We're trying to configure load balancing for certain integrations that listen on a specific TCP port for incoming traffic, for example Cisco FTD.
The LB is Nginx and it's load balancing on L4 (stream module). Here's the Nginx configuration:
# Cisco FTD LB
upstream ftd-ingress {
server server01:9003;
server server02:9003;
}
server {
listen 9003;
proxy_pass ftd-ingress;
proxy_protocol on;
proxy_connect_timeout 360s;
proxy_timeout 360s;
}
When proxy_protocol
is set to on
it doesn't work because the Agent does not understand the Proxy Protocol. Is there any way we can load balance the traffic while still having the source IP information in Elasticsearch?
On UDP it is possible with proxy_bind $remote_addr transparent;
, however that does not work on TCP.
Cheers,
Luka