mirror of
https://github.com/kiddin9/op-packages.git
synced 2026-11-15 21:13:00 +08:00
This commit is contained in:
1 parent
d1e51134a0
commit
afd9decbc5
26 files changed
+269
-97
No files matched your search
@@ -1,7 +1,26 @@
|
||||
config trafficctl 'main'
|
||||
option enabled '1'
|
||||
option default_mode 'limiter'
|
||||
option persist_rules '0'
|
||||
# Write per-device blocks and rate limits to /etc/trafficctl/rules.json so
|
||||
# the ifup-lan hook can restore them after a reboot.
|
||||
#
|
||||
# On by default since #105. It used to ship off, which made the same
|
||||
# intention behave two ways: a traffic SHAPE persists unconditionally (the
|
||||
# shaper has to record which HTB class belongs to which device, so
|
||||
# shapes.json is always written and always restored), while a block or a
|
||||
# limit vanished on reboot — and the default mode is the limiter, so the
|
||||
# out-of-the-box path was the one that forgot. That was an unexamined
|
||||
# default sitting next to a deliberate one, not a design.
|
||||
#
|
||||
# The cost is one small write to /etc per rule CHANGE. Not periodic, so
|
||||
# unlike the byte counters — which live in tmpfs precisely because they are
|
||||
# rewritten every poll — this is not a flash-wear concern.
|
||||
#
|
||||
# NOT the same switch as the global cut's "keep after reboot": that one is
|
||||
# its own opt-in because an all-devices cut can lock out whoever
|
||||
# administers the router through a LAN host, so a reboot must be able to
|
||||
# undo it. See the cut section below.
|
||||
option persist_rules '1'
|
||||
# Extra subnets to monitor (space-separated CIDRs), for downstream-router
|
||||
# clients that have no kernel route on this router. Subnets with a
|
||||
# "via" route over a LAN interface are picked up automatically.
|
||||
|
||||
Reference in new issue
Block a user