Subject
I would like some feedback concerning a use case
I have questions about TLS/SSL/STARTTLS/OpenSSL
Other
Description
I've deployed Docker Mailserver on Kubernetes by following the recommendations in the docs. This Mailserver is behind an Nginx ingress controller by using the proxy protocol.
This all works well, until trying to send mail from within the cluster's network, which doesn't go through the ingress controller.
When I try to send mail from another pod/container on the network (through an ExternalName) it will fail for any protocol used (smtp:25, smtps:465/587) by logging the following to the Mailserver logs:
smtp:25 postfix/postscreen[142071]: warning: haproxy read: time limit exceeded
-- OR --
smtp:465
postfix/smtps/smtpd[147024]: warning: haproxy read: timeout error
postfix/smtps/smtpd[147024]: connect from 10-1-39-118.some-svc.some-svc.svc.cluster.local[10.1.39.118]
postfix/smtps/smtpd[147024]: disconnect from 10-1-39-118.some-svc.some-svc.svc.cluster.local[10.1.39.118] commands=0/0
Question
The question here is: How would one configure the mailserver to accept mail from the cluster's network without (I suppose) trying to terminate the proxy protocol?
Configuration
apiVersion: v1
kind: ConfigMap
metadata:
name: config
namespace: mailserver
data:
postfix-master.cf: |
submission/inet/smtpd_upstream_proxy_protocol = haproxy
smtps/inet/smtpd_upstream_proxy_protocol = haproxy
postfix-main.cf: |
mydestination = localhost.$mydomain, localhost
postscreen_upstream_proxy_protocol = haproxy
dovecot.cf: |
# The following line should include the internal pod's network CIDR and localhost address
haproxy_trusted_networks = 10.1.0.0/16, 127.0.0.0/8
service imap-login {
inet_listener imaps {
haproxy = yes
}
}
TrustedHosts: |
127.0.0.1
localhost
Subject
I would like some feedback concerning a use case
I have questions about TLS/SSL/STARTTLS/OpenSSL
Other
Description
I've deployed Docker Mailserver on Kubernetes by following the recommendations in the docs. This Mailserver is behind an Nginx ingress controller by using the proxy protocol.
This all works well, until trying to send mail from within the cluster's network, which doesn't go through the ingress controller.
When I try to send mail from another pod/container on the network (through an ExternalName) it will fail for any protocol used (smtp:25, smtps:465/587) by logging the following to the Mailserver logs:
smtp:25
postfix/postscreen[142071]: warning: haproxy read: time limit exceeded-- OR --
smtp:465
Question
The question here is: How would one configure the mailserver to accept mail from the cluster's network without (I suppose) trying to terminate the proxy protocol?
Configuration