Kakao Reveals Solution for Traffic Inflow Issues Caused by L4 Device Session Table Retention During Deployment
Key point
To resolve traffic inflow issues during deployment caused by L4 device session table retention, a procedure was presented that involves waiting for HealthCheck failure before stopping the web server process to clear sessions.
Details
Kakao has published a deployment procedure to resolve the issue where requests from existing Keep-Alive clients continue to flow in even after a service is excluded, due to L4 devices in an L3DSR configuration maintaining the session table for the Idle Session Timeout (default 600 seconds). The 'Reset On Port Fail' option on L4 devices can immediately clear sessions when a port goes down; however, Kakao does not apply this by default due to the risk of fault propagation caused by resource load on the L4 device during mass failures. Instead, for VIPs using the L7 HealthCheck method, the health_check.html file is deleted to induce a 404 response, and after waiting up to 20 seconds (11–20 seconds) for HealthCheck failure detection, the web server process is stopped to clear server-side sessions (sending FIN packets) before proceeding with deployment. In the L4 HealthCheck method, stopping the web server process alone causes the port to go down, triggering HealthCheck failure detection and removal from the binding list. Through this process, the session table can be cleared and traffic inflow prevented without applying the option.
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.