Using regular expressions in Trace Recordings

With the help of regular expressions, textual search patterns can be defined.

When used as filters for Trace Recordings the resulting output can be vastly more helpful.

Regular Expressions are a broad topic and cannot be covered in full here. We will demonstrate the use of HiveMQ’s pattern matching capabilities by creating a Trace Recording, matching all clients with a clientId beginning with publisher0, followed by a single character.

Prerequisites

In order to verify, that our messages are being delivered as we expect we will subscribe to a test topic.

HiveMQ’s MQTT CLI lets us do this in one line like so:

mqtt sub -t 'our/topic' -h localhost -i subscriber00

To have an easier time differentiating our clients, we gave this one the clientId subscriber00.

Let this session remain active throughout the test.

Configuring the Trace Recording

Initiating, managing and downloading Trace Recordings is done in the HiveMQ Control Center.

Here we must Add New Trace Recording in order to reach the configuration mask.

First, choose a new, unique name. This need not match a topic name and only serves to help identify the recording we are about to create.

Next, we have to define a Client Filter. Here we will make use of the wildcard character . that acts as a quantifier for “any single character”.

After clicking Add Topic Filter it should become visible as an existing filter:

From here we can select the types of messages we are interested in. For this example, we are only interested in PUBLISH messages.

Finally, after making sure that Start Date is set to a past time and the selected End Date is far enough into the future to cover our efforts, we are ready to Start Trace Recording.

Upon success, your new recording will show up in the overview with the name you specified:Upon success, your new recording will show up in the overview with the name you specified:

Generate test data

In order to verify our recording works as expected, we will publish messages.

You can use the following loop to send messages using both clientIDs containing a single character after publisher0 as well ones with an additional second character.

for i in `seq 5 15`; do sleep 1; mqtt pub -t 'our/topic' -h 127.0.0.1 -m "message by publisher0$i" -i publisher0$i; done

Once $i reaches 10, the clientID will contain a character more than we specified in our filter. We expect not to see messages from publisher010 to publisher015 in our recording.

Inspecting the Trace Recording

Once our publishing run is complete, we may

• stop the Trace Recording by clicking stop

• have HiveMQ prepare the resulting output by clicking download

• and click save to receive the resulting compressed trace recording

After unpacking the downloaded file, the resulting .trace file can be inspected with any text editor.

You should see a Sent and corresponding Received for each message, similar to the following:

2020-04-09 15:52:43,526 - [publisher05] - Received PUBLISH message (topic: our/topic, QoS: 0, retain: false, duplicate delivery: false)
2020-04-09 15:52:45,297 - [publisher06] - Received PUBLISH message (topic: our/topic, QoS: 0, retain: false, duplicate delivery: false)
[...]
2020-04-09 15:53:16,852 - [publisher09] - Received PUBLISH message (topic: our/topic, QoS: 0, retain: false, duplicate delivery: false)

Please refer to HiveMQ Docs for a list of available regular expression quantifiers.