Re: aarch64 - netlink: Error: Could not process rule: No buffer space available

On Wed, Mar 04, 2026 at 11:49:41AM +1100, Steven Haigh wrote:
> On 4/3/26 11:42, Pablo Neira Ayuso wrote:
> > On Wed, Mar 04, 2026 at 11:34:39AM +1100, Steven Haigh wrote:
> > > On 4/3/26 11:17, Pablo Neira Ayuso wrote:
> > > > On Wed, Mar 04, 2026 at 01:02:14AM +0100, Pablo Neira Ayuso wrote:
> > > > > Hi,
> > > > > 
> > > > > On Wed, Mar 04, 2026 at 10:36:20AM +1100, Steven Haigh wrote:
> > > > > > Hi all,
> > > > > > 
> > > > > > Firstly, please CC me in replies as I'm not subscribed to the list.
> > > > > > 
> > > > > > I am currently loading some named sets into nftables using the following
> > > > > > configuration:
> > > > > > 
> > > > > > set au-ipv4 {
> > > > > >           type ipv4_addr
> > > > > >           flags interval
> > > > > >           auto-merge
> > > > > >           elements = { $AU.ipv4 }
> > > > > > }
> > > > > > 
> > > > > > set au-ipv6 {
> > > > > >           type ipv6_addr
> > > > > >           flags interval
> > > > > >           auto-merge
> > > > > >           elements = { $AU.ipv6 }
> > > > > > }
> > > > > > 
> > > > > > These sets are loaded in the config via:
> > > > > > include "/etc/nftables/firewall/geo-nft/countrysets/AU.ipv4";
> > > > > > include "/etc/nftables/firewall/geo-nft/countrysets/AU.ipv6";
> > > > > > 
> > > > > > The files are created using the geo-nft.sh script here:
> > > > > > https://raw.githubusercontent.com/wirefalls/geo-nft/main/geo-nft.sh
> > > > > > 
> > > > > > When loading these, I get the following fatal error:
> > > > > > netlink: Error: Could not process rule: No buffer space available
> > > > 
> > > > Just to be sure and discard something simple.
> > > > 
> > > > Maybe you made a mistake in your ruleset in the aarch64 box? With lots
> > > > of errors coming from the kernel, older userspace nftables versions
> > > > report ENOBUFS.
> > > > 
> > > > Try loading AU.ipv4 and AU.ipv6 with only one element to see if
> > > > userspace reports a different error.
> > > > 
> > > > commit 47e9aaf0227daf16f43a7442e1dceae8851817a5
> > > > Author: Pablo Neira Ayuso <[email protected]>
> > > > Date:   Tue Aug 26 10:09:13 2025 +0200
> > > >       mnl: continue on ENOBUFS errors when processing batch
> > > >       A user reports that:
> > > >         nft -f ruleset.nft
> > > >       fails with:
> > > >         netlink: Error: Could not process rule: No buffer space available
> > > >       This was triggered by:
> > > >       table ip6 fule {
> > > >         set domestic_ip6 {
> > > >           type ipv6_addr
> > > >           flags dynamic,interval
> > > >           elements = $domestic_ip6
> > > >         }
> > > >         chain prerouting {
> > > >           type filter hook prerouting priority 0;
> > > >           ip6 daddr @domestic_ip6 counter
> > > >         }
> > > >       }
> > > >       where $domestic_ip6 contains a large number of IPv6 addresses.
> > > >       This set declaration is not supported currently, because dynamic sets
> > > >       with intervals are not supported, then every IPv6 address that is added
> > > >       triggers an error, overruning the userspace socket buffer with lots of
> > > >       NLMSG_ERROR messages (or too big NLMSG_ERROR message to fit into the
> > > >       socket buffer)
> > > 
> > > --snip--
> > > 
> > > Interesting.
> > > 
> > > I have noticed that if I split the set into multiple 'chunks', then the set
> > > can be populated properly.
> > > 
> > > As an example, this crude claude code authored script here does function as
> > > expected and the entire set is loaded successfully:
> > > 	https://lamp.crc.id.au/paste/e0e9DD01E48E46e27F5ad1bc0e/
> > 
> > Unfortunately, I cannot reach this link.
> 
> Apologies - its behind the firewall that I'm debugging :)
> 
> Mirrored here: https://pastebin.com/iTa9XRCb

This is insane:
while IFS= read -r line; do
 ...
 echo "add element $TABLE_FAMILY $TABLE_NAME $set_name { $batch }" | nft -f -

this is one transaction per command.

This is as bad as a shell script with explicit iptables invocations,
one per line. This is an antipattern.