In this guide, you’ll learn how to send a log file to Elasticsearch using a standalone Elastic Agent, configure the Elastic Agent and your data streams using the elastic-agent.yml file, and query your logs using the data streams you’ve set up. Once your log files are in Elasticsearch, see the Get the most out of your log data section to learn how to parse your log data and extract fields so you can filter and sort your logs effectively.

Prerequisitesedit

To follow the steps in this guide, you need an Elastic Stack deployment that includes:

To get started quickly, spin up a deployment of our hosted Elasticsearch Service. The Elasticsearch Service is available on AWS, GCP, and Azure. Try it out for free.

Install and configure the standalone Elastic Agentedit

Complete these steps to install and configure the standalone Elastic Agent and send your log data to Elasticsearch:

Step 1: Download and extract the Elastic Agent installation packageedit

On your host, download and extract the installation package that corresponds with your system:

curl -L -O https://artifacts.elastic.co/downloads/beats/elastic-agent/elastic-agent-8.9.2-darwin-x86_64.tar.gz
tar xzvf elastic-agent-8.9.2-darwin-x86_64.tar.gz

curl -L -O https://artifacts.elastic.co/downloads/beats/elastic-agent/elastic-agent-8.9.2-linux-x86_64.tar.gz
tar xzvf elastic-agent-8.9.2-linux-x86_64.tar.gz

# PowerShell 5.0+
wget https://artifacts.elastic.co/downloads/beats/elastic-agent/elastic-agent-8.9.2-windows-x86_64.zip -OutFile elastic-agent-8.9.2-windows-x86_64.zip
Expand-Archive .\elastic-agent-8.9.2-windows-x86_64.zip

curl -L -O https://artifacts.elastic.co/downloads/beats/elastic-agent/elastic-agent-8.9.2-amd64.deb
sudo dpkg -i elastic-agent-8.9.2-amd64.deb

curl -L -O https://artifacts.elastic.co/downloads/beats/elastic-agent/elastic-agent-8.9.2-x86_64.rpm
sudo rpm -vi elastic-agent-8.9.2-x86_64.rpm

Step 2: Install and start the Elastic Agentedit

After downloading and extracting the installation package, you’re ready to install the Elastic Agent. From the agent directory, run the install command that corresponds with your system:

sudo ./elastic-agent install

sudo ./elastic-agent install

.\elastic-agent.exe install

sudo systemctl enable elastic-agent 
sudo systemctl start elastic-agent

sudo systemctl enable elastic-agent 
sudo systemctl start elastic-agent

During installation, you’re prompted with some questions:

Step 3: Configure the Elastic Agentedit

With your agent installed, configure it by updating the elastic-agent.yml file.

Locate your configuration fileedit

After installing the agent, you’ll find the elastic-agent.yml in one of the following locations according to your system:

Update your configuration fileedit

The following is an example of a standalone Elastic Agent configuration. To configure your Elastic Agent, replace the contents of the elastic-agent.yml file with this configuration:

outputs:
  default:
    type: elasticsearch
    hosts: '<your-elasticsearch-endpoint>:<port>'
    api_key: 'your-api-key'
inputs:
  - id: your-log-id
    type: filestream
    streams:
      - id: your-log-stream-id
        data_stream:
          dataset: generic
        paths:
          - /var/log/your-logs.log

Next, set the values for these fields:

Restart the Elastic Agentedit

After updating your configuration file, you need to restart the Elastic Agent:

First, stop the Elastic Agent and its related executables using the command that works with your system:

sudo launchctl unload /Library/LaunchDaemons/co.elastic.elastic-agent.plist

sudo service elastic-agent stop

Stop-Service Elastic Agent

sudo systemctl stop elastic-agent

sudo service elastic-agent stop

sudo systemctl stop elastic-agent

sudo service elastic-agent stop

Next, restart the Elastic Agent using the command that works with your system:

sudo launchctl load /Library/LaunchDaemons/co.elastic.elastic-agent.plist

sudo service elastic-agent start

Start-Service Elastic Agent

sudo systemctl start elastic-agent

sudo service elastic-agent start

sudo systemctl start elastic-agent

sudo service elastic-agent start

View and search your dataedit

With your Elastic Agent and data streams configured, you can now view, filter, and search your log data. In Kibana, navigate to Observability → Logs → Stream, and use the search bar to search for your data_stream.type and data_stream.dataset.

See the following examples for ways to search specific data types and datasets:

This example shows the search results for logs with an apm.error dataset and a default namespace:

Troubleshoot your Elastic Agent configurationedit

If you’re not seeing your log files in Kibana, verify the following in the elastic-agent.yml file:

If you’re still running into issues, see Elastic Agent troubleshooting and Configure standalone Elastic Agents.

Get the most out of your log dataedit

Make your logs more useful by extracting structured fields from your unstructured log data. Extracting structured fields makes it easier to search, analyze, and filter your log data.

Let’s look at this log example:

2023-08-08T13:45:12.123Z WARN 192.168.1.101 Disk usage exceeds 90%.

Add this log to Elasticsearch using the following command in Dev Tools, found in your deployment in the left navigation under Management:

POST logs-example-default/_doc
{
  "message": "2023-08-08T13:45:12.123Z WARN 192.168.1.101 Disk usage exceeds 90%."
}

The previous command stores the document in logs-example-default. Retrieve it with the following search:

GET /logs-example-default/_search

You see something like this:

{
  ...
  "hits": {
    ...
    "hits": [
      {
        "_index": ".ds-logs-example-default-2023.08.09-000001",
        ...
        "_source": {
          "message": "2023-08-08T13:45:12.123Z WARN 192.168.1.101 Disk usage exceeds 90%.",
          "@timestamp": "2023-08-09T17:19:27.73312243Z"
        }
      }
    ]
  }
}

Elasticsearch indexes the message field by default meaning you can search for phrases like WARN or Disk usage exceeds. For example, the following command searches for the phrase WARN in the log message field:

GET logs-example-default/_search
{
  "query": {
    "match": {
      "message": {
        "query": "WARN"
      }
    }
  }
}

While you can search for phrases in the message field, you can’t use this field to filter log data. Your message, however, contains all of the following potential fields you can extract and use to filter and aggregate your log data:

Extract the @timestamp fieldedit

When you ingested the document in the previous section, you’ll notice the @timestamp field shows when you added the data to Elasticsearch, not when the log occurred:

        ...
        "_source": {
          "message": "2023-08-08T13:45:12.123Z WARN 192.168.1.101 Disk usage exceeds 90%.",
          "@timestamp": "2023-08-09T17:19:27.73312243Z"
        }
        ...

This section shows you how to extract the @timestamp field from the log message so you can filter by when the logs and issues actually occurred.

To extract the timestamp, you need to:

Use an ingest pipeline to extract the @timestampedit

Ingest pipelines consist of a series of processors that perform common transformations on incoming documents before they are indexed. To extract the @timestamp field from the example log, use an ingest pipeline with a dissect processor. The dissect processor extracts structured fields from unstructured log messages based on a pattern you set.

Elasticsearch can parse string timestamps that are in yyyy-MM-dd'T'HH:mm:ss.SSSZ and yyyy-MM-dd formats into date fields. Since the log example’s timestamp is in one of these formats, you don’t need additional processors. More complex or nonstandard timestamps require a date processor to parse the timestamp into a date field. Date processors can also set the timezone, change the target field, and change the output format of the timestamp.

In the following command, the dissect processor extracts the timestamp from the message field to the @timestamp field and leaves the rest of the message in the message field:

PUT _ingest/pipeline/logs-example-default
{
  "description": "Extracts the timestamp from log",
  "processors": [
    {
      "dissect": {
        "field": "message",
        "pattern": "%{@timestamp} %{message}"
      }
    }
  ]
}

The previous command sets the following values for your ingest pipeline:

Test your pipeline with the simulate pipeline APIedit

The simulate pipeline API runs the ingest pipeline without storing any documents. This lets you verify your pipeline works using multiple documents. Run the following command to test your ingest pipeline with the simulate pipeline API.

POST _ingest/pipeline/logs-example-default/_simulate
{
  "docs": [
    {
      "_source": {
        "message": "2023-08-08T13:45:12.123Z WARN 192.168.1.101 Disk usage exceeds 90%."
      }
    }
  ]
}

The results should show the @timestamp field extracted from the message field:

{
  "docs": [
    {
      "doc": {
        "_index": "_index",
        "_id": "_id",
        "_version": "-3",
        "_source": {
          "message": "WARN 192.168.1.101 Disk usage exceeds 90%.",
          "@timestamp": "2023-08-08T13:45:12.123Z"
        },
        ...
      }
    }
  ]
}

Configure your data stream with an index templateedit

After creating your ingest pipeline, create an index template to point your log data to your pipeline using this command:

PUT _index_template/logs-example-default-template
{
  "index_patterns": [ "logs-example-*" ],
  "data_stream": { },
  "priority": 500,
  "template": {
    "settings": {
      "index.default_pipeline":"logs-example-default"
    }
  },
  "composed_of": [
    "logs-mappings",
    "logs-settings",
    "logs@custom",
    "ecs@dynamic_templates"
  ],
  "ignore_missing_component_templates": ["logs@custom"]
}

The previous command sets the following values for your index template:

The component templates that are set in the previous index template are defined as follows:

Create your data streamedit

Create your data stream using the data stream naming scheme. Since the name needs to match the name of your pipeline, name the data stream logs-example-default. Post the example log to your data stream with this command:

POST logs-example-default/_doc
{
  "message": "2023-08-08T13:45:12.123Z WARN 192.168.1.101 Disk usage exceeds 90%."
}

View your documents using this command:

GET /logs-example-default/_search

You should see the pipeline has extracted the @timestamp field:

{
...
{
  ...
  "hits": {
    ...
    "hits": [
      {
        "_index": ".ds-logs-example-default-2023.08.09-000001",
        "_id": "RsWy3IkB8yCtA5VGOKLf",
        "_score": 1,
        "_source": {
          "message": "WARN 192.168.1.101 Disk usage exceeds 90%.",
          "@timestamp": "2023-08-08T13:45:12.123Z"
        }
      }
    ]
  }
}

You can now use the @timestamp field to sort your logs by the date and time they happened.

Troubleshoot your @timestamp fieldedit

Check the following common issues and solutions with timestamps:

Extract the log.level fieldedit

Extracting the log.level field lets you filter by severity and focus on critical issues. This section shows you how to extract the log.level field from this example log:

2023-08-08T13:45:12.123Z WARN 192.168.1.101 Disk usage exceeds 90%.

To extract and use the log.level field:

Add log.level to your ingest pipelineedit

Add the %{log.level} option to the dissect processor pattern in the ingest pipeline you created in the Extract the @timestamp field section:

PUT _ingest/pipeline/logs-example-default
{
  "description": "Extracts the timestamp from log",
  "processors": [
    {
      "dissect": {
        "field": "message",
        "pattern": "%{@timestamp} %{log.level} %{message}"
      }
    }
  ]
}

Now your pipeline will extract these fields:

After creating your pipeline, an index template points your log data to your pipeline. Use the index template you created in the Extract the @timestamp field section.

Test the pipeline with the simulate APIedit

Test that your ingest pipeline works as expected with the simulate pipeline API:

POST _ingest/pipeline/logs-example-default/_simulate
{
  "docs": [
    {
      "_source": {
        "message": "2023-08-08T13:45:12.123Z WARN 192.168.1.101 Disk usage exceeds 90%."
      }
    }
  ]
}

The results should show the @timestamp and the log.level fields extracted from the message field:

{
  "docs": [
    {
      "doc": {
        "_index": "_index",
        "_id": "_id",
        "_version": "-3",
        "_source": {
          "message": "192.168.1.101 Disk usage exceeds 90%.",
          "log": {
            "level": "WARN"
          },
          "@timestamp": "2023-8-08T13:45:12.123Z",
        },
        ...
      }
    }
  ]
}

Query logs based on log.leveledit

Once you’ve extracted the log.level field, you can query for high-severity logs like WARN and ERROR, which may need immediate attention, and filter out less critical INFO and DEBUG logs.

Let’s say you have the following logs with varying severities:

2023-08-08T13:45:12.123Z WARN 192.168.1.101 Disk usage exceeds 90%.
2023-08-08T13:45:14.003Z ERROR 192.168.1.103 Database connection failed.
2023-08-08T13:45:15.004Z DEBUG 192.168.1.104 Debugging connection issue.
2023-08-08T13:45:16.005Z INFO 192.168.1.102 User changed profile picture.

Add them to your data stream using this command:

POST logs-example-default/_bulk
{ "create": {} }
{ "message": "2023-08-08T13:45:12.123Z WARN 192.168.1.101 Disk usage exceeds 90%." }
{ "create": {} }
{ "message": "2023-08-08T13:45:14.003Z ERROR 192.168.1.103 Database connection failed." }
{ "create": {} }
{ "message": "2023-08-08T13:45:15.004Z DEBUG 192.168.1.104 Debugging connection issue." }
{ "create": {} }
{ "message": "2023-08-08T13:45:16.005Z INFO 192.168.1.102 User changed profile picture." }

Then, query for documents with a log level of WARN or ERROR with this command:

GET logs-example-default/_search
{
  "query": {
    "terms": {
      "log.level": ["WARN", "ERROR"]
    }
  }
}

You should see the following results showing only your high-severity logs:

{
...
  },
  "hits": {
  ...
    "hits": [
      {
        "_index": ".ds-logs-example-default-2023.08.14-000001",
        "_id": "3TcZ-4kB3FafvEVY4yKx",
        "_score": 1,
        "_source": {
          "message": "192.168.1.101 Disk usage exceeds 90%.",
          "log": {
            "level": "WARN"
          },
          "@timestamp": "2023-08-08T13:45:12.123Z"
        }
      },
      {
        "_index": ".ds-logs-example-default-2023.08.14-000001",
        "_id": "3jcZ-4kB3FafvEVY4yKx",
        "_score": 1,
        "_source": {
          "message": "192.168.1.103 Database connection failed.",
          "log": {
            "level": "ERROR"
          },
          "@timestamp": "2023-08-08T13:45:14.003Z"
        }
      }
    ]
  }
}

Extract the host.ip fieldedit

Extracting the host.ip field lets you filter logs by host IP addresses allowing you to focus on specific hosts that you’re having issues with or find disparities between hosts.

The host.ip field is part of the Elastic Common Schema (ECS). Through the ECS, the host.ip field is mapped as an ip field type. ip field types allow range queries so you can find logs with IP addresses in a specific range. You can also query ip field types using CIDR notation to find logs from a particular network or subnet.

This section shows you how to extract the host.ip field from the following example logs and query based on the extracted fields:

2023-08-08T13:45:12.123Z WARN 192.168.1.101 Disk usage exceeds 90%.
2023-08-08T13:45:14.003Z ERROR 192.168.1.103 Database connection failed.
2023-08-08T13:45:15.004Z DEBUG 192.168.1.104 Debugging connection issue.
2023-08-08T13:45:16.005Z INFO 192.168.1.102 User changed profile picture.

To extract and use the host.ip field:

Add host.ip to your ingest pipelineedit

Add the %{host.ip} option to the dissect processor pattern in the ingest pipeline you created in the Extract the @timestamp field section:

PUT _ingest/pipeline/logs-example-default
{
  "description": "Extracts the timestamp from log",
  "processors": [
    {
      "dissect": {
        "field": "message",
        "pattern": "%{@timestamp} %{log.level} %{host.ip} %{message}"
      }
    }
  ]
}

Your pipeline will extract these fields:

After creating your pipeline, an index template points your log data to your pipeline. Use the index template you created in the Extract the @timestamp field section.

Test the pipeline with the simulate APIedit

Test that your ingest pipeline works as expected with the simulate pipeline API:

POST _ingest/pipeline/logs-example-default/_simulate
{
  "docs": [
    {
      "_source": {
        "message": "2023-08-08T13:45:12.123Z WARN 192.168.1.101 Disk usage exceeds 90%."
      }
    }
  ]
}

The results should show the host.ip, @timestamp, and log.level fields extracted from the message field:

{
  "docs": [
    {
      "doc": {
        ...
        "_source": {
          "host": {
            "ip": "192.168.1.101"
          },
          "@timestamp": "2023-08-08T13:45:12.123Z",
          "message": "Disk usage exceeds 90%.",
          "log": {
            "level": "WARN"
          }
        },
        ...
      }
    }
  ]
}

Query logs based on host.ipedit

You can query your logs based on the host.ip field in different ways, including using CIDR notation and range queries.

Before querying your logs, add them to your data stream using this command:

POST logs-example-default/_bulk
{ "create": {} }
{ "message": "2023-08-08T13:45:12.123Z WARN 192.168.1.101 Disk usage exceeds 90%." }
{ "create": {} }
{ "message": "2023-08-08T13:45:14.003Z ERROR 192.168.1.103 Database connection failed." }
{ "create": {} }
{ "message": "2023-08-08T13:45:15.004Z DEBUG 192.168.1.104 Debugging connection issue." }
{ "create": {} }
{ "message": "2023-08-08T13:45:16.005Z INFO 192.168.1.102 User changed profile picture." }

CIDR notationedit

You can use CIDR notation to query your log data using a block of IP addresses that fall within a certain network segment. CIDR notations uses the format of [IP address]/[prefix length]. The following command queries IP addresses in the 192.168.1.0/24 subnet meaning IP addresses from 192.168.1.0 to 192.168.1.255.

GET logs-example-default/_search
{
  "query": {
    "term": {
      "host.ip": "192.168.1.0/24"
    }
  }
}

Because all of the example logs are in this range, you’ll get the following results:

{
  ...
  },
  "hits": {
    ...
      {
        "_index": ".ds-logs-example-default-2023.08.16-000001",
        "_id": "ak4oAIoBl7fe5ItIixuB",
        "_score": 1,
        "_source": {
          "host": {
            "ip": "192.168.1.101"
          },
          "@timestamp": "2023-08-08T13:45:12.123Z",
          "message": "Disk usage exceeds 90%.",
          "log": {
            "level": "WARN"
          }
        }
      },
      {
        "_index": ".ds-logs-example-default-2023.08.16-000001",
        "_id": "a04oAIoBl7fe5ItIixuC",
        "_score": 1,
        "_source": {
          "host": {
            "ip": "192.168.1.103"
          },
          "@timestamp": "2023-08-08T13:45:14.003Z",
          "message": "Database connection failed.",
          "log": {
            "level": "ERROR"
          }
        }
      },
      {
        "_index": ".ds-logs-example-default-2023.08.16-000001",
        "_id": "bE4oAIoBl7fe5ItIixuC",
        "_score": 1,
        "_source": {
          "host": {
            "ip": "192.168.1.104"
          },
          "@timestamp": "2023-08-08T13:45:15.004Z",
          "message": "Debugging connection issue.",
          "log": {
            "level": "DEBUG"
          }
        }
      },
      {
        "_index": ".ds-logs-example-default-2023.08.16-000001",
        "_id": "bU4oAIoBl7fe5ItIixuC",
        "_score": 1,
        "_source": {
          "host": {
            "ip": "192.168.1.102"
          },
          "@timestamp": "2023-08-08T13:45:16.005Z",
          "message": "User changed profile picture.",
          "log": {
            "level": "INFO"
          }
        }
      }
    ]
  }
}

Range queriesedit

Use range queries to query logs in a specific range.

The following command searches for IP addresses greater than or equal to 192.168.1.100 and less than or equal to 192.168.1.102.

GET logs-example-default/_search
{
  "query": {
    "range": {
      "host.ip": {
        "gte": "192.168.1.100",
        "lte": "192.168.1.102"
      }
    }
  }
}

You’ll get the following results matching the range you’ve set:

{
  ...
  },
  "hits": {
    ...
      {
        "_index": ".ds-logs-example-default-2023.08.16-000001",
        "_id": "ak4oAIoBl7fe5ItIixuB",
        "_score": 1,
        "_source": {
          "host": {
            "ip": "192.168.1.101"
          },
          "@timestamp": "2023-08-08T13:45:12.123Z",
          "message": "Disk usage exceeds 90%.",
          "log": {
            "level": "WARN"
          }
        }
      },
      {
        "_index": ".ds-logs-example-default-2023.08.16-000001",
        "_id": "bU4oAIoBl7fe5ItIixuC",
        "_score": 1,
        "_source": {
          "host": {
            "ip": "192.168.1.102"
          },
          "@timestamp": "2023-08-08T13:45:16.005Z",
          "message": "User changed profile picture.",
          "log": {
            "level": "INFO"
          }
        }
      }
    ]
  }
}

Ignore malformed IP addressesedit

When you’re ingesting a large batch of log data, a single malformed IP address can cause the entire batch to fail. Prevent this by setting ignore_malformed to true for the host.ip field. Update the host.ip field to ignore malformed IPs using the update mapping API:

PUT /logs-example-default/_mapping
{
  "properties": {
    "host.ip": {
      "type": "ip",
      "ignore_malformed": true
    }
  }
}

Reroute log data to specific data streamedit

By default, an ingest pipeline sends your log data to a single data stream. To simplify log data management, use a reroute processor to route data from the generic data stream to a target data stream. For example, you might want to send high-severity logs to a specific data stream to help with categorization.

This section shows you how to use a reroute processor to send the high-severity logs (WARN or ERROR) from the following example logs to a specific data stream and keep the regular logs (DEBUG and INFO) in the default data stream:

2023-08-08T13:45:12.123Z WARN 192.168.1.101 Disk usage exceeds 90%.
2023-08-08T13:45:14.003Z ERROR 192.168.1.103 Database connection failed.
2023-08-08T13:45:15.004Z DEBUG 192.168.1.104 Debugging connection issue.
2023-08-08T13:45:16.005Z INFO 192.168.1.102 User changed profile picture.

To use a reroute processor:

Add a reroute processor to your ingest pipelineedit

Add a reroute processor to your ingest pipeline with the following command:

PUT _ingest/pipeline/logs-example-default
{
  "description": "Extracts fields and reroutes WARN",
  "processors": [
    {
      "dissect": {
        "field": "message",
        "pattern": "%{@timestamp} %{log.level} %{host.ip} %{message}"
      },
      "reroute": {
        "tag": "high_severity_logs",
        "if" : "ctx.log?.level == 'WARN' || ctx.log?.level == 'ERROR'",
        "dataset": "critical"
      }
    }
  ]
}

The previous command sets the following values for your reroute processor:

After creating your pipeline, an index template points your log data to your pipeline. Use the index template you created in the Extract the @timestamp field section.

Add logs to your data streamedit

Add the example logs to your data stream with this command:

POST logs-example-default/_bulk
{ "create": {} }
{ "message": "2023-08-08T13:45:12.123Z WARN 192.168.1.101 Disk usage exceeds 90%." }
{ "create": {} }
{ "message": "2023-08-08T13:45:14.003Z ERROR 192.168.1.103 Database connection failed." }
{ "create": {} }
{ "message": "2023-08-08T13:45:15.004Z DEBUG 192.168.1.104 Debugging connection issue." }
{ "create": {} }
{ "message": "2023-08-08T13:45:16.005Z INFO 192.168.1.102 User changed profile picture." }

Verify that the reroute processor workededit

The reroute processor should route any logs with a log.level of WARN or ERROR to the logs-critical-default data stream. Query the the data stream using the following command to verify the log data was routed as intended:

GET log-critical-default/_search

Your query should return similar results to the following:

{
  ...
  "hits": {
    ...
    "hits": [
        ...
        "_source": {
          "host": {
            "ip": "192.168.1.101"
          },
          "@timestamp": "2023-08-08T13:45:12.123Z",
          "message": "Disk usage exceeds 90%.",
          "log": {
            "level": "WARN"
          },
          "data_stream": {
            "namespace": "default",
            "type": "logs",
            "dataset": "critical"
          },
          {
        ...
        "_source": {
          "host": {
            "ip": "192.168.1.103"
           },
          "@timestamp": "2023-08-08T13:45:14.003Z",
          "message": "Database connection failed.",
          "log": {
            "level": "ERROR"
          },
          "data_stream": {
            "namespace": "default",
            "type": "logs",
            "dataset": "critical"
          }
        }
      }
    ]
  }
}

You can see the high-severity logs and that they’re now in the critical dataset.