Skip to content

Instantly share code, notes, and snippets.

@level323
Created August 3, 2016 21:14
Show Gist options
  • Star 39 You must be signed in to star a gist
  • Fork 9 You must be signed in to fork a gist
  • Save level323/54a921216f0baaa163127d960bfebbf0 to your computer and use it in GitHub Desktop.
Save level323/54a921216f0baaa163127d960bfebbf0 to your computer and use it in GitHub Desktop.
Run a command inside a customised networking environment (using cgroups)
#!/bin/bash
# === INFO ===
# altnetworking.sh
# Description: Run the specified application in a custom networking environment.
# Uses cgroups to run process(es) in a network environment of your own choosing (within limits!)
VERSION="0.1.0"
# Author: John Clark
# Requirements: Debian 8 Jessie (plus iptables 1.6 from unstable)
#
# This script was derived from the excellent "novpn.sh" script by KrisWebDev
# as found here: https://gist.github.com/kriswebdev/a8d291936fe4299fb17d3744497b1170
#
# === LICENSE ===
# This program is free software: you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
# the Free Software Foundation, either version 3 of the License, or
# (at your option) any later version.
#
# This program is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
# GNU General Public License for more details.
#
# You should have received a copy of the GNU General Public License
# along with this program. If not, see <http://www.gnu.org/licenses/>.
# This script will `source` a configuration script identified by the first command line argument.
# That configuration script is expected to define:
#
# 1. The name of the interface that we want the command to emit packets from
# e.g. desired_interface="eth0"
# 2. The default route we want to have for our cgroup
# e.g. desired_default_route=`ip route | grep "dev ${desired_interface}" | awk '/^default/ { print $3 }'`
# 3. The name for the cgroup we're going to create
# Note: Better keep it with purely lowercase alphabetic & underscore
# e.g. cgroup_name="vpntinclpgmtwireless"
# 4. The classid we're going to use for this cgroup
# Note: Anything from 0x00000001 to 0xFFFFFFFF (just needs to be unique)
# e.g. net_cls_classid="0x00110011"
# 5. The mark we're going to put on packets originating from this app/cgroup
# Note: Anything from 1 to 2147483647
# e.g. ip_table_fwmark="11"
# 6. The Routing table number we'll associate with packets on this cgroup
# Note: Anything from 1 to 252 (just needs to be unique)
# e.g. ip_table_number="11"
# 7. The routing table name we're going to use to formulate the full routing table name
# Note: Needs to be unique. Best to use cgroup name
# e.g. ip_table_name="$cgroup_name"
# 8. Define a post_up() function that will be called after everything about
# the cgroup is set up (including default route)
# Use it to create additional needed routes and/or iptables entries
# e,g post_up(){
# echo "Adding routes to LPGMT lan via vpn-tinc-telstra-wireless-01-node-11"
# sudo ip route add 10.0.1.0/24 via 10.11.12.9 dev "$desired_interface" table "$ip_table_name"
# sudo ip route add 10.0.99.0/24 via 10.11.12.9 dev "$desired_interface" table "$ip_table_name"
# }
# 9. Define a pre_down() function that will be called before everything about the cgroup is is torn down
# Use it to undo everything in post_up
# e.g. pre_down(){
# echo "Removing routes to LPGMT lan"
# sudo ip route del 10.0.1.0/24 via 10.11.12.9 dev "$desired_interface" table "$ip_table_name"
# sudo ip route del 10.0.99.0/24 via 10.11.12.9 dev "$desired_interface" table "$ip_table_name"
# }
#
# 10 Define the test_networking() function that will carry out tests to confirm that the
# networking environment that has been created is functioning properly. It should return 0 if
# networking is functioning correctly or 1 otherwise. If returning 0, set testresult=true else
# set it to false
# e.g. test_networking(){
# echo "Networking was not tested by the test_networking function. Confirm it's working manually if you feel the need"
# testresult=true
# return 0
# }
# === CODE ===
set -u
# Some defaults
force=false
testresult=true
# Handle options
action="command"
background=false
skip=false
init_nb_args="$#"
show_help() {
me=`basename "$0"`
echo -e "Usage : \e[1m$me \e[4mCONFIG\e[24m [\e[4mOPTIONS\e[24m] [\e[4mCOMMAND\e[24m [\e[4mCOMMAND PARAMETERS\e[24m]]\e[0m"
echo -e "Run command using different networking configuration (via cgroups)."
echo
echo -e "\e[1m\e[4mCONFIG\e[0m: Full path to the configuration file"
echo -e "\e[1m\e[4mOPTIONS\e[0m:"
echo -e "\e[1m-b, --background\e[0m Start \e[4mCOMMAND\e[24m as background process (release the shell)."
echo -e "\e[1m-l, --list\e[0m List processes running in this special cgroup namespace."
echo -e "\e[1m-s, --skip\e[0m Don't check/setup system config & don't ask for root,\n\
run \e[4mCOMMAND\e[24m even if network connectivity tests fail."
echo -e "\e[1m-c, --clean\e[0m Terminate all proceses inside cgroup and remove system config."
echo -e "\e[1m-v, --version\e[0m Print this program version."
echo -e "\e[1m-h, --help\e[0m This help."
}
config_file_name="$1"
if [ -f "$config_file_name" ]
then
source "$config_file_name"
else
show_help
exit 1
fi
shift
while [ "$#" -gt 0 ]; do
case "$1" in
-b|--background) background=true; shift 1;;
-l|--list) action="list"; shift 1;;
-s|--skip) skip=true; shift 1;;
-l|--clean) action="clean"; shift 1;;
-h|--help) action="help"; shift 1;;
-v|--version) echo "altnetworking.sh v$VERSION"; exit 0;;
-*) echo "Unknown option: $1. Try --help." >&2; exit 1;;
*) break;; # Start of COMMAND or LIST
esac
done
# Respond to --help
if [ "$init_nb_args" -lt 1 ] || [ "$action" = "help" ] ; then
show_help
exit 1
fi
# Helper functions
# Check the presence of required system packages
check_package(){
nothing_installed=1
for package_name in "$@"
do
if ! dpkg -l "$package_name" &> /dev/null; then
echo "Installing $package_name"
sudo apt-get install "$package_name"
nothing_installed=0
fi
done
return $nothing_installed
}
# List processes running inside the cgroup
list_processes(){
return_status=1
echo -e "PID""\t""CMD"
while read task_pid
do
echo -e "${task_pid}""\t""`ps -p ${task_pid} -o comm=`";
return_status=0
done < /sys/fs/cgroup/net_cls/${cgroup_name}/tasks
return $return_status
}
# Check and setup iptables - requires root even for check
iptable_checked=false
setup_iptables(){
if ! sudo iptables -t mangle -C OUTPUT -m cgroup --cgroup "$net_cls_classid" -j MARK --set-mark "$ip_table_fwmark" 2>/dev/null; then
echo "Adding iptables MANGLE rule to set firewall mark $ip_table_fwmark on packets with class identifier $net_cls_classid" >&2
sudo iptables -t mangle -A OUTPUT -m cgroup --cgroup "$net_cls_classid" -j MARK --set-mark "$ip_table_fwmark"
fi
if ! sudo iptables -t nat -C POSTROUTING -m cgroup --cgroup "$net_cls_classid" -o "$desired_interface" -j MASQUERADE 2>/dev/null; then
echo "Adding iptables NAT rule force the packets with class identifier $net_cls_classid to exit through $desired_interface" >&2
sudo iptables -t nat -A POSTROUTING -m cgroup --cgroup "$net_cls_classid" -o "$desired_interface" -j MASQUERADE
fi
iptable_checked=true
}
# Test if config is working, IPv4 only
test_connection(){
# Call the configuration function to test if networking is functioning
test_networking
}
check_iptables=false
if [ "$action" = "command" ]
then
# SETUP config
if [ "$skip" = false ]; then
echo "Checking/setting forced routing config (skip with $0 -s ...)" >&2
if check_package cgroupfs-mount cgmanager cgroup-tools inetutils-traceroute; then
echo "You may want to reboot now. But that's probably not necessary." >&2
exit 1
fi
if dpkg --compare-versions `sudo iptables --version | grep -oP "iptables v\K.*$"` "lt" "1.6"; then
echo -e "\e[31mYou need iptables 1.6.0+. Please install manually. Aborting.\e[0m" >&2
echo "Find latest iptables at http://www.netfilter.org/projects/iptables/downloads.html" >&2
echo "Commands to install iptables 1.6.0:" >&2
echo "sudo apt-get install iptables/unstable libxtables11/unstable" >&2
echo "... or compile from source as shown below:" >&2
echo -e "\e[34msudo apt-get install dh-autoreconf bison flex
cd /tmp
curl http://www.netfilter.org/projects/iptables/files/iptables-1.6.0.tar.bz2 | tar xj
cd iptables-1.6.0
./configure --prefix=/usr \\
--sbindir=/sbin \\
--disable-nftables \\
--enable-libipq \\
--with-xtlibdir=/lib/xtables \\
&& make \\
&& sudo make install
iptables --version\e[0m" >&2
exit 1
fi
if [ ! -d "/sys/fs/cgroup/net_cls/$cgroup_name" ]; then
echo "Creating net_cls control group $cgroup_name" >&2
sudo mkdir -p "/sys/fs/cgroup/net_cls/$cgroup_name"
check_iptables=true
fi
if [ `cat "/sys/fs/cgroup/net_cls/$cgroup_name/net_cls.classid" | xargs -n 1 printf "0x%08x"` != "$net_cls_classid" ]; then
echo "Applying net_cls class identifier $net_cls_classid to cgroup $cgroup_name" >&2
echo "$net_cls_classid" | sudo tee "/sys/fs/cgroup/net_cls/$cgroup_name/net_cls.classid" > /dev/null
fi
if ! grep -E "^${ip_table_number}\s+$ip_table_name" /etc/iproute2/rt_tables &>/dev/null; then
if grep -E "^${ip_table_number}\s+" /etc/iproute2/rt_tables; then
echo "ERROR: Table ${ip_table_number} already exists in /etc/iproute2/rt_tables with a different name than $ip_table_name" >&2
exit 1
fi
echo "Creating ip routing table: number=$ip_table_number name=$ip_table_name" >&2
echo "$ip_table_number $ip_table_name" | sudo tee -a /etc/iproute2/rt_tables > /dev/null
check_iptables=true
fi
if ! ip rule list | grep " lookup $ip_table_name" | grep " fwmark " &>/dev/null; then
echo "Adding rule to use ip routing table $ip_table_name for packets with firewall mark $ip_table_fwmark" >&2
sudo ip rule add fwmark "$ip_table_fwmark" table "$ip_table_name"
check_iptables=true
fi
if [ -z "`ip route list table "$ip_table_name" default via $desired_default_route dev ${desired_interface} 2>/dev/null`" ]; then
echo "Adding default route in ip routing table $ip_table_name via $desired_default_route dev $desired_interface" >&2
sudo ip route add default via "$desired_default_route" dev "$desired_interface" table "$ip_table_name"
# Now run custom post_up script
post_up
# Useless?
echo "Flushing ip route cache" >&2
sudo ip route flush cache
check_iptables=true
fi
if [ "`cat /proc/sys/net/ipv4/conf/all/rp_filter`" != "0" ] || [ "`cat /proc/sys/net/ipv4/conf/all/rp_filter`" != "2" ]; then
echo "Unset reverse path filtering for interface \"all\"" >&2
echo 2 | sudo tee "/proc/sys/net/ipv4/conf/all/rp_filter" > /dev/null
check_iptables=true
fi
if [ "`cat /proc/sys/net/ipv4/conf/${desired_interface}/rp_filter`" != "0" ] || [ "`cat /proc/sys/net/ipv4/conf/${desired_interface}/rp_filter`" != "2" ]; then
echo "Unset reverse path filtering for interface \"${desired_interface}\"" >&2
echo 2 | sudo tee "/proc/sys/net/ipv4/conf/${desired_interface}/rp_filter" > /dev/null
check_iptables=true
fi
if [ -z "`lscgroup net_cls:$cgroup_name`" ] || [ `stat -c "%U" /sys/fs/cgroup/net_cls/${cgroup_name}/tasks` != "$USER" ]; then
echo "Creating cgroup net_cls:${cgroup_name}. User $USER will be able to move tasks to it without root permissions." >&2
sudo cgcreate -t "$USER":"$USER" -a `id -g -n "$USER"`:`id -g -n "$USER"` -g net_cls:"$cgroup_name"
check_iptables=true
fi
if [ "$check_iptables" = true ]; then
setup_iptables
fi
fi
# TEST bypass
test_connection
if [ "$force" != true ]; then
if [ "$testresult" = false ]; then
if [ "$iptable_checked" = false ]; then
echo -e "Testing iptables..." >&2
setup_iptables
test_connection
fi
fi
if [ "$testresult" = false ]; then
exit 1
fi
fi
fi
# RUN command
if [ "$action" = "command" ]; then
if [ "$#" -eq 0 ]; then
echo "Error: COMMAND not provided." >&2
exit 1
fi
if [ "$background" = true ]; then
cgexec -g net_cls:"$cgroup_name" "$@" &>/dev/null &
exit 0
else
cgexec -g net_cls:"$cgroup_name" "$@"
exit $?
fi
# List processes using this cgroup
# Exit code 0 (true) if at least 1 process is running in the cgroup
elif [ "$action" = "list" ]; then
echo "List of processes using cgroup $cgroup_name:"
list_processes
exit $?
# CLEAN the mess
elif [ "$action" = "clean" ]; then
echo -e "Cleaning forced routing config generated by this script."
echo -e "Don't bother with errors meaning there's nothing to remove."
# Kill tasks in cgroup
if [ -f "/sys/fs/cgroup/net_cls/${cgroup_name}/tasks" ]; then
while read task_pid; do sudo kill ${task_pid} ; done < "/sys/fs/cgroup/net_cls/${cgroup_name}/tasks"
fi
# Run custom pre_down function
pre_down
# Delete cgroup
if [ -d "/sys/fs/cgroup/net_cls/${cgroup_name}" ]; then
sudo find "/sys/fs/cgroup/net_cls/${cgroup_name}" -depth -type d -print -exec rmdir {} \;
fi
# (DISABLED BECAUSE MY MACHINE DEFAULTS TO RPF BEING OFF) Re-enable Reverse Path Filtering
#echo 1 | sudo tee "/proc/sys/net/ipv4/conf/all/rp_filter" > /dev/null
#echo 1 | sudo tee "/proc/sys/net/ipv4/conf/${desired_interface}/rp_filter" > /dev/null
sudo iptables -t mangle -D OUTPUT -m cgroup --cgroup "$net_cls_classid" -j MARK --set-mark "$ip_table_fwmark"
sudo iptables -t nat -D POSTROUTING -m cgroup --cgroup "$net_cls_classid" -o "$desired_interface" -j MASQUERADE
sudo ip rule del fwmark "$ip_table_fwmark" table "$ip_table_name"
sudo ip route del default table "$ip_table_name"
sudo sed -i '/^${ip_table_number}\s\+${ip_table_name}\s*$/d' /etc/iproute2/rt_tables
if [ -n "`lscgroup net_cls:$cgroup_name`" ]; then
sudo cgdelete net_cls:"$cgroup_name"
fi
echo "All done."
fi
# BONUS: Useful commands:
# ./altnetworking.sh traceroute www.google.com
# ip=$(./altnetworking.sh curl 'https://wtfismyip.com/text' 2>/dev/null); echo "$ip"; whois "$ip" | grep -E "inetnum|route|netname|descr"
@ezbik
Copy link

ezbik commented Sep 21, 2020

https://github.com/ezbik/scripts/tree/master/altnetworking

improved version ( no sudo garbage, prevents IP leaks , ip rule & iptables workarounds).

@level323
Copy link
Author

https://github.com/ezbik/scripts/tree/master/altnetworking

improved version ( no sudo garbage, prevents IP leaks , ip rule & iptables workarounds).

Thanks for your contribution!

@viertelb
Copy link

viertelb commented Feb 2, 2021

Thanks for those! Unfortunately, I get an error when executing it on Raspbian Buster. I seem not to be the only one. https://superuser.com/questions/1569055/iptables-cgroup-module-not-found-with-kernel-4-19-118-v7l-and-iptables-v1-8-2

Also, I would like to ask, if I get it to work, should I be able to use it with a systemd process?

@ezbik
Copy link

ezbik commented Feb 3, 2021

@bvrulez
it is a known issue on Raspbian. If you don't want to rebuild its kernel, simply try Ubuntu for Raspberry PI. It has all the modules.

@akruis
Copy link

akruis commented Aug 23, 2021

https://gist.github.com/akruis/da385170f33799112df713ec7c3e9305 updated for Debian 10 bullseye, uses cgroups v2

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment