On 2025/08/28 14:27, Otto Moerbeek wrote:
>  WANTLIB+=		boost_context-mt boost_filesystem-mt boost_system-mt crypto
> -WANTLIB+=		curl fstrm m nghttp2 nghttp3 ngtcp2 ngtcp2_crypto_libressl
> +WANTLIB+=		curl fstrm m nghttp2 nghttp3 ngtcp2

looks like there are some others to remove? from port-lib-depends-check

Extra:  boost_atomic-mt.24 boost_system-mt.24 lua5.3.5 nghttp2.0
Extra:  nghttp3.3 ngtcp2.5 z.7

> -			--with-lua=${MODLUA_DEP} \
> +			-Dlua=auto \

Missing lib: luajit-5.1.1 (/usr/local/sbin/pdns_recursor) (NOT REACHABLE)

this has the same problem as dnsdist that we fixed with

.if ${PROPERTIES:Mluajit}
CONFIGURE_ARGS+=        -Dlua=luajit
LIB_DEPENDS+=           lang/luajit
WANTLIB+=               luajit-5.1
.else
CONFIGURE_ARGS+=        -Dlua=lua
MODULES+=               lang/lua
MODLUA_VERSION=         5.3
LIB_DEPENDS+=           ${MODLUA_LIB_DEPENDS}
WANTLIB+=               ${MODLUA_WANTLIB}
.endif

(and as with dnsdist, it'll need changing if 5.4 support is later
added to meson/lua/meson.build and picked up in preference to others)

> +BUILD_DEPENDS=          devel/boost

this is already in LIB_DEPENDS, no need to go in BUILD_DEPENDS as well

> +MODCARGO_CRATES +=	ring	0.17.13	# Apache-2.0 AND ISC

please add this near the top of the Makefile:

# ring-v0.17 does not support this arch
NOT_FOR_ARCHS =         sparc64

> Index: pkg/pdns_recursor.login
> ===================================================================
> RCS file: pkg/pdns_recursor.login
> diff -N pkg/pdns_recursor.login
> --- /dev/null	1 Jan 1970 00:00:00 -0000
> +++ pkg/pdns_recursor.login	28 Aug 2025 12:21:55 -0000
> @@ -0,0 +1,4 @@
> +pdns_recursor:\
> +	:openfiles-cur=8096:\
> +	:openfiles-max=8096:\
> +	:tc=daemon:

that's higher than the default kern.maxfiles (7030 on maxusers=80 archs
which are the only ones relevant to this; the smaller ones don't have
rust anyway); I think I'd drop by a bit (to 4k/6k maybe?) if possible,
otherwise mention the sysctl in pkg/README.

better if a single daemon is restricted by login.conf from eating so
many FDs that it's not possible to ssh in to fix.