diff --git a/packetbeat/docs/howto/howto.asciidoc b/packetbeat/docs/howto/howto.asciidoc
index cdadf3cb7b35..b7284ab3024b 100644
--- a/packetbeat/docs/howto/howto.asciidoc
+++ b/packetbeat/docs/howto/howto.asciidoc
@@ -23,6 +23,8 @@ include::{libbeat-dir}/howto/load-dashboards.asciidoc[]
 
 include::{libbeat-dir}/shared-geoip.asciidoc[]
 
+include::load-ingest-pipelines.asciidoc[]
+
 :standalone:
 include::{libbeat-dir}/shared-env-vars.asciidoc[]
 :standalone!:
diff --git a/packetbeat/docs/howto/load-ingest-pipelines.asciidoc b/packetbeat/docs/howto/load-ingest-pipelines.asciidoc
new file mode 100644
index 000000000000..acca824829c5
--- /dev/null
+++ b/packetbeat/docs/howto/load-ingest-pipelines.asciidoc
@@ -0,0 +1,28 @@
+[[load-ingest-pipelines]]
+== Load ingest pipelines
+
+{beatname_uc} modules are implemented using {es} ingest node
+pipelines.  The events receive their transformations within
+{es}.  The ingest node pipelines must be loaded
+into {es}.  This can happen one of several ways.
+
+[id="{beatname_lc}-load-pipeline-auto"]
+[float]
+=== On connection to {es}
+
+{beatname_uc} will send ingest pipelines automatically to {es} if the
+{es} output is enabled.
+
+Make sure the user specified in +{beatname_lc}.yml+ is
+<<privileges-to-setup-beats,authorized to set up {beatname_uc}>>.
+
+If {beatname_uc} is sending events to {ls} or another output you need
+to load the ingest pipelines with the `setup` command or manually.
+
+[id="{beatname_lc}-load-pipeline-manual"]
+[float]
+=== Manually install pipelines
+
+Pipelines can be loaded them into {es} with the `_ingest/pipeline` REST API
+call. The user making the REST API call will need to have the `ingest_admin`
+role assigned to them.
diff --git a/packetbeat/docs/modules.asciidoc b/packetbeat/docs/modules.asciidoc
new file mode 100644
index 000000000000..8e72454f9cff
--- /dev/null
+++ b/packetbeat/docs/modules.asciidoc
@@ -0,0 +1,41 @@
+[id="{beatname_lc}-modules"]
+[role="xpack"]
+= Modules
+
+[partintro]
+--
+This section contains detailed information about the available network packet
+log processing modules contained in {beatname_uc}.
+
+{beatname_uc} modules are implemented using Elasticsearch Ingest Node pipelines.
+The events receive their transformations within Elasticsearch. All events are
+sent through {beatname_uc}'s "routing" pipeline that routes events to specific
+module pipelines based on their network protocol.
+
+{beatname_uc}'s default config file contains the option to send all events to
+the routing pipeline. If you remove this option then the module processing
+will not be applied.
+
+[source,yaml,subs="attributes"]
+----
+output.elasticsearch.pipeline: packetbeat-%{[agent.version]}-routing
+----
+
+The general goal of each module is to transform events by renaming fields to
+comply with the {ecs-ref}/index.html[Elastic Common Schema] (ECS). The modules
+may also apply additional categorization, tagging, and parsing as necessary.
+about how to configure the language in `packetbeat`, refer to <<configuration-packetbeat-options>>.
+
+[id="{beatname_lc}-modules-setup"]
+[float]
+=== Setup of Ingest Node pipelines
+
+{beatname_uc}'s Ingest Node pipelines must be installed to Elasticsearch if you
+want to apply the module processing to events. The simplest way to get started
+is to use the Elasticsearch output and {beatname_uc} will automatically install
+the pipelines when it first connects to Elasticsearch.
+
+Installation Methods
+
+1. <<{beatname_lc}-load-pipeline-auto>>
+2. <<{beatname_lc}-load-pipeline-manual>>
diff --git a/packetbeat/docs/packetbeat-options.asciidoc b/packetbeat/docs/packetbeat-options.asciidoc
index c5cb4d95d6b8..c48b4a1b01d0 100644
--- a/packetbeat/docs/packetbeat-options.asciidoc
+++ b/packetbeat/docs/packetbeat-options.asciidoc
@@ -1650,3 +1650,12 @@ Example configuration:
 -------------------------------------------------------------------------------------
 packetbeat.shutdown_timeout: 5s
 -------------------------------------------------------------------------------------
+
+[float]
+==== `overwrite_pipelines`
+
+By default Ingest pipelines are not updated if a pipeline with the same ID
+already exists. If this option is enabled {beatname_uc} overwrites pipelines
+every time a new Elasticsearch connection is established.
+
+The default value is `false`.
diff --git a/winlogbeat/docs/howto/load-ingest-pipelines.asciidoc b/winlogbeat/docs/howto/load-ingest-pipelines.asciidoc
index fa795f0b6b2a..0d7f842249e1 100644
--- a/winlogbeat/docs/howto/load-ingest-pipelines.asciidoc
+++ b/winlogbeat/docs/howto/load-ingest-pipelines.asciidoc
@@ -24,7 +24,7 @@ to load the ingest pipelines with the `setup` command or manually.
 === setup command
 
 On a machine that has {beatname_uc} installed and has {es} configured
-as the outup, run the `setup` command with the `--pipelines` option
+as the output, run the `setup` command with the `--pipelines` option
 specified.  For example, the following command loads the ingest
 pipelines: