Jump to content

Sysctl: Difference between revisions

From SameTeem
No edit summary
No edit summary
Line 23: Line 23:
Allow more connections to be queued on the system, good for programs that establish a lot of connections, such as P2P applications.
Allow more connections to be queued on the system, good for programs that establish a lot of connections, such as P2P applications.
  net.core.somaxconn=8192
  net.core.somaxconn=8192
== Misc ==
Set a timer so the system will actually reboot once a kernel panic (BSOD/Bug Check) happens, this can be set as low as you like if you want to just reboot instantly, but 60 seconds allows the kernel watchdog to possibly recover the system if it's just something using all the CPU time.
kernel.panic=60

Revision as of 18:22, 6 August 2026

All files changes go in /etc/sysctl.conf

sudo nano /etc/sysctl.conf

A note on EAC

If you search for tips for sysctl tweaks/security improvements you will often find something a recommendation of

kernel.yama.ptrace_scope=2 

This will break EAC and stop any EAC game from starting on Linux, the highest you can set it is 1.

Memory Tweaks

Keep more data in RAM instead of swapping to disk, make the system more responsive, especially if your swap (paging file in Windows terminology) file/partition is on a hard disk drive.

vm.swappiness=1

Network

Enable BBR (https://en.wikipedia.org/wiki/TCP_congestion_control#TCP_BBR) Greater top speeds and more reliable, especially for higher latency connections like 3/4/5G and satellite.

net.core.default_qdisc = cake 
net.ipv4.tcp_congestion_control = bbr

Allow more packets to be queued up before the system drops them, pretty much required for Gigabit+ connections

net.core.netdev_max_backlog=16384

Allow more connections to be queued on the system, good for programs that establish a lot of connections, such as P2P applications.

net.core.somaxconn=8192

Misc

Set a timer so the system will actually reboot once a kernel panic (BSOD/Bug Check) happens, this can be set as low as you like if you want to just reboot instantly, but 60 seconds allows the kernel watchdog to possibly recover the system if it's just something using all the CPU time.

kernel.panic=60