Performance Tuning & Kernel Parameters (Sysctl)

Stop your high-traffic servers from dropping packets on the floor

🧩 The Challenge

Dealing with a massive influx of incoming connections on a busy load balancer, only to see the kernel silently drop them before they even hit the application. It makes you feel like you’re losing your mind when the app metrics look fine but users are getting connection resets.

💡 The Fix

Crank up the backlog limits for the network stack so the kernel stops panicking when a sudden burst of SYN packets hits your interface. It gives your application just enough breathing room to process the queue without cutting off the front door.

sysctl -w net.core.netdev_max_backlog=5000
sysctl -w net.ipv4.tcp_max_syn_backlog=2048

⚙️ Why It Works

These parameters define how many packets can be queued for processing before the NIC or the TCP stack starts tossing them into the trash bin. Bumping these values prevents that packet loss during those annoying, brief traffic spikes.

🚀 Pro-Tip: Always verify your changes persist across reboots by adding them to a file in /etc/sysctl.d/ so you aren’t chasing the same bug at 3 AM after a kernel update.

Linux Tips & Tricks | © ngelinux.com | 7/19/2026