Monitor Your PHP Applications with Logs, Metrics, Pings, and Traces

on YouTube ↗44m

" summary

Live demo of an Elastic stack that collects logs, metrics, pings and traces from a SilverStripe app on cheap AWS instances: filebeat, metricbeat, packetbeat, heartbeat send data to Kibana (shown as “Cabana”), producing ~60 dashboards and ~700,000 log events. Memory spikes hit ~1.5 GB, one host shows a Java cron causing the spikes, PHP‑FPM peaks at ~23 active processes and returns 502s, packetbeat shows TLS errors and some requests up to 18s, and structured JSON logs let queries and top slow MySQL statements be found.

AI-written summary · about

" chapters (18)

1. 0:00Start
2. 7:00Monitoring
3. 8:17Dashboards
4. 8:32System Overview
5. 9:43Visual Builder
6. 11:10Term Segregation
7. 16:12What Could I Monitor
8. 17:47Latency Histogram
9. 18:25Web Transactions
10. 20:57Configuration
11. 24:28Error Logs
12. 24:56Raw Data
13. 31:54What Is Nginx Up to
14. 37:24Heartbeat
15. 38:04Response Times
16. 41:58Luckily We Didn't Have any Mysql Errors and down Here You Can Also See Which Were the Most Common Queries That We'Ve Been Answering so that Would Then Tell You like Which Queries Are You Running and Which Ones Are Probably Slow or Might Need More Optimization so this Is Was It the Group with Him You Can See What Your System Is up to Also the Latency of How Long Do Your Queries Actually Take Okay if You Want To Play Around with the Dashboards Yourself There Is a Dashboard View You Will Be Logged In as a Dashboard on the User You CanNot Create Custom Visualizations
17. 42:32You CanNot Delete any Data or Anything like that but You Can Play Around with the Pre-Built Dashboards and See How They Are Interacting with What You'Re Doing So if You Start another a Denial of Service Attack You Will See What Is Happening and I Will Let that Run for the Rest of the Day so You Can Go Crazy and It's Not Going To Do Anything if You Want To Take a Look at the Code It's at that Github Url and There You Can See How Did I Set Up Everything
18. 42:46So You Can Go Crazy and It's Not Going To Do Anything if You Want To Take a Look at the Code It's at that Github Url and There You Can See How Did I Set Up Everything and Just Copy What Makes Sense Questions I Think We Have no Time for Questions Basically Grab Seekers if You Have Questions Come to Me Afterwards if There Is Anything on Slide or I Will Answer As Well Thank You Oh I'M Always Taking My Picture so I Can Prove that We'Ve Been Doing Something I'Ll Try To Get You all in One Picture

" related talks

• 35mFull Stack Visibility with Elastic: Logs, Metrics and TracesCarlos Pérez-Aradros · KubeCon + CloudNativeCon North America 2017
• 44mThe Holy Trinity of Observability: Logs, Metrics, and TracesPhilipp Krenn · Oredev 2017
• 35mHands-On ModSecurity and LoggingPhilipp Krenn · code.talks commerce 2019
• 39mLinux Server Deep DivesAmin Astaneh · DrupalCon Amsterdam 2019
• 40mA needle in a logstackTomasz Gągor, Paweł Torbus · JDD 2016
• 49mMonitor Your Microservices with Logs, Metrics, Pings and TracesDavid Pilato · Voxxed Days Microservices 2018
• 31m360° Monitoring of Your ServicesPhilipp Krenn · CloudNativeCon + KubeCon Europe 2017
• 01h 04mProfiling PHP ApplicationsBastian Hofmann · PHP UK Conference 2014