Elasticsearch logstash Search Guard

Log ingestion with Beats, logstash and Search Guard

by pmelo on March 20, 2019

Elasticsearch and its ecosystem provide a great stack for ingesting, storing and searching logs. Search Guard adds important safeguards to protect log data over its lifetime. It can be also easily integrated with all tools from the log analytics ecosystem. In this article we present a demo of Search Guard & Beats & logstash integration.


Beats is a “lightweight data shipper” that serves as an agent on a server where logs are produced. Ingesting logs with Beats into Elasticsearch can be done in two ways:

  1. We can define beats output to store data in Elasticsearch, which means we transport data directly from Beats agent to a cluster.
  2. We can export logs from Beats into Logstash, and then into ElasticSearch.

Both pipelines are supported by Search Guard and we start with the first approach. We will install filebeat and configure a log input from a local file.

We install a fresh demo version of ElasticSearch and Kibana, both with Search Guard plugins enabled. The sefault demo configuration already contains a user logstash (with a password logstash), and a sg_logstash role assigned to a user. The user has been granted permissions on indices logstash-* and *beat*. We do not need to apply any changes on Elasticsearch cluster.

We add the following entries into filebeat.yml that will fetch log lines from a local file /var/log/searchguard_test.log:


- type: log
enabled: true
- /var/log/searchguard_test.log
json.keys_under_root: true
json.add_error_key: true
json.message_key: log

We also define an output to Elasticsearch over HTTPS protocol with user and password provided:

setup.template.name: "filebeat"
setup.template.pattern: "filebeat-*"

  # Array of hosts to connect to.
  hosts: ["localhost:9200"]
  index: "beats-%{+yyyy.MM.dd}"
  protocol: "https"
  username: "logstash"
  password: "logstash"

As a demo, we can echo some JSON log into searchguard_test.log:

echo "{ \"level\": \"info\", \"message\": \"SearchGuard beat log sample\"}" >> /var/log/searchguard_test.log

After a moment, log will be available in Elasticsearch and Kibana. We can use a demo user kibanaro (password: kibanaro) to create an index pattern and browse incoming logs:

Search Guard Beats Elasticsearch


A downside of our previous approach is that in order to do modify our log pipeline, we need to modify the configuration on all Beats servers. Additionally, all the Beats servers store username and password to access Elasticsearch which can lead to security concerns.

That is why, it is so popular to use Beats & logstash together:

  • Use beats to capture logs on servers and pass them into logstash.
  • Use logstash to output the logs into ElasticSearch.

We will now modify our previous example to make it work this way. We modify filebeat.yml:

  • Remove ElasticSearch output from previous sections.
  • Add logstash output of the form:
 hosts: ["localhost:5044"]

We download logstash and create or modify sample pipeline. We can alter logstash-sample.conf:

input {
 beats {
   port => 5044

output {
 elasticsearch {
   hosts => ["https://localhost:9200"]
   index => "logstash-%{[@metadata][version]}-%{+YYYY.MM.dd}"
   user => "logstash"
   password => "logstash"
   ssl => true
   cacert => "path-to-root-CA/config/root-ca.pem"
   ssl_certificate_verification => true

We specify the root-ca certificate that has been generated for the demo configuration. The file should be located within config directory of Elasticsearch.

We start logstash with a given pipeline definition:

/bin/logstash -f logstash-sample.conf

We can now fire some logs:

echo "{ \"level\": \"info\", \"message\": \"SearchGuard beats & logstash log sample\"}" >> /var/log/searchguard_test.log

and check if they appear in Kibana. Note that we write into different indices and a separate index pattern needs to be created.

Search Guard logstash Elasticsearch


In this post we have described two possible log ingestion setups:

  • Beats -> ElasticSearch
  • Beats -> Logstash -> Elasticsearch

We explained differences between the approaches and we have shown that Search Guard is ready for both of them. Feel free to choose the one that suits your needs best.

Where to go next

Image: shutterstock / memej

pmeloLog ingestion with Beats, logstash and Search Guard

1 comment

Join the conversation
  • Masscv - April 9, 2019 reply

    Great post!

Join the conversation