `instance_name` in GCE add_cloud_metadata processor.
- Add `add_docker_metadata` processor.
- Add `logging.files` `permissions` option.
Filebeat
- Add Kubernetes manifests to deploy Filebeat.
- Changed the number of shards in the default configuration to 3.
- Don't start filebeat if external modules/prospectors config is wrong
and reload is disabled
- Add `filebeat.registry_flush` setting, to delay the registry
updates.
- Add experimental Redis module.
- Nginx module: use the first not-private IP address as the remote_ip.
- Load Ingest Node pipelines when the Elasticsearch connection is
established, instead of only once at startup.
- Add support for loading Xpack Machine Learning configurations from
the modules, and added sample configurations for the Nginx module.
- Add udp prospector type.
- Enabled Cgo which means libc is dynamically compiled.
- Add Beta module config reloading mechanism
- Remove spooler and publisher components and settings.
- Added ability to sort harvested files.
- Add experimental Redis slow log prospector type.
Winlogbeat
- Changed the number of shards in the default configuration to 3.
- Add the ability to use LevelRaw if Level isn't populated in the
event XML.
Metricbeat
- Add Kubernetes manifests to deploy Metricbeat.
- Auto-select a hostname (based on the host on which the Beat is
running) in the Host Overview dashboard.
- Add `filesystem.ignore_types` to system module for ignoring
filesystem types.
- Add support to exclude labels from kubernetes pod metadata.
- Add random startup delay to each metricset to avoid the thundering
herd problem.
- Add the ability to configure audit rules to the kernel module.
- Add the ability to configure kernel's audit failure mode.
- Add experimental Aerospike module.
- Vsphere module: collect custom fields from virtual machines.
- Add `test modules` command, to test modules expected output.
- Add `processors` setting to metricbeat modules.
- Support `npipe` protocol (Windows) in Docker module.
- Add macOS implementation of the system diskio metricset.
- Add process_summary metricset that records high level metrics about
processes.
- Add `kube-state-metrics` based metrics to `kubernetes` module
- Add debug logging to Jolokia JMX metricset.
- Add events metricset for kubernetes metricbeat module
- Change Metricbeat default configuration file to be better optimized
for most users.
- Add experimental RabbitMQ module.
- Add Kibana dashboard for the Kubernetes modules.
Heartbeat
- Changed the number of shards in the default configuration to 1.
- Enabled Cgo which means libc is dynamically compiled.
Packetbeat
- Changed the number of shards in the default configuration to 3.
Auditbeat
- Changed the number of shards in the default configuration to 3.
- Add support for receiving audit events using a multicast socket.
- Added `file.hash_types` config option for controlling the hash
types.
- Added the ability to specify byte unit suffixes to
`file.max_file_size`.
- Add file integrity metricset to the audit module.
==== Deprecated
Affecting all Beats
- The `@metadata.type` field, added by the Logstash output, is
deprecated, hardcoded to `doc` and will be removed in future
versions.
Filebeat
- The `filebeat.config_dir` option is deprecated. Use
`filebeat.config.prospector` options instead.
- Deprecate `input_type` prospector config. Use `type` config option
instead.
=== Beats version 5.6.4
- Fix race condition in internal logging rotator.
- Add support for enabling TLS renegotiation.
- Add setting to enable/disable the slow start in logstash output.
- Packetbeat: Fix missing length check in the PostgreSQL module.
777650d1b3