Scaling socket.io

Problem

socket.io requires sticky sessions in order to support multiple backends. For Meshblu in large scale settings, we have used haproxy to provide stickiness.

For use with Docker Swarm see: octoblu/swarm-haproxy-socket-io

Example

global
  stats socket /tmp/haproxy.sock
  maxconn 80000

resolvers dns
  nameserver dns1 8.8.8.8:53
  hold valid 2m

defaults
  log    global
  mode   http
  timeout client 60s            # Client and server timeout must match the longest
  timeout server 300s           # time we may wait for a response from the server.
  timeout queue  120s           # Don't queue requests too long if saturated.
  timeout connect 10s           # There's no reason to change this one.
  timeout http-request 300s     # A complete request may never take that long.
  timeout tunnel 2h
  retries         3
  option redispatch
  option httplog
  option dontlognull
  option http-server-close      # enable HTTP connection closing on the server side
  option abortonclose           # enable early dropping of aborted requests from pending queue
  option httpchk                # enable HTTP protocol to check on servers health
  stats auth opsworks:opsworks
  stats uri /haproxy?stats

backend meshblu-socket-io
  balance roundrobin
  timeout queue 5000
  timeout server 86400000
  timeout connect 86400000
  timeout check 1s
  option forwardfor
  no option httpclose
  option http-server-close
  option forceclose
  # this must be the partial url
  option httpchk GET /healthcheck
  stick-table type string len 40 size 20M expire 2m
  stick store-response set-cookie(io)
  stick on cookie(io)
  stick on url_param(sid)

  server server1 server1.local:80 resolvers dns resolve-prefer ipv4 check inter 1m
  server server2 server2.local:80 resolvers dns resolve-prefer ipv4 check inter 1m

frontend http-in
  bind :80
  default_backend meshblu-socket-io