testsAndMisc/linux_configuration/fixes/lib/arch_sysctl.sh
Krzysztof kuhy Rudnicki 4f2b0ccdf7 Dissolve linux_configuration/scripts/ and single_use/
Two levels of pure taxonomy carrying no information, and a 2-directory
cap leaves no room for them. Categories now sit directly under
linux_configuration/: features, fixes, utils, misc, fresh-install,
installers, diagnostics, gaming, boot_recovery, periodic_background, lib,
meta, tests. A script's own lib/ and tests/ are exempt from the cap, so
those stay put.

Depth violations 248 -> 139; roughly 40 of the remainder are inside
digital_wellbeing and hosts, deferred until the guard-lib repoint.

The moves are the easy half. What breaks is every path that counted
levels, and three classes of it had to be found by resolving paths rather
than reading diffs:

  - 52 relative `source` lines for lib/common.sh and lib/android.sh. The
    correct prefix differs per file: 40 were ../../ and 12 were ../../../,
    so a blanket replace would have fixed one set and broken the other.
  - REPO_ROOT in 7 fixes/lib/tests harnesses and setup_syncyomi_guard.sh,
    each walking up four levels from a directory now two levels shallower.
    This caused a real regression -- the fixes suite went red with
    "log_warn: command not found" -- caught by running it and comparing
    against the pre-move baseline worktree.
  - conftest.py's sys.path entries, which took the pytest gate down with
    14 collection errors, and the `# shellcheck source=` directives, which
    took shellcheck down with SC1091.

16 symlinks in $HOME broke, including the entire i3blocks bar. All
repointed, and cpu_monitor/memory/disk/time/volume were run to confirm
they emit real output rather than merely resolving.

Verified: 1798 tests pass at 100% coverage, 8 of 9 lib/tests suites pass
(the ninth needs the coverage jail, identically at baseline), pre-commit
is clean, and the two untracked secret configs are still ignored after
their gitignore paths changed.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PJrqKERAo5QUeEHWS5Asgp
2026-08-23 18:53:30 +02:00

125 lines
4.0 KiB
Bash
Executable File

#!/bin/bash
# VM and network sysctl tuning.
#
# Sourced by optimize_arch_desktop.sh; split out to keep it under the 250-line
# cap. Sourced rather than run, so it inherits the caller's strict mode
# and the variables defined above the source line.
# ===================================================================
# 3. Memory & swap tuning via sysctl
# ===================================================================
tweak_vm_sysctl() {
# SYSCTL_DROPIN_DIR defaults to the real directory and is overridden only
# by the test harness, which runs un-jailed in ci_mirror.sh and in CI.
local dropin="${SYSCTL_DROPIN_DIR:-/etc/sysctl.d}/90-desktop-performance.conf"
# Desktop workloads: low swappiness, aggressive VFS caching, tuned dirty ratios
local -A params=(
["vm.swappiness"]="10"
["vm.vfs_cache_pressure"]="50"
["vm.dirty_ratio"]="15"
["vm.dirty_background_ratio"]="5"
["vm.dirty_writeback_centisecs"]="1500"
["vm.page-cluster"]="0"
)
local needs_update=false
local key
for key in "${!params[@]}"; do
local current
current=$(sysctl -n "$key" 2>/dev/null || true)
if [[ $current != "${params[$key]}" ]]; then
needs_update=true
break
fi
done
if [[ $needs_update == "false" && -f $dropin ]]; then
log_ok "VM sysctl parameters already tuned — skipping."
return 0
fi
cat >"$dropin" <<'VMEOF'
# Desktop performance tuning — managed by optimize_arch_desktop.sh
#
# vm.swappiness=10 — prefer keeping data in RAM over swapping
# vm.vfs_cache_pressure=50 — favor keeping inode/dentry caches (speeds up file operations)
# vm.dirty_ratio=15 — allow up to 15% RAM dirty before synchronous writeback
# vm.dirty_background_ratio=5 — start async writeback at 5% dirty
# vm.dirty_writeback_centisecs=1500 — flush dirty pages every 15s (less I/O churn)
# vm.page-cluster=0 — read one page at a time from swap (reduces latency on SSD)
vm.swappiness = 10
vm.vfs_cache_pressure = 50
vm.dirty_ratio = 15
vm.dirty_background_ratio = 5
vm.dirty_writeback_centisecs = 1500
vm.page-cluster = 0
VMEOF
sysctl --system >/dev/null 2>&1
return 0
}
# ===================================================================
# 4. Network: TCP BBR + fastopen + buffer tuning
# ===================================================================
tweak_network_sysctl() {
local dropin="${SYSCTL_DROPIN_DIR:-/etc/sysctl.d}/91-desktop-network.conf"
# Check if BBR module is available
if ! modprobe tcp_bbr 2>/dev/null; then
log_warn "tcp_bbr kernel module unavailable — skipping network tuning."
return 0
fi
local -A params=(
["net.core.default_qdisc"]="fq"
["net.ipv4.tcp_congestion_control"]="bbr"
["net.ipv4.tcp_fastopen"]="3"
["net.core.rmem_max"]="16777216"
["net.core.wmem_max"]="16777216"
["net.ipv4.tcp_rmem"]="4096 1048576 16777216"
["net.ipv4.tcp_wmem"]="4096 1048576 16777216"
["net.ipv4.tcp_mtu_probing"]="1"
)
local needs_update=false
local key
for key in "${!params[@]}"; do
local current
current=$(sysctl -n "$key" 2>/dev/null || true)
# Normalize whitespace for comparison (kernel uses tabs)
current=$(echo "$current" | xargs)
local expected
expected=$(echo "${params[$key]}" | xargs)
if [[ $current != "$expected" ]]; then
needs_update=true
break
fi
done
if [[ $needs_update == "false" && -f $dropin ]]; then
log_ok "Network sysctl parameters already tuned — skipping."
return 0
fi
cat >"$dropin" <<'NETEOF'
# Network performance tuning — managed by optimize_arch_desktop.sh
#
# BBR congestion control — better throughput and lower latency than cubic
# TCP fastopen — saves one RTT on repeated connections (both client and server)
# Larger buffers — helps on high-bandwidth or high-latency links
net.core.default_qdisc = fq
net.ipv4.tcp_congestion_control = bbr
net.ipv4.tcp_fastopen = 3
net.core.rmem_max = 16777216
net.core.wmem_max = 16777216
net.ipv4.tcp_rmem = 4096 1048576 16777216
net.ipv4.tcp_wmem = 4096 1048576 16777216
net.ipv4.tcp_mtu_probing = 1
NETEOF
sysctl --system >/dev/null 2>&1
return 0
}