Problem Statement

In our organization, Error detection and troubleshooting are currently manual and time-consuming processes due to the lack of a unified view of application logs and metrics. This leads to delayed incident response and increased downtime.

Objective: To implement a comprehensive observability solution that provides a unified view of application logs and metrics, thereby automating error detection and troubleshooting processes. This initiative aims to significantly reduce the time spent on manual monitoring, enhance incident response times, and minimize system downtime by leveraging real-time data visualization and advanced analytics.

We adopted ELK stack to achieve above objective.

Setting up ELK on Ubuntu:

  1. Install and configure ELK on UBUNTU as single node cluster.

  2. Check Elasticsearch:

    • Test Elasticsearch by accessing its URL in a browser:  curl -X GET "localhost:9200"
    • If Elasticsearch is running correctly, you should see a JSON response with Elasticsearch cluster details.

    Check Kibana:

  3. Copy the apm_agent folder to the server (preferably in the wildfly installation location)

  4. apm_agent folder contains two files elastic-apm-agent.jar, elasticapm.properties 

    Add below lines in elasticapm.proerties

    server_url=http://<private-ip>:8200

    enable_log_correlation=true

    environment=hyd-sandbox

  5. Filebeat Setup on Application Nodes
    • Filebeat is used to forward logs from our application nodes to Logstash or Elasticsearch.
    • Download and install 
    • Download http_ca.crt to the server (preferably in the filebeat installation location)
    • Create a Logs folder inside Wildfly 

    • Do the changes in Filebeat.yml as mentioned here 

Switch to Wildfly/bin/ and edit standalone.conf.bat add below line under wildfly folder

set "JAVA_OPTS=%JAVA_OPTS% -javaagent:E:\\AppServer\\wildfly-30.0.0.Final\\wildfly-30.0.0.Final\\apm_agent\\elastic-apm-agent-1.52.0.jar"

restart wildfly 

check if traces are visible in APM section of kibana  

restart filebeat

check for logs under filebeat in kibana 

Do the changes in github repo under ci-properties & application properties for each service line.

Example:

Goto common-ci.properties

ADD a new line

logging.file.name=@env.HWC_API_LOGGING_FILE_NAME@ 

Goto application.properties 

Add below two lines:

logging. Path=logs/ 

logging.file.name=logs/hwc-api.log

4.1. Define Environment Variables in Jenkins Pipeline:

In the Jenkins pipeline, set the environment variable

env.HWC_API_LOGGING_FILE_NAME='E:/AppServer/wildfly-30.0.0.Final/wildfly-30.0.0.Final/Logs/hwc-api.log'             

5.4. Check Application Logs:

Check if the logs are being generated at the specified location (e.g., logs/inventory-api.log) and whether they are being processed by Filebeat and appear in Kibana.

Check Kibana:

  • Navigate to the "Discover" section and check if logs are being indexed in Elasticsearch.

Troubleshooting

  • Filebeat not sending logs: Check the Filebeat logs at /var/log/filebeat/ for errors.
  • Elasticsearch or Kibana not working: Review the logs in /var/log/elasticsearch/ and /var/log/kibana/.
  • Application not writing logs: Ensure that the application has access to the specified log path and that the environment variable is being correctly passed from Jenkins.

 


Conclusion

This document outlines the process of configuring ELK for log collection from our application. we have set up Elasticsearch, Logstash, and Kibana on Ubuntu, configured Filebeat for log forwarding, and adjusted our application’s properties file to use environment variables for log paths. The Jenkins pipeline is configured to pass the necessary environment variables to the application, ensuring that the logs are written to the correct location.


  • No labels