Merge pull request #15496 from kampfschlaefer/containers_more_veth_interfaces

Declarative containers: more veth interfaces
This commit is contained in:
Franz Pletz 2016-07-31 19:13:59 +02:00 committed by GitHub
commit d7f7ef4c21
7 changed files with 390 additions and 185 deletions

View File

@ -6,35 +6,212 @@ let
# The container's init script, a small wrapper around the regular
# NixOS stage-2 init script.
containerInit = pkgs.writeScript "container-init"
containerInit = (cfg:
let
renderExtraVeth = (name: cfg:
''
echo "Bringing ${name} up"
ip link set dev ${name} up
${optionalString (cfg . "localAddress" or null != null) ''
echo "Setting ip for ${name}"
ip addr add ${cfg . "localAddress"} dev ${name}
''}
${optionalString (cfg . "localAddress6" or null != null) ''
echo "Setting ip6 for ${name}"
ip -6 addr add ${cfg . "localAddress6"} dev ${name}
''}
${optionalString (cfg . "hostAddress" or null != null) ''
echo "Setting route to host for ${name}"
ip route add ${cfg . "hostAddress"} dev ${name}
''}
${optionalString (cfg . "hostAddress6" or null != null) ''
echo "Setting route6 to host for ${name}"
ip -6 route add ${cfg . "hostAddress6"} dev ${name}
''}
''
);
in
pkgs.writeScript "container-init"
''
#! ${pkgs.stdenv.shell} -e
# Initialise the container side of the veth pair.
if [ "$PRIVATE_NETWORK" = 1 ]; then
ip link set host0 name eth0
ip link set dev eth0 up
if [ -n "$LOCAL_ADDRESS" ]; then
ip addr add $LOCAL_ADDRESS dev eth0
fi
if [ -n "$LOCAL_ADDRESS6" ]; then
ip -6 addr add $LOCAL_ADDRESS6 dev eth0
fi
if [ -n "$HOST_ADDRESS" ]; then
ip route add $HOST_ADDRESS dev eth0
ip route add default via $HOST_ADDRESS
fi
if [ -n "$HOST_ADDRESS6" ]; then
ip -6 route add $HOST_ADDRESS6 dev eth0
ip -6 route add default via $HOST_ADDRESS6
fi
${concatStringsSep "\n" (mapAttrsToList renderExtraVeth cfg . "extraVeths" or {})}
ip a
ip r
fi
# Start the regular stage 1 script.
exec "$1"
''
);
nspawnExtraVethArgs = (name: cfg: "--network-veth-extra=${name}");
startScript = (cfg:
''
#! ${pkgs.stdenv.shell} -e
mkdir -p -m 0755 "$root/etc" "$root/var/lib"
mkdir -p -m 0700 "$root/var/lib/private" "$root/root" /run/containers
if ! [ -e "$root/etc/os-release" ]; then
touch "$root/etc/os-release"
fi
if ! [ -e "$root/etc/machine-id" ]; then
touch "$root/etc/machine-id"
fi
mkdir -p -m 0755 \
"/nix/var/nix/profiles/per-container/$INSTANCE" \
"/nix/var/nix/gcroots/per-container/$INSTANCE"
cp --remove-destination /etc/resolv.conf "$root/etc/resolv.conf"
# Initialise the container side of the veth pair.
if [ "$PRIVATE_NETWORK" = 1 ]; then
ip link set host0 name eth0
ip link set dev eth0 up
if [ -n "$LOCAL_ADDRESS" ]; then
ip addr add $LOCAL_ADDRESS dev eth0
fi
if [ -n "$LOCAL_ADDRESS6" ]; then
ip -6 addr add $LOCAL_ADDRESS6 dev eth0
fi
if [ -n "$HOST_ADDRESS" ]; then
ip route add $HOST_ADDRESS dev eth0
ip route add default via $HOST_ADDRESS
fi
if [ -n "$HOST_ADDRESS6" ]; then
ip -6 route add $HOST_ADDRESS6 dev eth0
ip -6 route add default via $HOST_ADDRESS6
extraFlags+=" --network-veth"
if [ -n "$HOST_BRIDGE" ]; then
extraFlags+=" --network-bridge=$HOST_BRIDGE"
fi
fi
# Start the regular stage 1 script.
exec "$1"
'';
${if cfg . "extraVeths" or null != null then
''extraFlags+=" ${concatStringsSep " " (mapAttrsToList nspawnExtraVethArgs cfg . "extraVeths" or {})}"''
else
''# No extra veth pairs to create''
}
for iface in $INTERFACES; do
extraFlags+=" --network-interface=$iface"
done
for iface in $MACVLANS; do
extraFlags+=" --network-macvlan=$iface"
done
# If the host is 64-bit and the container is 32-bit, add a
# --personality flag.
${optionalString (config.nixpkgs.system == "x86_64-linux") ''
if [ "$(< ''${SYSTEM_PATH:-/nix/var/nix/profiles/per-container/$INSTANCE/system}/system)" = i686-linux ]; then
extraFlags+=" --personality=x86"
fi
''}
# Run systemd-nspawn without startup notification (we'll
# wait for the container systemd to signal readiness).
EXIT_ON_REBOOT=1 \
exec ${config.systemd.package}/bin/systemd-nspawn \
--keep-unit \
-M "$INSTANCE" -D "$root" $extraFlags \
$EXTRA_NSPAWN_FLAGS \
--notify-ready=yes \
--bind-ro=/nix/store \
--bind-ro=/nix/var/nix/db \
--bind-ro=/nix/var/nix/daemon-socket \
--bind="/nix/var/nix/profiles/per-container/$INSTANCE:/nix/var/nix/profiles" \
--bind="/nix/var/nix/gcroots/per-container/$INSTANCE:/nix/var/nix/gcroots" \
--setenv PRIVATE_NETWORK="$PRIVATE_NETWORK" \
--setenv HOST_BRIDGE="$HOST_BRIDGE" \
--setenv HOST_ADDRESS="$HOST_ADDRESS" \
--setenv LOCAL_ADDRESS="$LOCAL_ADDRESS" \
--setenv HOST_ADDRESS6="$HOST_ADDRESS6" \
--setenv LOCAL_ADDRESS6="$LOCAL_ADDRESS6" \
--setenv PATH="$PATH" \
${containerInit cfg} "''${SYSTEM_PATH:-/nix/var/nix/profiles/system}/init"
''
);
preStartScript = (cfg:
''
# Clean up existing machined registration and interfaces.
machinectl terminate "$INSTANCE" 2> /dev/null || true
if [ "$PRIVATE_NETWORK" = 1 ]; then
ip link del dev "ve-$INSTANCE" 2> /dev/null || true
ip link del dev "vb-$INSTANCE" 2> /dev/null || true
fi
${concatStringsSep "\n" (
mapAttrsToList (name: cfg:
''ip link del dev ${name} 2> /dev/null || true ''
) cfg . "extraVeths" or {}
)}
''
);
postStartScript = (cfg:
let
ipcall = (cfg: ipcmd: variable: attribute:
if cfg . attribute or null == null then
''
if [ -n "${variable}" ]; then
${ipcmd} add ${variable} dev $ifaceHost
fi
''
else
''${ipcmd} add ${cfg . attribute} dev $ifaceHost''
);
renderExtraVeth = (name: cfg:
if cfg . "hostBridge" or null != null then
''
# Add ${name} to bridge ${cfg.hostBridge}
ip link set dev ${name} master ${cfg.hostBridge} up
''
else
''
# Set IPs and routes for ${name}
${optionalString (cfg . "hostAddress" or null != null) ''
ip addr add ${cfg . "hostAddress"} dev ${name}
''}
${optionalString (cfg . "hostAddress6" or null != null) ''
ip -6 addr add ${cfg . "hostAddress6"} dev ${name}
''}
${optionalString (cfg . "localAddress" or null != null) ''
ip route add ${cfg . "localAddress"} dev ${name}
''}
${optionalString (cfg . "localAddress6" or null != null) ''
ip -6 route add ${cfg . "localAddress6"} dev ${name}
''}
''
);
in
''
if [ "$PRIVATE_NETWORK" = 1 ]; then
if [ -z "$HOST_BRIDGE" ]; then
ifaceHost=ve-$INSTANCE
ip link set dev $ifaceHost up
${ipcall cfg "ip addr" "$HOST_ADDRESS" "hostAddress"}
${ipcall cfg "ip -6 addr" "$HOST_ADDRESS6" "hostAddress6"}
${ipcall cfg "ip route" "$LOCAL_ADDRESS" "localAddress"}
${ipcall cfg "ip -6 route" "$LOCAL_ADDRESS6" "localAddress6"}
fi
${concatStringsSep "\n" (mapAttrsToList renderExtraVeth cfg . "extraVeths" or {})}
fi
# Get the leader PID so that we can signal it in
# preStop. We can't use machinectl there because D-Bus
# might be shutting down. FIXME: in systemd 219 we can
# just signal systemd-nspawn to do a clean shutdown.
machinectl show "$INSTANCE" | sed 's/Leader=\(.*\)/\1/;t;d' > "/run/containers/$INSTANCE.pid"
''
);
system = config.nixpkgs.system;
@ -73,6 +250,63 @@ let
mkBindFlags = bs: concatMapStrings mkBindFlag (lib.attrValues bs);
networkOptions = {
hostBridge = mkOption {
type = types.nullOr types.string;
default = null;
example = "br0";
description = ''
Put the host-side of the veth-pair into the named bridge.
Only one of hostAddress* or hostBridge can be given.
'';
};
hostAddress = mkOption {
type = types.nullOr types.str;
default = null;
example = "10.231.136.1";
description = ''
The IPv4 address assigned to the host interface.
(Not used when hostBridge is set.)
'';
};
hostAddress6 = mkOption {
type = types.nullOr types.string;
default = null;
example = "fc00::1";
description = ''
The IPv6 address assigned to the host interface.
(Not used when hostBridge is set.)
'';
};
localAddress = mkOption {
type = types.nullOr types.str;
default = null;
example = "10.231.136.2";
description = ''
The IPv4 address assigned to the interface in the container.
If a hostBridge is used, this should be given with netmask to access
the whole network. Otherwise the default netmask is /32 and routing is
set up from localAddress to hostAddress and back.
'';
};
localAddress6 = mkOption {
type = types.nullOr types.string;
default = null;
example = "fc00::2";
description = ''
The IPv6 address assigned to the interface in the container.
If a hostBridge is used, this should be given with netmask to access
the whole network. Otherwise the default netmask is /128 and routing is
set up from localAddress6 to hostAddress6 and back.
'';
};
};
in
{
@ -133,56 +367,6 @@ in
'';
};
hostBridge = mkOption {
type = types.nullOr types.string;
default = null;
example = "br0";
description = ''
Put the host-side of the veth-pair into the named bridge.
Only one of hostAddress* or hostBridge can be given.
'';
};
hostAddress = mkOption {
type = types.nullOr types.str;
default = null;
example = "10.231.136.1";
description = ''
The IPv4 address assigned to the host interface.
(Not used when hostBridge is set.)
'';
};
hostAddress6 = mkOption {
type = types.nullOr types.string;
default = null;
example = "fc00::1";
description = ''
The IPv6 address assigned to the host interface.
(Not used when hostBridge is set.)
'';
};
localAddress = mkOption {
type = types.nullOr types.str;
default = null;
example = "10.231.136.2";
description = ''
The IPv4 address assigned to <literal>eth0</literal>
in the container.
'';
};
localAddress6 = mkOption {
type = types.nullOr types.string;
default = null;
example = "fc00::2";
description = ''
The IPv6 address assigned to <literal>eth0</literal>
in the container.
'';
};
interfaces = mkOption {
type = types.listOf types.string;
default = [];
@ -192,6 +376,15 @@ in
'';
};
extraVeths = mkOption {
type = types.attrsOf types.optionSet;
default = {};
options = networkOptions;
description = ''
Extra veth-pairs to be created for the container
'';
};
autoStart = mkOption {
type = types.bool;
default = false;
@ -214,7 +407,7 @@ in
'';
};
};
} // networkOptions;
config = mkMerge
[ (mkIf options.config.isDefined {
@ -272,108 +465,11 @@ in
environment.INSTANCE = "%i";
environment.root = "/var/lib/containers/%i";
preStart =
''
# Clean up existing machined registration and interfaces.
machinectl terminate "$INSTANCE" 2> /dev/null || true
preStart = preStartScript {};
if [ "$PRIVATE_NETWORK" = 1 ]; then
ip link del dev "ve-$INSTANCE" 2> /dev/null || true
ip link del dev "vb-$INSTANCE" 2> /dev/null || true
fi
'';
script = startScript {};
script =
''
mkdir -p -m 0755 "$root/etc" "$root/var/lib"
mkdir -p -m 0700 "$root/var/lib/private" "$root/root" /run/containers
if ! [ -e "$root/etc/os-release" ]; then
touch "$root/etc/os-release"
fi
if ! [ -e "$root/etc/machine-id" ]; then
touch "$root/etc/machine-id"
fi
mkdir -p -m 0755 \
"/nix/var/nix/profiles/per-container/$INSTANCE" \
"/nix/var/nix/gcroots/per-container/$INSTANCE"
cp --remove-destination /etc/resolv.conf "$root/etc/resolv.conf"
if [ "$PRIVATE_NETWORK" = 1 ]; then
extraFlags+=" --network-veth"
if [ -n "$HOST_BRIDGE" ]; then
extraFlags+=" --network-bridge=$HOST_BRIDGE"
fi
fi
for iface in $INTERFACES; do
extraFlags+=" --network-interface=$iface"
done
for iface in $MACVLANS; do
extraFlags+=" --network-macvlan=$iface"
done
# If the host is 64-bit and the container is 32-bit, add a
# --personality flag.
${optionalString (config.nixpkgs.system == "x86_64-linux") ''
if [ "$(< ''${SYSTEM_PATH:-/nix/var/nix/profiles/per-container/$INSTANCE/system}/system)" = i686-linux ]; then
extraFlags+=" --personality=x86"
fi
''}
# Run systemd-nspawn without startup notification (we'll
# wait for the container systemd to signal readiness).
EXIT_ON_REBOOT=1 \
exec ${config.systemd.package}/bin/systemd-nspawn \
--keep-unit \
-M "$INSTANCE" -D "$root" $extraFlags \
$EXTRA_NSPAWN_FLAGS \
--notify-ready=yes \
--bind-ro=/nix/store \
--bind-ro=/nix/var/nix/db \
--bind-ro=/nix/var/nix/daemon-socket \
--bind="/nix/var/nix/profiles/per-container/$INSTANCE:/nix/var/nix/profiles" \
--bind="/nix/var/nix/gcroots/per-container/$INSTANCE:/nix/var/nix/gcroots" \
--setenv PRIVATE_NETWORK="$PRIVATE_NETWORK" \
--setenv HOST_BRIDGE="$HOST_BRIDGE" \
--setenv HOST_ADDRESS="$HOST_ADDRESS" \
--setenv LOCAL_ADDRESS="$LOCAL_ADDRESS" \
--setenv HOST_ADDRESS6="$HOST_ADDRESS6" \
--setenv LOCAL_ADDRESS6="$LOCAL_ADDRESS6" \
--setenv PATH="$PATH" \
${containerInit} "''${SYSTEM_PATH:-/nix/var/nix/profiles/system}/init"
'';
postStart =
''
if [ "$PRIVATE_NETWORK" = 1 ]; then
if [ -z "$HOST_BRIDGE" ]; then
ifaceHost=ve-$INSTANCE
ip link set dev $ifaceHost up
if [ -n "$HOST_ADDRESS" ]; then
ip addr add $HOST_ADDRESS dev $ifaceHost
fi
if [ -n "$HOST_ADDRESS6" ]; then
ip -6 addr add $HOST_ADDRESS6 dev $ifaceHost
fi
if [ -n "$LOCAL_ADDRESS" ]; then
ip route add $LOCAL_ADDRESS dev $ifaceHost
fi
if [ -n "$LOCAL_ADDRESS6" ]; then
ip -6 route add $LOCAL_ADDRESS6 dev $ifaceHost
fi
fi
fi
# Get the leader PID so that we can signal it in
# preStop. We can't use machinectl there because D-Bus
# might be shutting down. FIXME: in systemd 219 we can
# just signal systemd-nspawn to do a clean shutdown.
machinectl show "$INSTANCE" | sed 's/Leader=\(.*\)/\1/;t;d' > "/run/containers/$INSTANCE.pid"
'';
postStart = postStartScript {};
preStop =
''
@ -425,15 +521,20 @@ in
[{ name = "container@"; value = unit; }]
# declarative containers
++ (mapAttrsToList (name: cfg: nameValuePair "container@${name}" (
unit // {
preStart = preStartScript cfg;
script = startScript cfg;
postStart = postStartScript cfg;
} // (
if cfg.autoStart then
unit // {
{
wantedBy = [ "multi-user.target" ];
wants = [ "network.target" ];
after = [ "network.target" ];
restartTriggers = [ cfg.path ];
reloadIfChanged = true;
}
else null
else {})
)) config.containers)
));
@ -462,11 +563,11 @@ in
LOCAL_ADDRESS6=${cfg.localAddress6}
''}
''}
INTERFACES="${toString cfg.interfaces}"
${optionalString cfg.autoStart ''
AUTO_START=1
''}
EXTRA_NSPAWN_FLAGS="${mkBindFlags cfg.bindMounts}"
INTERFACES="${toString cfg.interfaces}"
${optionalString cfg.autoStart ''
AUTO_START=1
''}
EXTRA_NSPAWN_FLAGS="${mkBindFlags cfg.bindMounts}"
'';
}) config.containers;

View File

@ -218,6 +218,7 @@ in rec {
tests.containers-ipv6 = callTest tests/containers-ipv6.nix {};
tests.containers-bridge = callTest tests/containers-bridge.nix {};
tests.containers-imperative = callTest tests/containers-imperative.nix {};
tests.containers-extra_veth = callTest tests/containers-extra_veth.nix {};
tests.docker = hydraJob (import tests/docker.nix { system = "x86_64-linux"; });
tests.dockerRegistry = hydraJob (import tests/docker-registry.nix { system = "x86_64-linux"; });
tests.dnscrypt-proxy = callTest tests/dnscrypt-proxy.nix { system = "x86_64-linux"; };

View File

@ -10,7 +10,7 @@ in
import ./make-test.nix ({ pkgs, ...} : {
name = "containers-bridge";
meta = with pkgs.stdenv.lib.maintainers; {
maintainers = [ aristid aszlig eelco chaoflow ];
maintainers = [ aristid aszlig eelco chaoflow kampfschlaefer ];
};
machine =

View File

@ -0,0 +1,103 @@
# Test for NixOS' container support.
import ./make-test.nix ({ pkgs, ...} : {
name = "containers-bridge";
meta = with pkgs.stdenv.lib.maintainers; {
maintainers = [ kampfschlaefer ];
};
machine =
{ config, pkgs, ... }:
{ imports = [ ../modules/installer/cd-dvd/channel.nix ];
virtualisation.writableStore = true;
virtualisation.memorySize = 768;
virtualisation.vlans = [];
networking.bridges = {
br0 = {
interfaces = [];
};
br1 = { interfaces = []; };
};
networking.interfaces = {
br0 = {
ip4 = [{ address = "192.168.0.1"; prefixLength = 24; }];
ip6 = [{ address = "fc00::1"; prefixLength = 7; }];
};
br1 = {
ip4 = [{ address = "192.168.1.1"; prefixLength = 24; }];
};
};
containers.webserver =
{
autoStart = true;
privateNetwork = true;
hostBridge = "br0";
localAddress = "192.168.0.100/24";
localAddress6 = "fc00::2/7";
extraVeths = {
veth1 = { hostBridge = "br1"; localAddress = "192.168.1.100/24"; };
veth2 = { hostAddress = "192.168.2.1"; localAddress = "192.168.2.100"; };
};
config =
{
networking.firewall.allowedTCPPorts = [ 80 ];
networking.firewall.allowPing = true;
};
};
virtualisation.pathsInNixDB = [ pkgs.stdenv ];
};
testScript =
''
$machine->waitForUnit("default.target");
$machine->succeed("nixos-container list") =~ /webserver/ or die;
# Status of the webserver container.
$machine->succeed("nixos-container status webserver") =~ /up/ or die;
# Debug
#$machine->succeed("nixos-container run webserver -- ip link >&2");
# Ensure that the veths are inside the container
$machine->succeed("nixos-container run webserver -- ip link show veth1") =~ /state UP/ or die;
$machine->succeed("nixos-container run webserver -- ip link show veth2") =~ /state UP/ or die;
# Debug
#$machine->succeed("ip link >&2");
# Ensure the presence of the extra veths
$machine->succeed("ip link show veth1") =~ /state UP/ or die;
$machine->succeed("ip link show veth2") =~ /state UP/ or die;
# Ensure the veth1 is part of br1 on the host
$machine->succeed("ip link show veth1") =~ /master br1/ or die;
# Debug
#$machine->succeed("ip -4 a >&2");
#$machine->succeed("ip -4 r >&2");
#$machine->succeed("nixos-container run webserver -- ip link >&2");
#$machine->succeed("nixos-container run webserver -- ip -4 a >&2");
#$machine->succeed("nixos-container run webserver -- ip -4 r >&2");
# Ping on main veth
$machine->succeed("ping -n -c 1 192.168.0.100");
$machine->succeed("ping6 -n -c 1 fc00::2");
# Ping on the first extra veth
$machine->succeed("ping -n -c 1 192.168.1.100 >&2");
# Ping on the second extra veth
$machine->succeed("ping -n -c 1 192.168.2.100 >&2");
# Stop the container.
$machine->succeed("nixos-container stop webserver");
$machine->fail("ping -n -c 1 192.168.1.100 >&2");
$machine->fail("ping -n -c 1 192.168.2.100 >&2");
# Destroying a declarative container should fail.
$machine->fail("nixos-container destroy webserver");
'';
})

View File

@ -3,7 +3,7 @@
import ./make-test.nix ({ pkgs, ...} : {
name = "containers-imperative";
meta = with pkgs.stdenv.lib.maintainers; {
maintainers = [ aristid aszlig eelco chaoflow ];
maintainers = [ aristid aszlig eelco chaoflow kampfschlaefer ];
};
machine =

View File

@ -3,7 +3,7 @@
import ./make-test.nix ({ pkgs, ...} : {
name = "containers-ipv4";
meta = with pkgs.stdenv.lib.maintainers; {
maintainers = [ aristid aszlig eelco chaoflow ];
maintainers = [ aristid aszlig eelco chaoflow kampfschlaefer ];
};
machine =

View File

@ -8,7 +8,7 @@ in
import ./make-test.nix ({ pkgs, ...} : {
name = "containers-ipv6";
meta = with pkgs.stdenv.lib.maintainers; {
maintainers = [ aristid aszlig eelco chaoflow ];
maintainers = [ aristid aszlig eelco chaoflow kampfschlaefer ];
};
machine =