update

docs/assets/setup.sh

@@ -0,0 +1,221 @@
+#!/bin/bash
+
+# Exit immediately if a command exits with a non-zero status
+set -e
+
+# Function to check if a namespace exists
+function ns_exists() {
+  ip netns list | grep -w "$1" &> /dev/null
+}
+
+# Function to add a namespace if it does not exist
+function create_namespace() {
+  if ns_exists "$1"; then
+    echo "Namespace $1 already exists."
+  else
+    ip netns add "$1"
+    echo "Namespace $1 created."
+  fi
+}
+
+# Function to create veth pairs and check for existence
+function create_veth_pair() {
+  if ip link show "$1" &> /dev/null; then
+    echo "$1 already exists."
+  else
+    ip link add "$1" type veth peer name "$2"
+    echo "Created veth pair: $1 <--> $2."
+  fi
+}
+
+# Function to assign IP address to veth interface in a namespace
+function assign_ip_in_ns() {
+  if ip netns exec "$1" ip addr show "$2" | grep "$3" &> /dev/null; then
+    echo "IP address $3 is already set for $2 in namespace $1."
+  else
+    ip netns exec "$1" ip addr add "$3" dev "$2"
+    echo "Assigned IP address $3 to $2 in namespace $1."
+  fi
+}
+
+# Function to manually add ARP entry and verify success
+function add_arp_entry() {
+  local ip_addr=$1
+  local mac_addr=$2
+  sudo arp -s "$ip_addr" "$mac_addr"
+  if arp -n | grep "$ip_addr" &> /dev/null; then
+    echo "Successfully added ARP entry for $ip_addr with MAC $mac_addr."
+  else
+    echo "Failed to add ARP entry for $ip_addr."
+    exit 1
+  fi
+}
+
+# Create namespaces for the load balancer and backends
+create_namespace "lb"
+create_namespace "h2"
+create_namespace "h3"
+
+# Create veth pairs for LB to Local Machine, and LB to backends
+create_veth_pair "veth0" "veth0_lb"
+create_veth_pair "veth2" "veth2_lb"
+create_veth_pair "veth3" "veth3_lb"
+
+# Attach veth pairs to their respective namespaces
+ip link set veth0_lb netns lb
+ip link set veth2 netns h2   # Attach veth2 to h2 namespace
+ip link set veth3 netns h3   # Attach veth3 to h3 namespace
+ip link set veth2_lb netns lb
+ip link set veth3_lb netns lb
+
+# Assign IP addresses with separate subnets to namespaces and host
+assign_ip_in_ns "lb" "veth0_lb" "10.0.0.10/24"  # Load balancer subnet
+assign_ip_in_ns "lb" "veth2_lb" "10.0.1.1/24"   # h2 subnet
+assign_ip_in_ns "lb" "veth3_lb" "10.0.2.1/24"   # h3 subnet
+
+assign_ip_in_ns "h2" "veth2" "10.0.1.2/24"      # h2 namespace IP
+assign_ip_in_ns "h3" "veth3" "10.0.2.2/24"      # h3 namespace IP
+
+# Assign IP for veth0 on local machine (direct connection to LB)
+if ip addr show veth0 | grep "10.0.0.11" &> /dev/null; then
+  echo "IP address 10.0.0.11 is already set for veth0."
+else
+  ip addr add 10.0.0.11/24 dev veth0
+  echo "Assigned IP address 10.0.0.11 to veth0."
+fi
+
+# Bring up interfaces
+ip link set veth0 up
+ip netns exec lb ip link set veth0_lb up
+ip netns exec lb ip link set veth2_lb up
+ip netns exec lb ip link set veth3_lb up
+
+# Enable IP forwarding in the load balancer namespace
+ip netns exec lb sysctl -w net.ipv4.ip_forward=1
+
+# Add ARP entries for load balancer connection
+mac_lb_veth0=$(sudo ip netns exec lb ip link show veth0_lb | grep ether | awk '{print $2}')
+sudo arp -s 10.0.0.10 $mac_lb_veth0
+echo "Added ARP entry for 10.0.0.10 with MAC $mac_lb_veth0."
+
+# Add static ARP entries for the backend namespaces
+mac_h2=$(sudo ip netns exec lb ip link show veth2_lb | grep ether | awk '{print $2}')
+sudo arp -s 10.0.1.2 $mac_h2
+echo "Added ARP entry for 10.0.1.2 with MAC $mac_h2."
+
+mac_h3=$(sudo ip netns exec lb ip link show veth3_lb | grep ether | awk '{print $2}')
+sudo arp -s 10.0.2.2 $mac_h3
+echo "Added ARP entry for 10.0.2.2 with MAC $mac_h3."
+
+# Bring up interfaces in h2 and h3 namespaces
+ip netns exec h2 ip link set dev veth2 up
+ip netns exec h2 ip link set dev lo up
+ip netns exec h3 ip link set dev veth3 up
+ip netns exec h3 ip link set dev lo up
+
+# Enable IP masquerading on the lb namespace (NAT)
+ip netns exec lb iptables -t nat -A POSTROUTING -o veth2_lb -j MASQUERADE
+ip netns exec lb iptables -t nat -A POSTROUTING -o veth3_lb -j MASQUERADE
+
+# Disable reverse path filtering on veth0 and lb interfaces
+sysctl -w net.ipv4.conf.veth0.rp_filter=0
+ip netns exec lb sysctl -w net.ipv4.conf.veth0_lb.rp_filter=0
+ip netns exec lb sysctl -w net.ipv4.conf.veth2_lb.rp_filter=0
+ip netns exec lb sysctl -w net.ipv4.conf.veth3_lb.rp_filter=0
+
+# Allow forwarding between veth0_lb and veth2_lb in lb namespace
+ip netns exec lb iptables -A FORWARD -i veth0_lb -o veth2_lb -j ACCEPT
+ip netns exec lb iptables -A FORWARD -i veth2_lb -o veth0_lb -j ACCEPT
+
+# Allow forwarding between veth0_lb and veth3_lb in lb namespace
+ip netns exec lb iptables -A FORWARD -i veth0_lb -o veth3_lb -j ACCEPT
+ip netns exec lb iptables -A FORWARD -i veth3_lb -o veth0_lb -j ACCEPT
+
+########################################################################
+echo ""
+echo "Start checking..."
+echo ""
+
+# Check ARP table on the host
+echo "Checking ARP table on the host..."
+arp -n
+
+# Check the status of veth2 in the h2 namespace
+echo "Checking interface veth2 inside h2 namespace..."
+ip netns exec h2 ip addr show veth2
+ip netns exec h2 ip link show veth2
+
+# Check the status of the routing table on the host
+echo "Checking routing table on the host..."
+ip route show
+
+# Check IP forwarding status on the host
+echo "Checking IP forwarding status on the host..."
+sysctl net.ipv4.ip_forward
+
+# Check IP forwarding status in the load balancer namespace (lb)
+echo "Checking IP forwarding status in lb namespace..."
+ip netns exec lb sysctl net.ipv4.ip_forward
+
+# Additional checks for troubleshooting
+echo "Checking interface states inside lb namespace..."
+ip netns exec lb ip link show
+
+echo "Checking ARP table inside lb namespace..."
+ip netns exec lb ip neigh show
+
+echo "Checking routing table inside lb namespace..."
+ip netns exec lb ip route show
+
+echo "Checking packet statistics for veth interfaces in lb namespace..."
+ip netns exec lb ip -s link show veth0_lb
+ip netns exec lb ip -s link show veth2_lb
+ip netns exec lb ip -s link show veth3_lb
+
+echo "Checking NAT rules in lb namespace..."
+ip netns exec lb iptables -t nat -L -v
+
+echo "Checking reverse path filtering in h2 and h3..."
+ip netns exec h2 sysctl net.ipv4.conf.all.rp_filter
+ip netns exec h3 sysctl net.ipv4.conf.all.rp_filter
+
+# Check if any firewall rules are blocking traffic
+echo "Checking firewall rules on the host..."
+sudo iptables -L
+sudo iptables -t nat -L
+
+# Debug routing table inside LB namespace
+echo "LB namespace routing table:"
+ip netns exec lb ip route
+
+# Connectivity tests
+
+# Function to test connectivity
+function test_ping() {
+  local target=$1
+  local source=$2
+
+  echo "Testing connectivity from $source to $target..."
+  if ping -c 3 "$target" &> /dev/null; then
+    echo "Ping from $source to $target succeeded."
+  else
+    echo "Ping from $source to $target failed."
+    exit 1
+  fi
+}
+
+# Test connectivity from host to load balancer (veth0 -> veth0_lb)
+test_ping "10.0.0.10" "host"
+
+# Test connectivity from host to h2 and h3
+test_ping "10.0.1.2" "host"
+test_ping "10.0.2.2" "host"
+
+# Test connectivity within the load balancer namespace (lb -> h2 and lb -> h3)
+echo "Testing connectivity from lb to h2..."
+ip netns exec lb ping -c 3 10.0.1.2 || echo "Ping from lb to h2 failed."
+
+echo "Testing connectivity from lb to h3..."
+ip netns exec lb ping -c 3 10.0.2.2 || echo "Ping from lb to h3 failed."
+
+echo "Network setup and tests completed!"

docs/assets/teardown.sh

@@ -0,0 +1,33 @@
+#!/bin/bash
+
+set -xe
+
+rm_bridge () {
+  if ip link show $1 &> /dev/null; then
+    ip link set dev $1 down
+    brctl delbr $1
+  fi
+}
+
+rm_pair () {
+  if ip link show $1 &> /dev/null; then
+    ip link delete $1 type veth
+  fi
+}
+
+rm_ns () {
+  if [ -e "/var/run/netns/$1" ]; then
+    ip netns delete $1
+  fi
+}
+
+rm_bridge br0
+
+rm_pair veth0
+rm_pair veth2
+rm_pair veth4
+rm_pair veth6
+
+rm_ns h2
+rm_ns h3
+rm_ns lb

docs/blogs/index.md

@@ -1,5 +1,6 @@
 # Blogs about eunomia-bpf community
 
+- [Modernizing eBPF for the Next Decade: Past, Present, and Future](ten-years.md)
 - [Simplifying Kernel Programming: The LLM-Powered eBPF Tool](kgent.md)
 - [Expanding eBPF Compile Once, Run Everywhere(CO-RE) to Userspace Compatibility](../tutorials/38-btf-uprobe/README.md)
 - [Implementing an Inline Hook in C in 5 minutes](inline-hook.md)