public inbox for pve-devel@lists.proxmox.com
 help / color / mirror / Atom feed
From: Wolfgang Bumiller <w.bumiller@proxmox.com>
To: Alexandre Derumier <aderumier@odiso.com>
Cc: pve-devel@lists.proxmox.com
Subject: Re: [pve-devel] [PATCH pve-container 1/1] add ipam support
Date: Wed, 12 May 2021 14:27:02 +0200	[thread overview]
Message-ID: <20210512122702.3relfzag2yrqv2lo@wobu-vie.proxmox.com> (raw)
In-Reply-To: <20210510153709.2271214-2-aderumier@odiso.com>

On Mon, May 10, 2021 at 05:37:07PM +0200, Alexandre Derumier wrote:
> This add ipam support for nic using sdn vnets.
> 
> - if ips are specified manally, we verify that subnet exist on vnet, and we register ip in ipam
> - if nic is on a vnet, but no ip is specified, we auto find the next available ips in this vnet subnet
> - if a gateway is defined on the subnet, we override current vm nc gateway
> - extra informations like mac address,hostname are registered in external ipam.
> - if dns server exist in the zone, we register vm hostname in ipam for each ip address
> 
> - ips addresses are removed on ct destroy, nic destroy, ip change or vnet change
> 
> Signed-off-by: Alexandre Derumier <aderumier@odiso.com>
> ---
>  src/PVE/LXC.pm        |   3 +
>  src/PVE/LXC/Config.pm | 168 ++++++++++++++++++++++++++++++++++++++++++
>  2 files changed, 171 insertions(+)
> 
> diff --git a/src/PVE/LXC.pm b/src/PVE/LXC.pm
> index 7e6f378..f6bd806 100644
> --- a/src/PVE/LXC.pm
> +++ b/src/PVE/LXC.pm
> @@ -847,6 +847,9 @@ sub destroy_lxc_container {
>      rmdir "/var/lib/lxc/$vmid/rootfs";
>      unlink "/var/lib/lxc/$vmid/config";
>      rmdir "/var/lib/lxc/$vmid";
> +
> +    PVE::LXC::Config->delete_ipam_ifaces($conf);
> +
>      if (defined $replacement_conf) {
>  	PVE::LXC::Config->write_config($vmid, $replacement_conf);
>      } else {
> diff --git a/src/PVE/LXC/Config.pm b/src/PVE/LXC/Config.pm
> index 7b82f65..c0be257 100644
> --- a/src/PVE/LXC/Config.pm
> +++ b/src/PVE/LXC/Config.pm
> @@ -12,6 +12,14 @@ use PVE::INotify;
>  use PVE::JSONSchema qw(get_standard_option);
>  use PVE::Tools;
>  
> +my $have_sdn;
> +eval {
> +    require PVE::Network::SDN::Vnets;
> +    require PVE::Network::SDN::Subnets;
> +    $have_sdn = 1;
> +};
> +
> +
>  use base qw(PVE::AbstractConfig);
>  
>  use constant {FIFREEZE => 0xc0045877,
> @@ -1044,6 +1052,10 @@ sub update_pct_config {
>  	    $class->check_protection($conf, "can't remove CT $vmid drive '$opt'");
>  	} elsif ($opt eq 'unprivileged') {
>  	    die "unable to delete read-only option: '$opt'\n";
> +	} elsif ($opt =~ m/^net(\d+)$/) {
> +	    my $netid = $1;
> +	    my $oldnet = $class->parse_lxc_network($conf->{$opt});
> +	    delete_net_ip($conf->{hostname}, $oldnet, "vm:$vmid net:$netid");
>  	}
>  	$class->add_to_pending_delete($conf, $opt);
>      }
> @@ -1071,7 +1083,24 @@ sub update_pct_config {
>  	    $value = PVE::LXC::verify_searchdomain_list($value);
>  	} elsif ($opt eq 'unprivileged') {
>  	    die "unable to modify read-only option: '$opt'\n";
> +	} elsif ($opt =~ m/^net(\d+)$/) {
> +		my $netid = $1;
> +		my $net = $class->parse_lxc_network($value);
> +		my $oldnet = $class->parse_lxc_network($conf->{$opt});
> +		my $hostname = $param->{hostname} ? $param->{hostname} : $conf->{hostname};
> +		update_net_ip($net, $oldnet, $hostname, $hostname, "vm:$vmid net:$netid");
> +		$value = $class->print_lxc_network($net);
> +	} elsif ($opt eq 'hostname') {
> +	    #if hostname change, update ipam + dns for each ip
> +	    foreach my $netopt (sort keys %$conf) {
> +	        next if $netopt !~ m/^net(\d+)$/;
> +		my $netid = $1;
> +		my $net = $class->parse_lxc_network($conf->{$netopt});
> +		my $oldnet = $class->parse_lxc_network($conf->{$netopt});

^ you can juse use $net twice, no need to parse the same value twice

> +		update_net_ip($net, $oldnet, $value, $conf->{hostname}, "vm:$vmid net:$netid");
> +	    }
>  	}
> +
>  	$conf->{pending}->{$opt} = $value;
>  	$class->remove_from_pending_delete($conf, $opt);
>      }
> @@ -1663,4 +1692,143 @@ sub get_backup_volumes {
>      return $return_volumes;
>  }
>  
> +sub delete_ipam_ifaces {
> +    my ($class, $conf) = @_;
> +
> +    return if !$have_sdn;
> +
> +    foreach my $opt (sort keys %$conf) {

^ do we need this to be sorted?

> +	next if $opt !~ m/^net(\d+)$/;
> +	my $netid = $1;
> +	my $net = $class->parse_lxc_network($conf->{$opt});
> +	my $bridge = $net->{bridge};
> +	my $hostname = $conf->{hostname};
> +	my $description = '';
> +
> +	my $subnets = PVE::Network::SDN::Vnets::get_subnets($bridge);
> +	next if !keys %{$subnets};
> +
> +	eval {
> +	    PVE::Network::SDN::Vnets::del_cidr($bridge, $net->{ip}, $hostname, $description) if $net->{ip} && $net->{ip} ne 'dhcp' && $net->{ip} ne 'manual';
> +	    PVE::Network::SDN::Vnets::del_cidr($bridge, $net->{ip6}, $hostname, $description) if $net->{ip6} && $net->{ip6} ne 'dhcp' && $net->{ip6} ne 'manual' && $net->{ip6} ne 'auto';

^ Can we please factor out the above conditions in the post-fix if
statements into a separate helper (ie. 'is_static_ip' or something?),
because you use the same condition multiple times, but there's a case
further down below where it's inconsistent and probably wrong...
(btw. a helper wouldn't necessarily need to bother distinguishing
between ipv4 and ipv6 since in ipv4 'auto' simply never happens)

> +	};
> +	if ($@) {
> +	    warn $@;
> +	}
> +    }
> +}
> +
> +sub update_net_ip {
> +    my ($net, $oldnet, $hostname, $oldhostname, $description) = @_;
> +
> +    return if !$have_sdn;
> +
> +    my $oldbridge = $oldnet->{bridge};
> +    my $bridge = $net->{bridge};
> +    my $mac = $net->{hwaddr};
> +
> +    my $subnets = PVE::Network::SDN::Vnets::get_subnets($bridge);
> +
> +    return if !keys %{$subnets};

I have a feeling the code below might get a little more readable if we
pull out some conditions into variables here, not sure, something like
$ip_changed, $new_ip_not_static, $old_ip_not_static for both ipv4 and ipv6
Though I'm not sure...

> +
> +    #add new ip first
> +    my $new_ipv4_allocated = undef;
> +    eval {
> +	if (!$net->{ip}) {
> +	    $net->{ip} = PVE::Network::SDN::Vnets::get_next_free_cidr($bridge, $hostname, $mac, $description);
> +	    $new_ipv4_allocated = 1;
> +	} elsif ($net->{ip} ne 'dhcp' && $net->{ip} ne 'manual') {
> +
> +	    if ($oldnet->{ip} && $net->{ip} eq $oldnet->{ip}) {

would be !$ip_changed

> +		#update ip attributes if no ip address change
> +		PVE::Network::SDN::Vnets::update_cidr($bridge, $net->{ip}, $hostname, $oldhostname, $mac, $description);
> +	    } else {
> +		PVE::Network::SDN::Vnets::add_cidr($bridge, $net->{ip}, $hostname, $mac, $description) if !$oldnet->{ip} || $net->{ip} ne $oldnet->{ip};

I think it's better to move the post-fix condition up to replace the
`else` with an `elsif`, since IMO it should also cover the line below,
right?
Note that the 2nd half of the condition is already implied at this
point.

> +		$new_ipv4_allocated = 1;
> +	    }
> +	}
> +    };
> +    if ($@) {
> +	die $@;
> +    }
> +
> +    #then add ip6
> +    eval {
> +	if (!$net->{ip6}) {
> +	    $net->{ip6} = PVE::Network::SDN::Vnets::get_next_free_cidr($bridge, $hostname, $mac, $description, 6);
> +	} elsif ($net->{ip6} ne 'dhcp' && $net->{ip6} ne 'manual' && $net->{ip6} ne 'auto') {
> +	    if ($oldnet->{ip6} && $net->{ip6} eq $oldnet->{ip6}) {
> +		#update ipv6 attributes if no ip address change
> +		PVE::Network::SDN::Vnets::update_cidr($bridge, $net->{ip6}, $hostname, $oldhostname, $mac, $description);
> +	    } else {
> +		PVE::Network::SDN::Vnets::add_cidr($bridge, $net->{ip6}, $hostname, $mac, $description) if !$oldnet->{ip6} || $net->{ip6} ne $oldnet->{ip6};

^ same here

> +	    }
> +	}
> +    };
> +    if ($@) {
> +	my $err = $@;
> +	#if error, delete previously added ipv4
> +	if ($new_ipv4_allocated) {
> +	    eval {
> +		PVE::Network::SDN::Vnets::del_cidr($bridge, $net->{ip}, $hostname, $description);
> +	    };
> +	}
> +	die $err;
> +    }
> +
> +    #delete old ip
> +    if ($oldnet->{ip} && $oldnet->{ip} ne 'dhcp' && $oldnet->{ip} ne 'manual') {
> +	my $deletebridge = $oldbridge ne $bridge ? $oldbridge : $bridge;
> +	eval {
> +	    PVE::Network::SDN::Vnets::del_cidr($deletebridge, $oldnet->{ip}, $hostname, $description) if !$net->{ip} || $net->{ip} ne $oldnet->{ip};
> +	};
> +	warn $@ if $@;
> +    }
> +
> +    #delete old ip6
> +    if ($oldnet->{ip6} && $oldnet->{ip6} ne 'dhcp' && $oldnet->{ip6} ne 'manual' && $net->{ip6} ne 'auto') {
> +	my $deletebridge = $oldbridge ne $bridge ? $oldbridge : $bridge;
> +	eval {
> +	    PVE::Network::SDN::Vnets::del_cidr($deletebridge, $oldnet->{ip6}, $hostname) if !$net->{ip6} || $net->{ip6} ne $oldnet->{ip6};
> +	};
> +	warn $@ if $@;
> +    }
> +
> +    #update gateway
> +    if ($net->{ip} && $net->{ip} ne 'dhcp' && $net->{ip} ne 'manual') {
> +	my ($ip, $mask) = split(/\//, $net->{ip});
> +	my ($subnetidv4, $subnetv4) = PVE::Network::SDN::Subnets::find_ip_subnet($ip, $mask, $subnets);
> +	$net->{gw} = $subnetv4->{gateway} if $subnetv4->{gateway};
> +    }
> +
> +    if ($net->{ip6} && $net->{ip6} ne 'dhcp' && $net->{ip6} ne 'manual' && $net->{ip6} ne 'auto') {
> +	my ($ip6, $mask6) = split(/\//, $net->{ip6});
> +	my ($subnetidv6, $subnetv6) = PVE::Network::SDN::Subnets::find_ip_subnet($ip6, $mask6, $subnets);
> +	$net->{gw6} = $subnetv6->{gateway} if $subnetv6->{gateway};
> +    }
> +
> +}
> +
> +sub delete_net_ip {
> +    my ($hostname, $net, $description) = @_;
> +
> +    return if !$have_sdn;
> +
> +    my $bridge = $net->{bridge};
> +    my $subnets = PVE::Network::SDN::Vnets::get_subnets($bridge);
> +    return if !keys %{$subnets};
> +
> +    if ($net->{ip6} && ($net->{ip6} eq 'auto' || $net->{ip6} eq 'manual' || $net->{ip6} eq 'dhcp')) {

So here this reads as: "$net->{ip6} is one of 'auto', 'manual' or 'dhcp'"
with the helper mentioned above:
  if ($net->{ip6} && !ip_is_static($net->{ip6}))

> +	eval {
> +	    PVE::Network::SDN::Vnets::del_cidr($bridge, $net->{ip6}, $hostname, $description);
> +	};
> +	warn $@ if $@;
> +    } elsif ($net->{ip} && $net->{ip} ne 'dhcp' && $net->{ip} ne 'manual') {

but the IPv4 case here reads as: "$net->{ip} is set but is neither 'dhcp' nor 'manual'"
with the helper mentioned above:
  if ($net->{ip} && ip_is_static($net->{ip}))

this seems inconsistent (note the '!' on the `ip_is_static` call differ
between the ipv4 & ipv6 cases)

I think you meant to use the same condition for both?

> +	eval {
> +	    PVE::Network::SDN::Vnets::del_cidr($bridge, $net->{ip}, $hostname, $description);
> +	};
> +	warn $@ if $@;
> +    }
> +}
> +
>  1;
> -- 
> 2.20.1




  reply	other threads:[~2021-05-12 12:27 UTC|newest]

Thread overview: 6+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-05-10 15:37 [pve-devel] [PATCH container/manager/network] RFC: lxc: " Alexandre Derumier
2021-05-10 15:37 ` [pve-devel] [PATCH pve-container 1/1] " Alexandre Derumier
2021-05-12 12:27   ` Wolfgang Bumiller [this message]
2021-05-18 11:43     ` aderumier
2021-05-10 15:37 ` [pve-devel] [PATCH pve-network 1/1] sdn: get_local_vnets : add ipam && vlanaware values Alexandre Derumier
2021-05-10 15:37 ` [pve-devel] [PATCH pve-manager 1/1] ui: lxc: network: sdn display improvements Alexandre Derumier

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=20210512122702.3relfzag2yrqv2lo@wobu-vie.proxmox.com \
    --to=w.bumiller@proxmox.com \
    --cc=aderumier@odiso.com \
    --cc=pve-devel@lists.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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
Service provided by Proxmox Server Solutions GmbH | Privacy | Legal