1. jmeter-logstash

Jmeter JTL ans statiscts file parsing with Logstash and elasticsearch, you can find image on Docker Hub, statistics.json file is generated with jmeter html report.

2. Quick reference

Where to get help:

2.1. Image version

2.2. Features

Parse Standard JTL (CSV Format).
Possibility to filter requests based on regex filter (include and exclude filter) .
Flag samplers generated by TransactionController based on regex (by default '.+').
For TransactionController, calculate number of failing sampler and total.
Add relative time to compare different Executions, variable TESTSTART.MS should be logged with property sample_variables
Add Project name, test name, environment and executionId to organize results and compare different execution.
Split Label name to have multi tags by request (by default split by '/').
Flag subresult, when there is a redirection 302,.. Subrequest has a have a suffix like "-xx" when xx is the order number
Supporting ElasticSearch and can be adapted for other tools.
can also index custom field logged in file with property : sample_variables

2.4. Image Variants

The jmeter-logstash images come in many flavors, each designed for a specific use case. The images version are based on component used to build image, default use elasticsearch output:

Logstash Version: 7.17.9 -> default for 7.17.

3. Getting Started

3.1. Create ElasticSearch stack (Only if using ElasticSearch & Kibana)

Create Optional docker network (Called jmeter). If not used remove "--net jmeter " from all following docker command and adapt Elasticsearch url.

docker network create jmeter

Start elastic search container , Or use any Elasticsearch instance you have already installed.

docker run --name jmeter-elastic  --net jmeter \
-p 9200:9200 -p 9300:9300 \
-e "ES_JAVA_OPTS=-Xms1024m -Xmx1024m" \
-e "xpack.security.enabled=false" \
-e "discovery.type=single-node" \
docker.elastic.co/elasticsearch/elasticsearch:8.4.1

Start Kibana and connect it to elastic search Using environnement variable ELASTICSEARCH_HOSTS.

docker run --name jmeter-kibana  --net jmeter -p 5601:5601 -e "ELASTICSEARCH_HOSTS=http://jmeter-elastic:9200" docker.elastic.co/kibana/kibana:8.4.1

3.2. Run Logstash

In the project folder create a folder named 'input' or you can use any input folder in your machine.
If you choose to use a different input folder, you should change "${PWD}/input" on the following command by your input folder.

3.2.1. Run With image from docker hub for Elasticsearch (way 1, preferred)

#Run Image
docker run --rm -it --net jmeter -e "ELASTICSEARCH_HOSTS=http://jmeter-elastic:9200" -v ${PWD}/input:/input/ anasoid/jmeter-logstash

3.3. Dashboards

3.3.1. Kibana

Download Dashboards from Kibana Dashboards and go to Stack management kibana -> saved object for import.

Main Dashboard	Compare Dashboard

3.4. HOW-TO

To exit after all files parsed use (-e "FILE_EXIT_AFTER_READ=true") should be used with ( -e " FILE_READ_MODE=read") ..
To not remove container logstash after execution not use --rm from arguments.
Logstash keep information on position f last line parsed in a file sincedb, this file by default is on the path /usr/share/logstash/data/plugins/inputs/file, if you use the same container this file will be persisted even you restart logstash cotainer, and if you need to maintain this file even you remove container you can mount volume folder in the path (/usr/share/logstash/data/plugins/inputs/file)
To have relative time for comparison test start time should be logged : in user.properties file (sample_variables=TESTSTART.MS,...) or add properties with file using -q argument or directly in command line with (-Jsample_variables=TESTSTART.MS,..) see Full list of command-line options

3.4.1. Example

Run Logstash without remove container after stop.

docker run   -it --net jmeter -e "ELASTICSEARCH_HOSTS=http://jmeter-elastic:9200" \
-v ${PWD}/input:/input/ \
anasoid/jmeter-logstash

Run Logstash and remove file after the end of reading.

docker run --rm -it --net jmeter -e "ELASTICSEARCH_HOSTS=http://jmeter-elastic:9200" \
-v ${PWD}/input:/input/ \
-e "FILE_READ_MODE=read" \
-e "FILE_COMPLETED_ACTION=delete" \
anasoid/jmeter-logstash


Run Logstash with external sincedb folder.

```shell
docker run --rm -it --net jmeter -e "ELASTICSEARCH_HOSTS=http://jmeter-elastic:9200" \
-v ${PWD}/input:/input/ \
-v ${PWD}/.sincedb:/usr/share/logstash/data/plugins/inputs/file \
anasoid/jmeter-logstash

3.5. Parameters

3.5.1. ElasticSearch configuration

Environment variables	Description	Default
`ELASTICSEARCH_HOSTS`	Elasticsearch output configuration hosts (ex: http://elasticsearch:9200 )
`ELASTICSEARCH_INDEX`	Elasticsearch output configuration index	jmeter-jtl-%{+YYYY.MM.dd}
`ELASTICSEARCH_USER`	Elasticsearch output configuration user
`ELASTICSEARCH_PASSWORD`	Elasticsearch output configuration password
`ELASTICSEARCH_SSL_VERIFICATION`	Elasticsearch output configuration ssl_certificate_verification	true
`ELASTICSEARCH_HTTP_COMPRESSION`	Elasticsearch output configuration http_compression	false
`ELASTICSEARCH_VERSION`	Elasticsearch template version, shoud be the same as elasticsearch version (not logstash version), valid values are 7 and 8. Only logstash 7 who can work with Elasticsearch 7 and 8, logstash 8 work only with elasticsearch 8.	7 for logstash 7.x et 8 for logstash 8.x

3.5.2. Logstash

Environment variables	Description	Default
`INPUT_PATH`	Default folder input used for JTL and statistics file.	/input
`INPUT_PATH_JTL`	Default folder input used for JTL, pattern : (`["${INPUT_PATH:/input}/.jtl","${INPUT_PATH_JTL:/input}/.jtl"]`)	/input
`INPUT_PATH_STAT`	Default folder input used statistics , pattern : (`["${INPUT_PATH:/input}/.json","${INPUT_PATH_STAT:/input}/.json"]`)	/input
`PROJECT_NAME`	Project name	undefined
`ENVIRONMENT_NAME`	Environment name, if not provided will try to extract value from file name ( {test_name}-{environment-name}-{execution_id} )	undefined
`TEST_NAME`	Test name, if not provided will try to extract value from file name ( {test_name}-{environment-name}-{execution_id} or {test_name}-{execution_id} or {test_name})	undefined
`TEST_METADATA`	Test metadata as key value ex : (version=v1,type=daily,region=europe).
`TEST_TAGS`	Test tags, can add testtags fields as liste of value ex : (v1,daily,europe) version.
`EXECUTION_ID`	Execution Id, if not provided will try to extract value from file name ( {test_name}-{environment-name}-{execution_id} or {test_name}-{execution_id} )	undefined
`FILE_READ_MODE`	File input configuration mode	tail
`FILE_START_POSITION`	File input configuration start_position	beginning
`FILE_EXIT_AFTER_READ`	File input configuration exit_after_read	false
`FILE_COMPLETED_ACTION`	File input configuration file_completed_action	log
`MISSED_RESPONSE_CODE`	Default response code when not present in response like on timeout case	510
`PARSE_LABELS_SPLIT_CHAR`	Char to split label into labels	/
`PARSE_TRANSACTION_REGEX`	Regex to identify transaction Label	_.+_
`PARSE_TRANSACTION_AUTO`	Detect transaction controller based on URL null, and message format.	true
`PARSE_FILTER_INCLUDE_SAMPLER_REGEX`	Regex used to include samplers and transactions.
`PARSE_FILTER_EXCLUDE_SAMPLER_REGEX`	Regex used to exclude samplers and transactions.
`PARSE_REMOVE_TRANSACTION`	Remove transaction.	false
`PARSE_REMOVE_SAMPLER`	Remove sampler, not transaction.	false
`PARSE_REMOVE_MESSAGE_FIELD`	Remove field message.	true
`PARSE_CLEANUP_FIELDS`	Remove fields : host, path.	true
`PARSE_WITH_FLAG_SUBRESULT`	Flag result with prefix like have a suffix like "-xx" when xx is the order number	true
`PCT1`	percent 1 value for statistics report	90
`PCT2`	percent 1 value for statistics report	95
`PCT3`	percent 1 value for statistics report	99

3.6. Fields

3.6.1. Common

Common field are integrated during file parsing for both JTL ans statistics file.

Fields	Type ELK	source	Description
`origin`	string	-	Origin of message jtl or stat
`@timestamp`	date	elk	Insertion time in Elastic search
`environment`	string	input/parsing	Target environment (Ex: dev, stage ..), as input using environment variable or extracted from filename.
`executionid`	string	input/parsing	Unique id to identify data for a test, as input using environment variable or extracted from filename.
`filename`	string	parsing	file name without extension.
`path`	string	logstash	Path of file.
`project`	string	input	Project name.
`testname`	string	parsing	Test name, as input using environment variable or extracted from filename.
`testtags`	string	parsing	List of keywords extracted by splitting environnement variable "TEST_TAGS" from environment variable
`timestamp`	date	csv	Request time. Accept timestamp format in ms or "yyyy/MM/dd HH:mm:ss.SSS" .
`label`	string	csv	sampler label
`labels`	string	parsing	List of keywords extracted by splitting label using the char "PARSE_LABELS_SPLIT_CHAR" from environment variable , default to "/"
`globalLabel`	string	parsing	Normalized label for subresults, when there redirection (ex 302) jmeter log all redirects requests and the parent one by default (jmeter.save.saveservice.subresults=false to disable), the parent will have the normal label and other subresut will have a suffix like "-xx" when xx is the order number, in this field you will find the original label for all subresult without the number suffix (see field : subresult, redirectLevel )

3.6.2. JTL

Fields for JTL File. For csv field see documentation on CSV Log format.

For additional fields see documentation on Results file configuration.

Fields	Type ELK	source	Description
`Connect`	long	csv	time to establish connection
`IdleTime`	long	csv	number of milliseconds of 'Idle' time (normally 0)
`Latency`	long	csv	time to first response
`URL`	string	csv
`allThreads`	long	csv	total number of active threads in all groups
`bytes`	long	csv	number of bytes in the sample
`dataType`	string	csv	e.g. text
`domain`	string	parsing	domain name or ip which is extracted from url.
`elapsed`	long	csv	elapsed - in milliseconds
`failureMessage`	string	csv
`grpThreads`	long	csv	number of active threads in this thread group
`host`	string	elk	hostname of logstash node.
`redirectLevel`	long	parsing	redirect number (see field: globalLabel)
`relativetime`	float	parsing	Number of milliseconds from test started. Useful to compare test. this field need to have started test time logged to csv (add this variable name TESTSTART.MS to property sample_variables)
`request`	string	parsing	Request path if Http/s request.
`responseCode`	string	csv
`responseMessage`	string	csv
`responseStatus`	long	parsing	Numeric responseCode , if responseCode is not numeric (case on timeout) using value "MISSED_RESPONSE_CODE" from environment variable , default to 510.
`sentBytes`	long	csv	number of bytes sent for the sample.
`subresult`	boolean	parsing	true if sample is a sub result (see field: globalLabel)
`success`	boolean	csv	true or false.
`teststart`	date	csv	Test start time. This field need to have started test time logged to csv (add this variable name TESTSTART.MS to property sample_variables)
`threadGrpId`	long	parsing	The number of thread group. (Extract from threadName) .
`threadGrpName`	string	parsing	The name of thread group.(Extract from threadName) .
`workerNode`	string	parsing	host port of worker node. (Extract from threadName) .
`threadNumber`	long	parsing	The number of thread (unique in thread group). (Extract from threadName) .
`threadName`	string	csv	Thread name, unique on test.
`transaction`	boolean	parsing	IS Sampler transaction , generated by transaction controller, to identify transaction label should start and and with "_", the regex used to this is "_.+_"
`transactionFailingSampler`	long	parsing	If sample is transaction, this value represent number of failing sampler.
`transactionTotalSampler`	long	parsing	If sample is transaction, this value represent count of total sampler.

3.6.3. Statistics (Only ElasticSearch)

Fields for Statistics

For percentiles configuration (pc1,pc2,pc3) see Percentiles configuration.

Fields	Type ELK	source	Description
`isTotal`	boolean	Parsing	Is total line, label will be "Total"
`sampleCount`	long	Parsing	Sample count
`errorPct`	long	Parsing	Percent of error
`errorCount`	long	Parsing	Error count
`receivedKBytesPerSec`	long	Parsing	Received Kilo Bytes per seconds
`sentKBytesPerSec`	long	Parsing	Sent Kilo Bytes per seconds
`throughput`	long	Parsing	Throughput
`pct1ResTime`	long	Parsing	Percentile 1 response time (aggregate_rpt_pct1)
`pct90`	long	Parsing	percentile 1 value, field name can be changed with PCT1 see configuration
`pct2ResTime`	long	Parsing	Percentile 2 response time (aggregate_rpt_pct2)
`pct95`	long	Parsing	percentile 2 value, field name can be changed with PCT1 see configuration
`pct3ResTime`	long	Parsing	Percentile 3 response time (aggregate_rpt_pct3)
`pct99`	long	Parsing	percentile 3 value, field name can be changed with PCT1 see configuration
`minResTime`	long	Parsing	Minimum response time.
`meanResTime`	long	Parsing	Mean response time.
`medianResTime`	long	Parsing	Median response time.
`maxResTime`	long	Parsing	MAximum response time.

4. Troubleshooting & Limitation

Logstash instance can't parse CSV file with different header Format, as first header will be used for all file, if you have files with different format you should use each time a new instance or restart the instance.
Change sincedb file can't done on logstash with Elasticsearch without building image.
Label with suffix '-{number}' will be considered as subresult, so don't prefix label with '-{number}' or disable subresult flag with PARSE_WITH_FLAG_SUBRESULT.

anasoid / jmeter-logstash