Sticky Sessions Configuration in Load Balancers

Sticky sessions (session persistence) ensure that requests from the same client route to the same backend server for the duration of a session. This prevents session loss when applications maintain in-memory state rather than using centralized session storage. This guide covers cookie-based persistence in Nginx and HAProxy, IP hash techniques, app-specific cookies, and session replication alternatives.

Sticky Sessions Overview
Cookie-Based Persistence in Nginx
Cookie-Based Persistence in HAProxy
IP Hash Load Balancing
Application-Specific Cookies
Session Replication
Session Affinity Timeouts
Testing Sticky Sessions
Troubleshooting

Sticky Sessions Overview

Sticky sessions use multiple mechanisms:

Cookie-Based: Proxy sets/modifies cookie to route requests
IP-Based: Hash client IP to determine server
App-Cookie-Based: Use existing application cookie for routing
Source IP + Port: Hash connection source and port

Limitations of sticky sessions:

Prevents load distribution changes
Makes server maintenance harder
Reduces effective capacity
Increases request latency (hash calculations)
Session loss on server failure

Better alternatives:

Distributed session storage (Redis, Memcached)
Stateless application design
Session database (PostgreSQL, MySQL)
Centralized cache

Use sticky sessions only when necessary for stateful applications.

Nginx requires third-party modules for native cookie-based persistence. Use the sticky directive or implement with map and route:

Using Sticky Module (if compiled in)

upstream backend {
    least_conn;
    
    server 192.168.1.100:8000;
    server 192.168.1.101:8000;
    server 192.168.1.102:8000;
    
    sticky cookie srv_route expires=1h domain=.example.com path=/ httponly secure;
}

server {
    listen 443 ssl http2;
    server_name app.example.com;
    
    ssl_certificate /etc/nginx/ssl/example.com.crt;
    ssl_certificate_key /etc/nginx/ssl/example.com.key;
    
    location / {
        proxy_pass http://backend;
        proxy_set_header Host $host;
        proxy_set_header X-Real-IP $remote_addr;
        proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
    }
}

Using Map-Based Routing (without module)

# Define upstream servers with identifiers
upstream backend_1 { server 192.168.1.100:8000; }
upstream backend_2 { server 192.168.1.101:8000; }
upstream backend_3 { server 192.168.1.102:8000; }

# Create map to determine backend based on session ID
map $cookie_session_id $upstream {
    # Hash the session cookie to one of three backends
    ~*^(?<prefix>[a-f0-9]{2}) $prefix;
    default "00";
}

server {
    listen 80;
    server_name app.example.com;
    
    # Extract session hash value
    set $session_route $upstream;
    
    location / {
        # Route based on session hash
        if ($session_route ~* "^00$") { proxy_pass http://backend_1; }
        if ($session_route ~* "^01$") { proxy_pass http://backend_2; }
        if ($session_route ~* "^02$") { proxy_pass http://backend_3; }
        
        # Set session cookie if not exists
        add_header Set-Cookie "session_id=$request_id; Path=/; HttpOnly; Max-Age=3600" always;
        
        proxy_set_header Host $host;
        proxy_set_header X-Real-IP $remote_addr;
    }
}

Using Consistent Hash

Implement consistent hashing for better distribution:

# Lua-based consistent hashing (requires ngx_lua module)
upstream backend {
    server 192.168.1.100:8000;
    server 192.168.1.101:8000;
    server 192.168.1.102:8000;
}

# Nginx Lua script for consistent hashing
location / {
    set_by_lua_file $backend_pool /etc/nginx/lua/consistent_hash.lua $cookie_session_id;
    proxy_pass http://$backend_pool;
}

# Create /etc/nginx/lua/consistent_hash.lua
# Local consistent hash function
local function crc32(data)
    local CRC32_POLY = 0xEDB88320
    local crc = 0xFFFFFFFF
    for i = 1, #data do
        crc = bit.bxor(crc, string.byte(data, i))
        for _ = 1, 8 do
            if bit.band(crc, 1) == 1 then
                crc = bit.bxor(bit.rshift(crc, 1), CRC32_POLY)
            else
                crc = bit.rshift(crc, 1)
            end
        end
    end
    return bit.bxor(crc, 0xFFFFFFFF)
end

local session_id = ngx.arg[1]
local servers = {"192.168.1.100", "192.168.1.101", "192.168.1.102"}

local hash = crc32(session_id)
local selected = servers[hash % #servers + 1]
return selected .. ":8000"

HAProxy provides native cookie-based persistence:

global
    log stdout local0
    stats socket /run/haproxy/admin.sock

defaults
    mode http
    timeout connect 5000
    timeout client 50000
    timeout server 50000

frontend web_in
    bind *:80
    default_backend web_servers

backend web_servers
    balance roundrobin
    
    # Enable cookie-based persistence
    cookie SERVERID insert indirect secure httponly
    
    server srv1 192.168.1.100:8000 check cookie srv1
    server srv2 192.168.1.101:8000 check cookie srv2
    server srv3 192.168.1.102:8000 check cookie srv3

Parameters:

SERVERID: Cookie name
insert: Add new cookie if missing
indirect: Don't remove HAProxy-set cookie
secure: Set secure flag for HTTPS
httponly: Set HttpOnly flag
cookie srv1: Server identifier

backend web_servers
    balance roundrobin
    
    # Cookie with domain, path, and expiration
    cookie SERVERID insert indirect secure httponly nocache domain .example.com path /
    
    # Use existing application cookie for routing
    appsession JSESSIONID len 52 timeout 1h
    
    server srv1 192.168.1.100:8000 check cookie srv1
    server srv2 192.168.1.101:8000 check cookie srv2

backend web_servers
    balance roundrobin
    
    cookie SERVERID insert indirect httponly
    
    # Primary servers
    server srv1 192.168.1.100:8000 check cookie srv1
    server srv2 192.168.1.101:8000 check cookie srv2
    
    # Backup servers (if session lost)
    server srv3 192.168.1.102:8000 check cookie srv3 backup
    server srv4 192.168.1.103:8000 check cookie srv4 backup

Sticky Sessions with Multiple Paths

backend web_servers
    balance roundrobin
    cookie SERVERID insert indirect secure httponly
    
    stick-table type string len 32 size 100k expire 30m
    stick on cookie(JSESSIONID)
    
    server srv1 192.168.1.100:8000 check cookie srv1
    server srv2 192.168.1.101:8000 check cookie srv2
    server srv3 192.168.1.102:8000 check cookie srv3

IP Hash Load Balancing

IP hash (source IP based routing) provides persistence without cookies:

Nginx IP Hash

upstream backend {
    ip_hash;
    
    server 192.168.1.100:8000 weight=3;
    server 192.168.1.101:8000 weight=2;
    server 192.168.1.102:8000 weight=1;
}

server {
    listen 80;
    server_name app.example.com;
    
    location / {
        proxy_pass http://backend;
        # Important: Use X-Forwarded-For only if trusted sources
        proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
    }
}

IP hash characteristics:

Deterministic: Same client IP always routes to same server
No cookies required
Survives proxy/NAT transitions (if behind same NAT)
Server down causes remapping for ~1/N clients

HAProxy Source Hash

backend web_servers
    balance source
    
    server srv1 192.168.1.100:8000 check
    server srv2 192.168.1.101:8000 check
    server srv3 192.168.1.102:8000 check

With client connection tracking:

backend web_servers
    balance source
    
    # Track connections by source IP
    stick-table type ip size 100k expire 1h
    stick on src
    
    server srv1 192.168.1.100:8000 check
    server srv2 192.168.1.101:8000 check

Application-Specific Cookies

Route based on application session cookies:

HAProxy appsession

backend web_servers
    # Use existing JSESSIONID (Java)
    appsession JSESSIONID len 52 timeout 1h
    
    server srv1 192.168.1.100:8000 check
    server srv2 192.168.1.101:8000 check
    server srv3 192.168.1.102:8000 check

# Map application session ID to backend
map $cookie_phpsessionid $php_backend {
    ~(?P<hash>.+) http://backend;
}

upstream backend {
    server 192.168.1.100:8000;
    server 192.168.1.101:8000;
    server 192.168.1.102:8000;
}

server {
    listen 80;
    
    location / {
        proxy_pass http://backend;
        proxy_set_header X-Session-ID $cookie_phpsessionid;
    }
}

frontend web_in
    bind *:80
    
    # Extract customer ID from session cookie
    set-var(sess.customer_id) cookie(sessionid)
    
    use_backend gold_servers if { var(sess.customer_id) -m reg -i ^gold_ }
    use_backend silver_servers if { var(sess.customer_id) -m reg -i ^silver_ }
    default_backend bronze_servers

backend gold_servers
    balance roundrobin
    server srv1 192.168.1.110:8000 check
    server srv2 192.168.1.111:8000 check

backend silver_servers
    balance roundrobin
    server srv3 192.168.1.120:8000 check

backend bronze_servers
    balance roundrobin
    server srv4 192.168.1.130:8000 check

Session Replication

Move away from sticky sessions by replicating sessions:

Redis-Based Session Storage

Configure application to use Redis:

# Install Redis
sudo apt install redis-server
sudo systemctl start redis-server

# Verify Redis
redis-cli ping

Example with Spring Boot (Java):

# application.yml
spring:
  session:
    store-type: redis
  redis:
    host: localhost
    port: 6379
    timeout: 2000ms

Example with Node.js (Express):

const session = require('express-session');
const RedisStore = require('connect-redis').default;
const { createClient } = require('redis');

const redisClient = createClient();
redisClient.connect();

app.use(session({
    store: new RedisStore({ client: redisClient }),
    secret: 'secret-key',
    resave: false,
    saveUninitialized: false,
    cookie: { 
        secure: true, 
        maxAge: 1800000 
    }
}));

Memcached Session Storage

Use Memcached for distributed session cache:

# Install Memcached
sudo apt install memcached
sudo systemctl start memcached

Configure application (PHP example):

<?php
ini_set('session.save_handler', 'memcached');
ini_set('session.save_path', 'localhost:11211');

session_start();
$_SESSION['user_id'] = 123;
?>

Database Session Storage

Store sessions in shared database:

-- Create sessions table
CREATE TABLE sessions (
    id VARCHAR(255) PRIMARY KEY,
    user_id INT,
    data TEXT,
    created_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP,
    updated_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP ON UPDATE CURRENT_TIMESTAMP,
    expires_at TIMESTAMP,
    INDEX(expires_at)
);

Configure HAProxy to use database backend:

backend session_db
    mode tcp
    balance roundrobin
    server db1 192.168.1.200:5432 check
    server db2 192.168.1.201:5432 check

Session Affinity Timeouts

Configure session persistence durations:

Nginx Timeout

upstream backend {
    least_conn;
    keepalive 32;
    keepalive_timeout 60s;
    
    server 192.168.1.100:8000;
    server 192.168.1.101:8000;
}

# Sticky using Traefik-style header
map $http_x_session_id $session_backend {
    ~(?P<sid>.+) http://backend;
}

server {
    location / {
        proxy_pass http://backend;
        
        # Session timeout
        proxy_read_timeout 30s;
        proxy_send_timeout 30s;
        proxy_connect_timeout 5s;
    }
}

HAProxy Timeout

backend web_servers
    balance roundrobin
    
    # Cookie expires in 1 hour
    cookie SERVERID insert indirect max-age 3600
    
    # Stick table expires in 30 minutes
    stick-table type ip size 100k expire 1800s
    stick on src
    
    timeout server 30s
    timeout connect 5s
    
    server srv1 192.168.1.100:8000 check inter 2000

Testing Sticky Sessions

Test cookie-based persistence:

# Extract session cookie
COOKIE=$(curl -s -c - http://app.example.com/ | grep -i server | awk '{print $NF}')

# Make multiple requests with same cookie
for i in {1..5}; do
    curl -s -b "SERVERID=$COOKIE" http://app.example.com/ | grep -i server
done

Test IP hash routing:

# Multiple requests from same IP should go to same server
for i in {1..5}; do
    curl -s http://app.example.com/ | head -1
done

# Test from different IPs (using VPN or proxy)
for ip in 1.2.3.4 5.6.7.8 9.10.11.12; do
    curl -s --interface $ip http://app.example.com/ | head -1
done

Verify session replication:

# Check Redis sessions
redis-cli
> KEYS *
> GET session:*

# Check session count
> DBSIZE

# Monitor session access
redis-cli MONITOR

Troubleshooting

Check sticky session configuration:

# Nginx check
nginx -T | grep -A 10 "upstream"

# HAProxy check
haproxy -f /etc/haproxy/haproxy.cfg -c
echo "show backend" | socat - /run/haproxy/admin.sock

Monitor session cookies:

# Capture cookie traffic
tcpdump -A -s 1024 'tcp port 80' | grep -i "set-cookie"

# Monitor cookie with curl
curl -v -c cookies.txt http://app.example.com/ | head -20
cat cookies.txt

# Verify cookie attributes
curl -i http://app.example.com/ | grep -i "set-cookie"

Verify server routing:

# Add tracking headers in proxy
# Test multiple requests capture the Server header
for i in {1..10}; do
    curl -s -b "SERVERID=srv1" http://app.example.com/ | grep -i "X-Backend-Server"
done

# Check HAProxy stats
curl http://localhost:8404/stats | grep -i server

Test session loss scenarios:

# Kill a backend server
ssh 192.168.1.100 "sudo systemctl stop application"

# Attempt request with existing session
curl -b "SERVERID=srv1" http://app.example.com/

# Verify failover to backup
curl -b "SERVERID=srv1" http://app.example.com/ | grep -i server

Conclusion

Sticky sessions enable stateful application deployments but limit scalability and operational flexibility. While cookie-based and IP hash persistence solve immediate session problems, distributed session storage with Redis or Memcached provides superior scalability and resilience. Evaluate stateless application architecture as the preferred solution, implementing sticky sessions only when necessary and with clear timeout policies and backup mechanisms.

Sticky sessions configuration in load balancers

Sticky Sessions Configuration in Load Balancers

Table of Contents

Sticky Sessions Overview

Using Sticky Module (if compiled in)

Using Map-Based Routing (without module)

Using Consistent Hash

Sticky Sessions with Multiple Paths

IP Hash Load Balancing

Nginx IP Hash

HAProxy Source Hash

Application-Specific Cookies

HAProxy appsession

Session Replication

Redis-Based Session Storage

Memcached Session Storage

Database Session Storage

Session Affinity Timeouts

Nginx Timeout

HAProxy Timeout

Testing Sticky Sessions

Troubleshooting

Conclusion

Último Video

Obtén $20 de Crédito Gratis

Sticky sessions configuration in load balancers

Sticky Sessions Configuration in Load Balancers

Table of Contents

Sticky Sessions Overview

Cookie-Based Persistence in Nginx

Using Sticky Module (if compiled in)

Using Map-Based Routing (without module)

Using Consistent Hash

Cookie-Based Persistence in HAProxy

Basic Cookie Persistence

Advanced Cookie Configuration

Cookie with Failure Handling

Sticky Sessions with Multiple Paths

IP Hash Load Balancing

Nginx IP Hash

HAProxy Source Hash

Application-Specific Cookies

HAProxy appsession

Nginx with Application Cookie

Route Based on Cookie Value

Session Replication

Redis-Based Session Storage

Memcached Session Storage

Database Session Storage

Session Affinity Timeouts

Nginx Timeout

HAProxy Timeout

Testing Sticky Sessions

Troubleshooting

Conclusion

Último Video

Obtén $20 de Crédito Gratis