-
Notifications
You must be signed in to change notification settings - Fork 100
Description
When running bpftune with IPv6 disabled at kernel level (ipv6.disable=1 boot parameter), the neigh_table_tuner (and potentially other tuners) generate errors on startup because they try to access /proc/sys/net/ipv6/ paths that do not exist:
bpftune: could not open /proc/sys/net/ipv6/neigh/default/gc_interval (netns fd 0) for reading: No such file or directory
bpftune: could not open /proc/sys/net/ipv6/neigh/default/gc_stale_time (netfs fd 0) for reading: No such file or directory
bpftune: could not open /proc/sys/net/ipv6/neigh/default/gc_thresh1 (netns fd 0) for reading: No such file or directory
bpftune: could not open /proc/sys/net/ipv6/neigh/default/gc_thresh2 (netns fd 0) for reading: No such file or directory
This is a known situation also reported in #90.
Environment
- OS: Arch Linux
- Kernel: 6.18.13-xanmod1-1-x64v3
- CPU: AMD Ryzen Threadripper 3970X
- Boot parameter:
ipv6.disable=1
Expected behavior
Bpftune should handle gracefully the case where IPv6 is disabled at kernel level, either by:
-
Option A (preferred): Detecting at startup that IPv6 is disabled (e.g. checking
/proc/sys/net/ipv6availability orCONFIG_IPV6state) and skipping IPv6-related tunables silently, without logging errors. -
Option B: Adding a CLI flag such as
-D <tuner_name>(or--disable <tuner_name>) to allow the administrator to explicitly exclude specific tuners. This would also be useful for other scenarios where a tuner is not applicable to a given system configuration.