On Wed, Oct 8, 2025 at 2:28 AM Gleb Smirnoff <[email protected]> wrote: > > On Tue, Oct 07, 2025 at 09:44:42PM -0300, Jose Luis Duran wrote: > J> Here is the warning on a fresh VM with 8GB: > J> > J> # sysctl hw.physmem > J> hw.physmem: 8551202816 > J> # service local_unbound onestart > J> Starting local_unbound. > J> [1759883902] local-unbound[3237:0] warning: setsockopt(..., > J> SO_SNDBUF, ...) was not granted: No buffer space available > J> [1759883902] local-unbound[3237:0] warning: so-sndbuf 4194304 was > J> not granted. Got 9216. To fix: start with root permissions(linux) or > J> sysctl bigger net.core.wmem_max(linux) or kern.ipc.maxsockbuf(bsd) > J> values. or set so-sndbuf: 0 (use system value). > > Why would local-unbound want such a big send buffer? Doesn't make sense > to me.
Thank you for the quick fix/workaround. I think it is worth opening a ticket, as it indeed does not make sense, let's say I want to use the port dns/unbound, on a very busy server, and want to bump the so-sndbuf to 8m. It'll be very interesting to set net.inet.udp.maxdgram=8388608 in order to appease the warnings. > -- > Gleb Smirnoff
