personal_interest_mirrors/nebula
1
0
Fork 0
mirror of https://github.com/slackhq/nebula.git synced 2026-08-15 07:26:58 +02:00
Code Issues Actions Packages Projects Releases Wiki Activity

tun: default pin_threads off

Browse Source 
Thread pinning trades scheduler freedom for TX-ring ordering; that's the
right trade on dedicated forwarders but not as a surprise default on
hosts sharing cores with other workloads. Make it opt-in and document
the default in the example config.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
This commit is contained in:
JackDoan
2026-07-27 15:39:14 -05:00
parent 6c0305c7ae
commit 56d2a3841d
3 changed files with 6 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files
examples/config.yml
+3 -2
View File
@@ -264,8 +264,9 @@ tun:
# Linux only. pin_threads pins each tun reader/encrypt OS thread to a single CPU. This keeps every goroutine's # Linux only. pin_threads pins each tun reader/encrypt OS thread to a single CPU. This keeps every goroutine's
# batched sends flowing through one XPS-selected NIC TX ring, so packets within a flow stay ordered on the wire # batched sends flowing through one XPS-selected NIC TX ring, so packets within a flow stay ordered on the wire
# instead of being sprayed across multiple TX rings and reordered. Not reloadable. # instead of being sprayed across multiple TX rings and reordered. Off by default; pinning takes CPUs away from
#pin_threads: true # the scheduler, which can hurt hosts that share cores with other workloads. Not reloadable.
#pin_threads: false
# Linux only. cpu_affinity overrides which CPUs the tun reader threads pin to: a list of CPU IDs, one per routine # Linux only. cpu_affinity overrides which CPUs the tun reader threads pin to: a list of CPU IDs, one per routine
# (see the top-level `routines` setting). Lists shorter than `routines` are modulo-cycled across the queues; extra # (see the top-level `routines` setting). Lists shorter than `routines` are modulo-cycled across the queues; extra
interface.go
+2 -2
View File
@@ -59,7 +59,7 @@ type InterfaceConfig struct {
// pin-to-(i % NumCPU) behavior. Only consulted when PinThreads is true. // pin-to-(i % NumCPU) behavior. Only consulted when PinThreads is true.
CpuAffinity []int CpuAffinity []int
// PinThreads controls whether each TUN reader OS thread is pinned to a // PinThreads controls whether each TUN reader OS thread is pinned to a
// single CPU (via tun.pin_threads, default true). Pinning keeps each // single CPU (via tun.pin_threads, default false). Pinning keeps each
// goroutine's sendmmsg on one XPS-selected NIC TX ring so per-flow // goroutine's sendmmsg on one XPS-selected NIC TX ring so per-flow
// packets stay ordered on the wire. // packets stay ordered on the wire.
PinThreads bool PinThreads bool
@@ -94,7 +94,7 @@ type Interface struct {
// Only consulted when pinThreads is true. // Only consulted when pinThreads is true.
cpuAffinity []int cpuAffinity []int
// pinThreads controls whether listenIn pins each TUN reader OS thread to // pinThreads controls whether listenIn pins each TUN reader OS thread to
// a CPU at all (tun.pin_threads, default true). When false, threads are // a CPU at all (tun.pin_threads, default false). When false, threads are
// left free to migrate as on stock nebula. // left free to migrate as on stock nebula.
pinThreads bool pinThreads bool
// ecnEnabled gates RFC 6040 underlay ECN propagation. When true, // ecnEnabled gates RFC 6040 underlay ECN propagation. When true,
main.go
+1 -1
View File
@@ -236,7 +236,7 @@ func Main(c *config.C, configTest bool, buildVersion string, l *slog.Logger, dev
punchy: punchy, punchy: punchy,
ConntrackCacheTimeout: conntrackCacheTimeout, ConntrackCacheTimeout: conntrackCacheTimeout,
CpuAffinity: parseCpuAffinity(c, l, routines), CpuAffinity: parseCpuAffinity(c, l, routines),
PinThreads: c.GetBool("tun.pin_threads", true), PinThreads: c.GetBool("tun.pin_threads", false),
l: l, l: l,
} }