From: "DERUMIER, Alexandre" <Alexandre.DERUMIER@groupe-cyllene.com>
To: "w.bumiller@proxmox.com" <w.bumiller@proxmox.com>,
"aderumier@odiso.com" <aderumier@odiso.com>
Cc: "pve-devel@lists.proxmox.com" <pve-devel@lists.proxmox.com>
Subject: Re: [pve-devel] [PATCH V2 pve-firewall] add cluster ebtables_dst_macfilter option.
Date: Fri, 18 Mar 2022 18:59:05 +0000 [thread overview]
Message-ID: <b3e0d6601676114e6dc9ee8ad19c32142e76a280.camel@groupe-cyllene.com> (raw)
In-Reply-To: <20220318132605.mevywhdgakluajz4@wobu-vie.proxmox.com>
Hi Wolfgang,
I think we don't need this patch, now that bridge disabling learning
patch has been applied
https://git.proxmox.com/?p=pve-common.git;a=commit;h=354ec8dee37d481ebae49b488349a8e932dce736
(The problem with hetzner was unicast flood to all ports, and iptables
reject replied with fwbr mac).
(I'll do more tests on nftables next week. I think that I'm currently
able to do nat/routing/ bridging with in/out direction with only 1
extra bridge but I need to do more tests to be sure)
Le vendredi 18 mars 2022 à 14:26 +0100, Wolfgang Bumiller a écrit :
> Sorry for the late reply.
>
> On Fri, Sep 10, 2021 at 05:34:29PM +0200, Alexandre Derumier wrote:
> > This new option allow filtering of destination macs for ingress
> > traffic.
> >
> > This is a protection from bad/hosting networks (like hetzner)
> > flooding
> > traffic with non-hosted mac.
> >
> > To be fast, one rule, this use the "--among-dst mac,mac,mac,mac,"
> > syntax.
> > broadcast mac ff:ff:ff:ff:ff:ff is always allowed
> >
> > currently, ebtables-restore segfault if too many are defined
> > https://antiphishing.cetsi.fr/proxy/v3?i=SGI0YVJGNmxZNE90Z2thMFYLWS
> > xJOfIERJocpmb73Vs&r=SW5LV3JodE9QZkRVZ3JEYaKhfBhKBzRXSL89azwXC1T82d4
> > SHYTQZhKJK2pOWOed&f=bnJjU3hQT3pQSmNQZVE3aPZk7pd95tMIq-
> > 3WY1DAs1r9IrKi7Hir7rLvpxC8B0uY&u=https%3A//www.spinics.net/lists/ne
> > tfilter/msg55995.html&k=dFBm
> >
> > So, I'm using "--among-dst-file", loading macs from an external
> > file.
>
> It's a little awkward but works, I guess, however, it does mess with
> the
> digests we use to verify the ruleset and therefore keeps logging
> errors
> in syslog. This will need fixing.
>
> How many entries can you have before this starts happening anyway?
> And how many entries do you expect there to be?
>
> > Note that "ebtables-save" still show the syntax with "--among-dst
> > mac,mac,mac,"
> > (with a comma at the end), so I compile the full mac list with --
> > among-dst to
> > compare, and if update is needed, I'm writing the dst file in
> > /var/lib/pve-firewall/chain-macfilter, and replace among-dst syntax
> > by among-dst-file
> >
> > Changelog v2:
> > - as we use one rule for performance, add all vms/ct macaddress
> > when vm firewall is enabled.
> > (even if vm macfilter option is disabled).
> >
> > Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
> > ---
> > src/PVE/Firewall.pm | 40 ++++++++++++++++++++++++++++++++++++----
> > 1 file changed, 36 insertions(+), 4 deletions(-)
> >
> > diff --git a/src/PVE/Firewall.pm b/src/PVE/Firewall.pm
> > index edc5336..8277ee0 100644
> > --- a/src/PVE/Firewall.pm
> > +++ b/src/PVE/Firewall.pm
> > @@ -1221,6 +1221,11 @@ our $cluster_option_properties = {
> > default => 1,
> > optional => 1,
> > },
> > + ebtables_dst_macfilter => {
> > + description => "Filtering VM/CT destination mac for ingress
> > traffic.",
> > + type => 'boolean',
> > + optional => 1,
> > + },
> > policy_in => {
> > description => "Input policy.",
> > type => 'string',
> > @@ -2867,7 +2872,7 @@ sub parse_clusterfw_option {
> > if (($value > 1) && ((time() - $value) > 60)) {
> > $value = 0
> > }
> > - } elsif ($line =~ m/^(ebtables):\s*(0|1)\s*$/i) {
> > + } elsif ($line =~
> > m/^(ebtables|ebtables_dst_macfilter):\s*(0|1)\s*$/i) {
> > $opt = lc($1);
> > $value = int($2);
> > } elsif ($line =~
> > m/^(policy_(in|out)):\s*(ACCEPT|DROP|REJECT)\s*$/i) {
> > @@ -3948,11 +3953,19 @@ sub compile_ebtables_filter {
> > ruleset_create_chain($ruleset, "PVEFW-FORWARD");
> >
> > ruleset_create_chain($ruleset, "PVEFW-FWBR-OUT");
> > +
> > + if ($cluster_conf->{options}->{ebtables_dst_macfilter}) {
> > + #filtering destination mac for ipv4/ipv6
> > + ruleset_create_chain($ruleset, "PVEFW-FWBR-IN");
> > + ruleset_addrule($ruleset, 'PVEFW-FORWARD', '-i fwln+', '-j
> > PVEFW-FWBR-IN');
> > + }
> > +
> > #for ipv4 and ipv6, check macaddress in iptables, so we use
> > conntrack 'ESTABLISHED', to speedup rules
> > ruleset_addrule($ruleset, 'PVEFW-FORWARD', '-p IPv4', '-j
> > ACCEPT');
> > ruleset_addrule($ruleset, 'PVEFW-FORWARD', '-p IPv6', '-j
> > ACCEPT');
> > ruleset_addrule($ruleset, 'PVEFW-FORWARD', '-o fwln+', '-j
> > PVEFW-FWBR-OUT');
> >
> > + my $maclist = [];
> > # generate firewall rules for QEMU VMs
> > foreach my $vmid (sort keys %{$vmdata->{qemu}}) {
> > eval {
> > @@ -3975,7 +3988,7 @@ sub compile_ebtables_filter {
> > push(@$arpfilter, $ip);
> > }
> > }
> > - generate_tap_layer2filter($ruleset, $iface,
> > $macaddr, $vmfw_conf, $vmid, $arpfilter);
> > + generate_tap_layer2filter($ruleset, $iface,
> > $macaddr, $vmfw_conf, $vmid, $arpfilter, $maclist);
> > }
> > };
> > warn $@ if $@; # just to be sure - should not happen
> > @@ -4012,17 +4025,23 @@ sub compile_ebtables_filter {
> > push @$arpfilter, $ip;
> > }
> > }
> > - generate_tap_layer2filter($ruleset, $iface,
> > $macaddr, $vmfw_conf, $vmid, $arpfilter);
> > + generate_tap_layer2filter($ruleset, $iface,
> > $macaddr, $vmfw_conf, $vmid, $arpfilter, $maclist);
> > }
> > };
> > warn $@ if $@; # just to be sure - should not happen
> > }
> >
> > + if ($cluster_conf->{options}->{'ebtables_dst_macfilter'} &&
> > @$maclist > 0) {
> > + push @$maclist, 'ff:ff:ff:ff:ff:ff'; #allow broadcast mac
> > + my $maclist_str = join ',',sort(@$maclist);
> > + ruleset_addrule($ruleset, 'PVEFW-FWBR-IN', "--among-dst !
> > $maclist_str,", '-j DROP');
> > + }
> > +
> > return $ruleset;
> > }
> >
> > sub generate_tap_layer2filter {
> > - my ($ruleset, $iface, $macaddr, $vmfw_conf, $vmid, $arpfilter)
> > = @_;
> > + my ($ruleset, $iface, $macaddr, $vmfw_conf, $vmid, $arpfilter,
> > $maclist) = @_;
> > my $options = $vmfw_conf->{options};
> >
> > my $tapchain = $iface."-OUT";
> > @@ -4037,6 +4056,10 @@ sub generate_tap_layer2filter {
> > ruleset_addrule($ruleset, $tapchain, "-s ! $macaddr",
> > '-j DROP');
> > }
> >
> > + if (defined($macaddr)) {
> > + push @$maclist, $macaddr;
> > + }
> > +
> > if (@$arpfilter){
> > my $arpchain = $tapchain."-ARP";
> > ruleset_addrule($ruleset, $tapchain, "-p ARP", "-j
> > $arpchain");
> > @@ -4225,6 +4248,15 @@ sub get_ebtables_cmdlist {
> > next if ! $pve_include;
> > $pve_include = 0;
> > }
> > +
> > + if ($cmd =~ m/^-A (\S+) --among-dst ! (\S+) -j DROP/) {
> > + my $chain = $1;
> > + my $maclist_raw = $2."\n";
> > + my $filename =
> > "$pve_fw_status_dir/ebtables_macfilter-$chain";
> > + PVE::Tools::file_set_contents($filename,
> > $maclist_raw);
> > + $cmd = "-A $1 --among-dst-file ! $filename -j
> > DROP";
> > + }
> > +
> > $cmdlist .= "$cmd\n";
> > }
> > }
> > --
> > 2.30.2
>
prev parent reply other threads:[~2022-03-18 18:59 UTC|newest]
Thread overview: 3+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-09-10 15:34 Alexandre Derumier
2022-03-18 13:26 ` Wolfgang Bumiller
2022-03-18 18:59 ` DERUMIER, Alexandre [this message]
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=b3e0d6601676114e6dc9ee8ad19c32142e76a280.camel@groupe-cyllene.com \
--to=alexandre.derumier@groupe-cyllene.com \
--cc=aderumier@odiso.com \
--cc=pve-devel@lists.proxmox.com \
--cc=w.bumiller@proxmox.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.