Collect Apache Hadoop logs

This document explains how to ingest Apache Hadoop logs to Google Security Operations using Bindplane. The parser first extracts fields from raw Hadoop logs using Grok patterns based on common Hadoop log formats. Then, it maps the extracted fields to the corresponding fields in the Unified Data Model (UDM) schema, performs data type conversions, and enriches the data with additional context.

Before you begin

Make sure you have the following prerequisites:

• A Google SecOps instance
• A Windows 2016 or later or Linux host with systemd
• If running behind a proxy, ensure firewall ports are open per the Bindplane agent requirements
• Privileged access to the Apache Hadoop cluster configuration files

Get Google SecOps ingestion authentication file

1. Sign in to the Google SecOps console.
2. Go to SIEM Settings > Collection Agents.
3. Download the Ingestion Authentication File. Save the file securely on the system where Bindplane will be installed.

Get Google SecOps customer ID

1. Sign in to the Google SecOps console.
2. Go to SIEM Settings > Profile.
3. Copy and save the Customer ID from the Organization Details section.

Install the Bindplane agent

Install the Bindplane agent on your Windows or Linux operating system according to the following instructions.

Windows installation

1. Open the Command Prompt or PowerShell as an administrator.

2. Run the following command:

   msiexec /i "https://github.com/observIQ/bindplane-agent/releases/latest/download/observiq-otel-collector.msi" /quiet 

Linux installation

1. Open a terminal with root or sudo privileges.

2. Run the following command:

   sudo sh -c "$(curl -fsSlL https://github.com/observiq/bindplane-agent/releases/latest/download/install_unix.sh)" install_unix.sh 

Additional installation resources

• For additional installation options, consult this installation guide.

Configure the Bindplane agent to ingest Syslog and send to Google SecOps

1. Access the configuration file:

   1. Locate the config.yaml file. Typically, it's in the /etc/bindplane-agent/ directory on Linux or in the installation directory on Windows.
   2. Open the file using a text editor (for example, nano, vi, or Notepad).

2. Edit the config.yaml file as follows:

   receivers:  udplog:  # Replace the port and IP address as required  listen_address: "0.0.0.0:514" exporters:  chronicle/chronicle_w_labels:  compression: gzip  # Adjust the path to the credentials file you downloaded in Step 1  creds_file_path: '/path/to/ingestion-authentication-file.json'  # Replace with your actual customer ID from Step 2  customer_id: <CUSTOMER_ID>  endpoint: malachiteingestion-pa.googleapis.com  # Add optional ingestion labels for better organization  log_type: 'HADOOP'  raw_log_field: body  ingestion_labels: service:  pipelines:  logs/source0__chronicle_w_labels-0:  receivers:  - udplog  exporters:  - chronicle/chronicle_w_labels 

   • Replace the port and IP address as required in your infrastructure.
   • Replace <CUSTOMER_ID> with the actual customer ID.
   • Update /path/to/ingestion-authentication-file.json to the path where the authentication file was saved in the Get Google SecOps ingestion authentication file section.

Restart the Bindplane agent to apply the changes

• To restart the Bindplane agent in Linux, run the following command:

  sudo systemctl restart bindplane-agent 

• To restart the Bindplane agent in Windows, you can either use the Services console or enter the following command:

  net stop BindPlaneAgent && net start BindPlaneAgent 

Configure Syslog forwarding on Apache Hadoop

Apache Hadoop uses Log4j for logging. Configure the appropriate Syslog appender based on your Log4j version so that Hadoop daemons (NameNode, DataNode, ResourceManager, NodeManager, etc.) forward logs directly to your syslog receiver (Bindplane host). Log4j is configured via files (no web UI).

Option 1: Log4j 1.x configuration

1. Locate the log4j.properties file (typically in $HADOOP_CONF_DIR/log4j.properties).

2. Add the following SyslogAppender configuration to the file:

   # Syslog appender (UDP example) log4j.appender.SYSLOG=org.apache.log4j.net.SyslogAppender log4j.appender.SYSLOG.SyslogHost=<BINDPLANE_HOST_IP>:514 log4j.appender.SYSLOG.Facility=LOCAL0 log4j.appender.SYSLOG.FacilityPrinting=true log4j.appender.SYSLOG.layout=org.apache.log4j.PatternLayout log4j.appender.SYSLOG.layout.ConversionPattern=%d{ISO8601} level=%p logger=%c thread=%t msg=%m%n # Example: send NameNode logs to syslog log4j.logger.org.apache.hadoop.hdfs.server.namenode=INFO,SYSLOG log4j.additivity.org.apache.hadoop.hdfs.server.namenode=false # Or attach to root logger to send all Hadoop logs # log4j.rootLogger=INFO, SYSLOG 

3. Replace <BINDPLANE_HOST_IP> with the IP address of your Bindplane host.

4. Save the file.

5. Restart Hadoop daemons to apply the configuration changes.

Option 2: Log4j 2.x configuration

1. Locate the log4j2.xml file (typically in $HADOOP_CONF_DIR/log4j2.xml).

2. Add the following Syslog appender configuration to the file:

   <Configuration status="WARN">  <Appenders>  <!-- UDP example; for TCP use protocol="TCP" -->  <Syslog name="SYSLOG" format="RFC5424"  host="<BINDPLANE_HOST_IP>"  port="514"  protocol="UDP"  facility="LOCAL0"  appName="hadoop"  enterpriseNumber="18060"  mdcId="mdc">  <PatternLayout pattern="%d{ISO8601} level=%p logger=%c thread=%t msg=%m %X%n"/>  </Syslog>  </Appenders>  <Loggers>  <!-- Send NameNode logs to syslog -->  <Logger name="org.apache.hadoop.hdfs.server.namenode" level="info" additivity="false">  <AppenderRef ref="SYSLOG"/>  </Logger>  <!-- Or send all Hadoop logs -->  <Root level="info">  <AppenderRef ref="SYSLOG"/>  </Root>  </Loggers> </Configuration> 

   • Replace <BINDPLANE_HOST_IP> with the IP address of your Bindplane host.

3. Save the file.

4. Restart Hadoop daemons to apply the configuration changes.

UDM Mapping Table

Log Field	UDM Mapping	Logic
allowed	security_result.action	If "false", action is "BLOCK". If "true", action is "ALLOW".
auth_type	additional.fields.key = "auth_type", additional.fields.value.string_value	Extracted from "ugi" field using grok pattern "%{DATA:suser}@.*auth:%{WORD:auth_type}". Parentheses and "auth:" are removed.
call	additional.fields.key = "Call#", additional.fields.value.string_value	Directly mapped.
call_context	additional.fields.key = "callerContext", additional.fields.value.string_value	Directly mapped.
cliIP	principal.ip	Mapped only when "json_data" field exists and is successfully parsed as JSON.
cmd	principal.process.command_line	Directly mapped.
cluster_name	target.hostname	Used as target hostname if present.
day	metadata.event_timestamp.seconds	Used with month, year, hours, minutes, and seconds to construct event_timestamp.
description	metadata.description	Directly mapped.
driver	additional.fields.key = "driver", additional.fields.value.string_value	Directly mapped.
dst	target.ip OR target.hostname OR target.file.full_path	If successfully parsed as IP, mapped to target IP. If the value starts with "/user", mapped to target file path. Otherwise, mapped to target hostname.
dstport	target.port	Directly mapped and converted to integer.
enforcer	security_result.rule_name	Directly mapped.
event_count	additional.fields.key = "event_count", additional.fields.value.string_value	Directly mapped and converted to string.
fname	src.file.full_path	Directly mapped.
hours	metadata.event_timestamp.seconds	Used with month, day, year, minutes, and seconds to construct event_timestamp.
id	additional.fields.key = "id", additional.fields.value.string_value	Directly mapped.
ip	principal.ip	Mapped to principal IP after removing any leading "/" character.
json_data		Parsed as JSON. Extracted fields are mapped to corresponding UDM fields.
logType	additional.fields.key = "logType", additional.fields.value.string_value	Directly mapped.
message		Used for extracting various fields using grok patterns.
method	network.http.method	Directly mapped.
minutes	metadata.event_timestamp.seconds	Used with month, day, year, hours, and seconds to construct event_timestamp.
month	metadata.event_timestamp.seconds	Used with day, year, hours, minutes, and seconds to construct event_timestamp.
observer	observer.hostname OR observer.ip	If successfully parsed as IP, mapped to observer IP. Otherwise, mapped to observer hostname.
perm	additional.fields.key = "perm", additional.fields.value.string_value	Directly mapped.
policy	security_result.rule_id	Directly mapped and converted to string.
product	metadata.product_name	Directly mapped.
product_event	metadata.product_event_type	Directly mapped. If "rename", the "dst" field is mapped to "target.file.full_path".
proto	network.application_protocol	Directly mapped and converted to uppercase if it's not "webhdfs".
reason	security_result.summary	Directly mapped.
repo	additional.fields.key = "repo", additional.fields.value.string_value	Directly mapped.
resType	additional.fields.key = "resType", additional.fields.value.string_value	Directly mapped.
result	additional.fields.key = "result", additional.fields.value.string_value	Directly mapped and converted to string.
Retry	additional.fields.key = "Retry#", additional.fields.value.string_value	Directly mapped.
seconds	metadata.event_timestamp.seconds	Used with month, day, year, hours, and minutes to construct event_timestamp.
seq_num	additional.fields.key = "seq_num", additional.fields.value.string_value	Directly mapped and converted to string.
severity	security_result.severity	Mapped to different severity levels based on the value: "INFO", "Info", "info" -> "INFORMATIONAL"; "Low", "low", "LOW" -> "LOW"; "error", "Error", "WARN", "Warn" -> "MEDIUM"; "High", "high", "HIGH" -> "HIGH"; "Critical", "critical", "CRITICAL" -> "CRITICAL".
shost	principal.hostname	Used as principal hostname if different from "src".
src	principal.ip OR principal.hostname OR observer.ip	If successfully parsed as IP, mapped to principal and observer IP. Otherwise, mapped to principal hostname.
srcport	principal.port	Directly mapped and converted to integer.
summary	security_result.summary	Directly mapped.
suser	principal.user.userid	Directly mapped.
tags	additional.fields.key = "tags", additional.fields.value.string_value	Directly mapped.
thread	additional.fields.key = "thread", additional.fields.value.string_value	Directly mapped.
tip	target.ip	Directly mapped.
ugi	target.hostname	Used as target hostname if "log_data" field doesn't contain "·".
url	target.url	Directly mapped.
vendor	metadata.vendor_name	Directly mapped.
version	metadata.product_version	Directly mapped.
year	metadata.event_timestamp.seconds	Used with month, day, hours, minutes, and seconds to construct event_timestamp.
N/A	metadata.event_type	Set to "NETWORK_CONNECTION" by default. Changed to "STATUS_UPDATE" if no target is identified.
N/A	metadata.log_type	Set to "HADOOP".
N/A	security_result.alert_state	Set to "ALERTING" if severity is "HIGH" or "CRITICAL".
N/A	is_alert	Set to "true" if severity is "HIGH" or "CRITICAL".
N/A	is_significant	Set to "true" if severity is "HIGH" or "CRITICAL".

Need more help? Get answers from Community members and Google SecOps professionals.