Linux Sysctl & Ulimit Tuner Reference

Interactive kernel tunables reference for sysctl.conf and limits.conf covering network performance and memory — entirely on this device.

0 bytes uploaded

Target Workload Profile

Expands socket backlogs, ephemeral port ranges, and file descriptors for massive concurrent connections.

Curated Kernel Tunables

Parameter	Recommended	Kernel Default	Why tune this?
`net.core.somaxconn`	65535	4096	Maximum socket listen backlog queue size. 💡 Prevents connection drop spikes under heavy burst traffic.
`net.ipv4.tcp_max_syn_backlog`	65535	1024	Maximum half-open SYN connection backlog. 💡 Ensures TCP three-way handshakes complete without dropping incoming requests.
`net.ipv4.ip_local_port_range`	1024 65535	32768 60999	Range of local ephemeral outbound ports. 💡 Expands available ports for reverse-proxy upstream connections to avoid TIME_WAIT exhaustion.
`net.ipv4.tcp_tw_reuse`	1	2	Allow reusing TIME_WAIT sockets for outgoing connections. 💡 Safe protocol-compliant reuse of closed TCP connections for outgoing proxy traffic.
`net.ipv4.tcp_fin_timeout`	15	60	Seconds to hold socket in FIN-WAIT-2 state. 💡 Reclaims closed socket memory four times faster.
`fs.file-max`	2097152	1048576	System-wide maximum open file handles. 💡 Supports hundreds of thousands of concurrent client connections.

Generated /etc/sysctl.d/99-custom.conf

# /etc/sysctl.d/99-custom-tuning.conf # Generated by Developer Utility SaaS - https://devxcode.com/sysctl-parameters # --- NETWORK TUNABLES --- # Maximum socket listen backlog queue size. (default: 4096) net.core.somaxconn = 65535 # Maximum half-open SYN connection backlog. (default: 1024) net.ipv4.tcp_max_syn_backlog = 65535 # Range of local ephemeral outbound ports. (default: 32768 60999) net.ipv4.ip_local_port_range = 1024 65535 # Allow reusing TIME_WAIT sockets for outgoing connections. (default: 2) net.ipv4.tcp_tw_reuse = 1 # Seconds to hold socket in FIN-WAIT-2 state. (default: 60) net.ipv4.tcp_fin_timeout = 15 # --- FILESYSTEM TUNABLES --- # System-wide maximum open file handles. (default: 1048576) fs.file-max = 2097152

Matching /etc/security/limits.d/99-custom.conf

# /etc/security/limits.d/99-custom-limits.conf # <domain> <type> <item> <value> * soft nofile 65536 # Per-process open file limit * hard nofile 1048576 # Max ceiling open file limit

Apply Tunables Immediately

Reload and apply kernel settings without rebooting the server:

sudo sysctl --system

Do it offline:$

How to use it

Select your target infrastructure profile (High-Concurrency Web / Reverse Proxy, Database & In-Memory Store, or Kubernetes Worker Host).
Inspect the curated parameters, comparing kernel defaults with production-tested recommended values.
Review technical rationales explaining how each tunable prevents connection drops or memory stalls under load.
Copy the generated /etc/sysctl.d/99-custom.conf and matching /etc/security/limits.d/ configuration files.
Apply changes live without a reboot using sudo sysctl --system.

Interactive Linux kernel tuning and resource limit generator for sysctl.conf and limits.conf. Optimize network socket queues, virtual memory swappiness, open file descriptor ceilings, and container node parameters.

How it works

The Linux kernel exposes hundreds of runtime configuration toggles under /proc/sys/ that can be modified via the sysctl interface.

Under high concurrency, default socket listen queues (net.core.somaxconn) and SYN backlogs (tcp_max_syn_backlog) are quickly saturated, causing silent client connection dropouts.

Adjusting vm.swappiness down to 10 prevents the kernel from aggressively swapping active database memory pages to slow disks while still providing swap space during emergency memory pressure.

Persistent changes should be placed in files under /etc/sysctl.d/*.conf, which systemd-sysctl parses and loads automatically at system boot time.

Examples

Every example on this page is executed against the same engine that powers the tool, on every build.

Gotchas worth knowing

Common errors

FAQ

How do I apply sysctl changes without rebooting?

Run 'sudo sysctl --system' on systemd-based Linux distributions (Ubuntu, Debian, RHEL, Rocky, Fedora). This command reloads configuration files across /etc/sysctl.d/, /run/sysctl.d/, and /usr/lib/sysctl.d/ in priority order.

Why does Redis recommend setting vm.overcommit_memory=1?

Redis relies on the fork() system call to create background child processes for RDB snapshots and AOF rewriting. By default, Linux heuristics (mode 0) may reject fork() allocations if total virtual memory exceeds physical RAM, causing snapshot failures. Setting overcommit_memory=1 tells the kernel to always grant memory allocations optimistically.

What is the difference between sysctl file-max and limits.conf nofile?

fs.file-max in sysctl controls the maximum total file handles the entire Linux operating system can allocate across all processes combined. In contrast, 'nofile' in /etc/security/limits.conf sets the per-process limit for a specific user or session.

Nothing you paste here leaves this tab. State only ever travels in links you explicitly create.

Last reviewed: 2026-10-09