Apache Nifi: Send From MySQL to Syslog

Our prior post in this series, Ingest Syslog Data Via Apache NiFi, covered the basics of using Apache NiFi to load balance incoming syslog events. Here you’ll go a step further by querying a MySQL database (our example contains entity events from Home Assistant, the open source home automation toolset). Shown in Figure 1, several linked NiFi processors will reformat the output to JSON on its way to a syslog server.

Java Connector – Assuming you’ve installed NiFi in /opt/nifi per instructions in the earlier post, start by downloading the required Java connector.

mkdir -p /opt/nifi/drivers
cd /opt/nifi/drivers
wget https://repo1.maven.org/maven2/mysql/mysql-connector-java/8.0.19/mysql-connector-java-8.0.19.jar

QueryDatabaseTable – You now have the MySQL connection drivers. Next you’ll create a QueryDatabaseTableprocessor.

Drag the processor onto the canvas, then double-click it and go to settings. Click the Scheduling tab, then Run Schedule. Change it to query the table every 15 seconds.

Now go to the Properties tab to configure the MySQL service. Click the blank space adjacent to Database Connection Pooling Service, then select Create New Service (Figure 2).

Rename the Controller Service Name to something such as MySQL Connection (Figure 3). Click Create when finished.

On the settings page you’ll see a → to the right of the Database Connection Pooling Service in the processor settings. Click the arrow, then click the gear icon adjacent to your MySQL Connection service.

Next change the hostname and database variables in the connection URL to reflect your own environment. Figure 4 shows a completed example. 

Database Connection URL:
jdbc:mysql://:3306/?serverTimezone=UTC
Database Driver Class Name: com.mysql.jdbc.Driver
Database Driver Location(s):
/opt/nifi/drivers/mysql-connector-java-8.0.19.jar
Database User: A user with permissions to access the DB.
Database Password: Password for above.

Also ensure the service is enabled (Figure 5).

Now return to the QueryDatabaseTable processor settings where you’ll plug in the database settings on its Properties tab.

Here, Maximum-value Columns acts as a placeholder so the processor knows which data had previously been collected (you want the connection to repeatedly pull the same data). Whatever its value, the column should auto-increment as tabular rows are added (e.g., the state_id column in Home Assistant increments by 1 in Figure 6).


Database Connection Pooling Service: The one you'd created above.
Database Type: MySQL
Table Name: The table of the database you're querying.
Maximum-value Columns: The incrementing value column.

With your database connection configured, you’ll now want to convert that processor’s data output from Avro to JSON format.

ConvertAvroToJSON – Create a ConvertAvroToJSON processor and connect the QueryDatabaseTable to it. Select success under Automatically Terminate Relationships in its settings (likely the only option).

SplitJSON – Unmodified, the flow will include a huge number of database events in a giant blob. You’ll split these into individual events using a SplitJSON processor to pass along to the next series processor. There is only one value to change for SplitJSON (figure 7):

JsonPath Expression: $.*

You’ll also want to select failure and original under Automatically Terminate Relationships in its settings.

Connect ConvertAvroToJSON to the SplitJSON processor using both success and failure for its Automatically Terminate Relationships settings.

The last step is to output the data in pseudo-syslog format. That is, you’ll send it somewhere on 514/UDP and leave it to the syslog receiver to timestamp it.

PutUDP – Last in the chain, create a PutUDP processor, for which you’ll only plug in its hostname and port. Also, go to the processor settings and check failure and success under Automatically Terminate Relationships (Figure 8).

Finally, connect the SplitJSON and PutUDP processors. Select split under Automatically Terminate Relationships since you only want to send over the split JSON data.

The final step is to right-click the canvas and select Start to fire up your NiFi processor chain. The processor indicators should turn green.

This NiFi scenario can likely be adapted to just about any JDBC-supported database.

I hope you found this useful. Hit me up with questions on NiFi with a comment on our social media platforms.