Kubernetes
Introduction
This article describes how to deploy DMS to Kubernetes. We highly recommend everyone to use our community DMS Helm chart.
Requirements
- Basic knowledge about Kubernetes from the reader.
- A basic understanding of mail servers.
- Ideally, the reader has already deployed DMS before with a simpler setup (
docker run
or Docker Compose).
Limited Support
DMS does not officially support Kubernetes. This content is entirely community-supported. If you find errors, please open an issue and raise a PR.
Manually Writing Manifests
If using our Helm chart is not viable for you, here is some guidance to start with your own manifests.
Provide the basic configuration via environment variables with a ConfigMap
.
Example
Below is only an example configuration, adjust the ConfigMap
to your own needs.
---
apiVersion: v1
kind: ConfigMap
metadata:
name: mailserver.environment
immutable: false
data:
TLS_LEVEL: modern
POSTSCREEN_ACTION: drop
OVERRIDE_HOSTNAME: mail.example.com
FAIL2BAN_BLOCKTYPE: drop
POSTMASTER_ADDRESS: postmaster@example.com
UPDATE_CHECK_INTERVAL: 10d
POSTFIX_INET_PROTOCOLS: ipv4
ENABLE_CLAMAV: '1'
ENABLE_POSTGREY: '0'
ENABLE_FAIL2BAN: '1'
AMAVIS_LOGLEVEL: '-1'
SPOOF_PROTECTION: '1'
MOVE_SPAM_TO_JUNK: '1'
ENABLE_UPDATE_CHECK: '1'
ENABLE_SPAMASSASSIN: '1'
SUPERVISOR_LOGLEVEL: warn
SPAMASSASSIN_SPAM_TO_INBOX: '1'
# here, we provide an example for the SSL configuration
SSL_TYPE: manual
SSL_CERT_PATH: /secrets/ssl/rsa/tls.crt
SSL_KEY_PATH: /secrets/ssl/rsa/tls.key
You can also make use of user-provided configuration files (e.g. user-patches.sh
, postfix-accounts.cf
, etc), to customize DMS to your needs.
Providing config files
Here is a minimal example that supplies a postfix-accounts.cf
file inline with two users:
---
apiVersion: v1
kind: ConfigMap
metadata:
name: mailserver.files
data:
postfix-accounts.cf: |
test@example.com|{SHA512-CRYPT}$6$someHashValueHere
other@example.com|{SHA512-CRYPT}$6$someOtherHashValueHere
Static Configuration
The inline postfix-accounts.cf
config example above provides file content that is static. It is mounted as read-only at runtime, thus cannot support modifications.
For production deployments, use persistent volumes instead (via PersistentVolumeClaim
). That will enable files like postfix-account.cf
to add and remove accounts, while also persisting those changes externally from the container.
Modularize your ConfigMap
Kustomize can be a useful tool as it supports creating a ConfigMap
from multiple files.
To persist data externally from the DMS container, configure a PersistentVolumeClaim
(PVC).
Make sure you have a storage system (like Longhorn, Rook, etc.) and that you choose the correct storageClassName
(according to your storage system).
Example
---
apiVersion: v1
kind: PersistentVolumeClaim
metadata:
name: data
spec:
storageClassName: local-path
accessModes:
- ReadWriteOnce
resources:
requests:
storage: 25Gi
A Service
is required for getting the traffic to the pod itself. It configures a load balancer with the ports you'll need.
The configuration for a Service
affects if the original IP from a connecting client is preserved (this is important). More about this further down below.
Example
---
apiVersion: v1
kind: Service
metadata:
name: mailserver
labels:
app: mailserver
spec:
type: LoadBalancer
selector:
app: mailserver
ports:
# smtp
- name: smtp
port: 25
targetPort: smtp
protocol: TCP
# submissions (ESMTP with implicit TLS)
- name: submission
port: 465
targetPort: submissions
protocol: TCP
# submission (ESMTP with explicit TLS)
- name: submission
port: 587
targetPort: submission
protocol: TCP
# imaps (implicit TLS)
- name: imaps
port: 993
targetPort: imaps
protocol: TCP
Using cert-manager
to supply TLS certificates
---
apiVersion: cert-manager.io/v1
kind: Certificate
metadata:
name: mail-tls-certificate-rsa
spec:
secretName: mail-tls-certificate-rsa
isCA: false
privateKey:
algorithm: RSA
encoding: PKCS1
size: 2048
dnsNames: [mail.example.com]
issuerRef:
name: mail-issuer
kind: Issuer
The TLS docs page provides guidance when it comes to certificates and transport layer security.
ECDSA + RSA (fallback)
You could supply RSA certificates as fallback certificates instead, with ECDSA as the primary. DMS supports dual certificates via the ENV SSL_ALT_CERT_PATH
and SSL_ALT_KEY_PATH
.
Always provide sensitive information via a Secret
For storing OpenDKIM keys, TLS certificates, or any sort of sensitive data - you should be using Secret
s.
A Secret
is similar to ConfigMap
, it can be used and mounted as a volume as demonstrated in the Deployment
manifest tab.
The Deployment
config is the most complex component.
- It instructs Kubernetes how to run the DMS container and how to apply your
ConfigMap
s, persisted storage, etc. - Additional options can be set to enforce runtime security.
Example
---
apiVersion: apps/v1
kind: Deployment
metadata:
name: mailserver
annotations:
ignore-check.kube-linter.io/run-as-non-root: >-
'mailserver' needs to run as root
ignore-check.kube-linter.io/privileged-ports: >-
'mailserver' needs privileged ports
ignore-check.kube-linter.io/no-read-only-root-fs: >-
There are too many files written to make the root FS read-only
spec:
replicas: 1
selector:
matchLabels:
app: mailserver
template:
metadata:
labels:
app: mailserver
annotations:
container.apparmor.security.beta.kubernetes.io/mailserver: runtime/default
spec:
hostname: mail
containers:
- name: mailserver
image: ghcr.io/docker-mailserver/docker-mailserver:latest
imagePullPolicy: IfNotPresent
securityContext:
# `allowPrivilegeEscalation: true` is required to support SGID via the `postdrop`
# executable in `/var/mail-state` for Postfix (maildrop + public dirs):
# https://github.com/docker-mailserver/docker-mailserver/pull/3625
allowPrivilegeEscalation: true
readOnlyRootFilesystem: false
runAsUser: 0
runAsGroup: 0
runAsNonRoot: false
privileged: false
capabilities:
add:
# file permission capabilities
- CHOWN
- FOWNER
- MKNOD
- SETGID
- SETUID
- DAC_OVERRIDE
# network capabilities
- NET_ADMIN # needed for F2B
- NET_RAW # needed for F2B
- NET_BIND_SERVICE
# miscellaneous capabilities
- SYS_CHROOT
- KILL
drop: [ALL]
seccompProfile:
type: RuntimeDefault
# Tune this to your needs.
# If you disable ClamAV, you can use less RAM and CPU.
# This becomes important in case you're low on resources
# and Kubernetes refuses to schedule new pods.
resources:
limits:
memory: 4Gi
cpu: 1500m
requests:
memory: 2Gi
cpu: 600m
volumeMounts:
- name: files
subPath: postfix-accounts.cf
mountPath: /tmp/docker-mailserver/postfix-accounts.cf
readOnly: true
# PVCs
- name: data
mountPath: /var/mail
subPath: data
readOnly: false
- name: data
mountPath: /var/mail-state
subPath: state
readOnly: false
- name: data
mountPath: /var/log/mail
subPath: log
readOnly: false
# certificates
- name: certificates-rsa
mountPath: /secrets/ssl/rsa/
readOnly: true
ports:
- name: smtp
containerPort: 25
protocol: TCP
- name: submissions
containerPort: 465
protocol: TCP
- name: submission
containerPort: 587
- name: imaps
containerPort: 993
protocol: TCP
envFrom:
- configMapRef:
name: mailserver.environment
restartPolicy: Always
volumes:
# configuration files
- name: files
configMap:
name: mailserver.files
# PVCs
- name: data
persistentVolumeClaim:
claimName: data
# certificates
- name: certificates-rsa
secret:
secretName: mail-tls-certificate-rsa
items:
- key: tls.key
path: tls.key
- key: tls.crt
path: tls.crt
Exposing your Mail Server to the Outside World
The more difficult part with Kubernetes is to expose a deployed DMS instance to the outside world.
The major problem with exposing DMS to the outside world in Kubernetes is to preserve the real client IP. The real client IP is required by DMS for performing IP-based DNS and spam checks.
Kubernetes provides multiple ways to address this; each has its upsides and downsides.
Advantages / Disadvantages
- Simple
- Requires the node to have a dedicated, publicly routable IP address
- Limited to a single node (associated to the dedicated IP address)
- Your deployment requires an explicit IP in your configuration (or an entire Load Balancer).
Requirements
- You can dedicate a publicly routable IP address for the DMS configured
Service
. - A dedicated IP is required to allow your mail server to have matching
A
andPTR
records (which other mail servers will use to verify trust when they receive mail sent from your DMS instance).
Example
Assign the DMS Service
an external IP directly, or delegate an LB to assign the IP on your behalf.
The DMS Service
is configured with an "external IP" manually. Append your externally reachable IP address to spec.externalIPs
.
---
apiVersion: v1
kind: Service
metadata:
name: mailserver
labels:
app: mailserver
spec:
selector:
app: mailserver
ports:
- name: smtp
port: 25
targetPort: smtp
# ...
externalIPs:
- 10.20.30.40
The config differs depending on your choice of load balancer. This example uses MetalLB.
---
apiVersion: v1
kind: Service
metadata:
name: mailserver
labels:
app: mailserver
annotations:
metallb.universe.tf/address-pool: mailserver
# ...
---
apiVersion: metallb.io/v1beta1
kind: IPAddressPool
metadata:
name: mail
namespace: metallb-system
spec:
addresses: [ <YOUR PUBLIC DEDICATED IP IN CIDR NOTATION> ]
autoAssign: true
---
apiVersion: metallb.io/v1beta1
kind: L2Advertisement
metadata:
name: mail
namespace: metallb-system
spec:
ipAddressPools: [ mailserver ]
Advantages / Disadvantages
- Simple
- Requires the node to have a dedicated, publicly routable IP address
- Limited to a single node (associated to the dedicated IP address)
- It is not possible to access DMS via other cluster nodes, only via the node that DMS was deployed on
- Every port within the container is exposed on the host side
Example
Using hostPort
and hostNetwork: true
is a similar approach to network_mode: host
with Docker Compose.
---
apiVersion: apps/v1
kind: Deployment
metadata:
name: mailserver
# ...
spec:
hostNetwork: true
# ...
containers:
# ...
ports:
- name: smtp
containerPort: 25
hostPort: 25
- name: submissions
containerPort: 465
hostPort: 465
- name: submission
containerPort: 587
hostPort: 587
- name: imaps
containerPort: 993
hostPort: 993
Advantages / Disadvantages
- Preserves the origin IP address of clients (which is crucial for DNS related checks)
- Aligns with a best practice for Kubernetes by using a dedicated ingress, routing external traffic to the k8s cluster (with the benefits of flexible routing rules)
- Avoids the restraint of a single node (as a workaround to preserve the original client IP)
- Introduces complexity by requiring:
- A reverse-proxy / ingress controller (potentially extra setup)
- Kubernetes manifest changes for the DMS configured
Service
- DMS configuration changes for Postfix and Dovecot
- To keep support for direct connections to DMS services internally within cluster, service ports must be "duplicated" to offer an alternative port for connections using PROXY protocol
- Custom Fail2Ban required: Because the traffic to DMS is now coming from the proxy, banning the origin IP address will have no effect; you'll need to implement a custom solution for your setup.
What is the PROXY protocol?
PROXY protocol is a network protocol for preserving a client’s IP address when the client’s TCP connection passes through a proxy.
It is a common feature supported among reverse-proxy services (NGINX, HAProxy, Traefik), which you may already have handling ingress traffic for your cluster.
flowchart LR
A(External Mail Server) -->|Incoming connection| B
subgraph cluster
B("Ingress Acting as a Proxy") -->|PROXY protocol connection| C(DMS)
end
For more information on the PROXY protocol, refer to our dedicated docs page on the topic.
Configure the Ingress Controller
On Traefik's side, the configuration is very simple.
- Create an entrypoint for each port that you want to expose (probably 25, 465, 587 and 993).
- Each entrypoint should configure an
IngressRouteTCP
that routes to the equivalent internal DMSService
port which supports PROXY protocol connections.
The below snippet demonstrates an example for two entrypoints, submissions
(port 465) and imaps
(port 993).
---
apiVersion: v1
kind: Service
metadata:
name: mailserver
spec:
# This an optimization to get rid of additional routing steps.
# Previously "type: LoadBalancer"
type: ClusterIP
---
apiVersion: traefik.io/v1alpha1
kind: IngressRouteTCP
metadata:
name: smtp
spec:
entryPoints: [ submissions ]
routes:
- match: HostSNI(`*`)
services:
- name: mailserver
namespace: mail
port: subs-proxy # note the 15 character limit here
proxyProtocol:
version: 2
---
apiVersion: traefik.io/v1alpha1
kind: IngressRouteTCP
metadata:
name: imaps
spec:
entryPoints: [ imaps ]
routes:
- match: HostSNI(`*`)
services:
- name: mailserver
namespace: mail
port: imaps-proxy
proxyProtocol:
version: 2
*-proxy
port name suffix
The IngressRouteTCP
example configs above reference ports with a *-proxy
suffix.
- These port variants will be defined in the
Deployment
manifest, and are scoped to themailserver
service (viaspec.routes.services.name
). - The suffix is used to distinguish that these ports are only compatible with connections using the PROXY protocol, which is what your ingress controller should be managing for you by adding the correct PROXY protocol headers to TCP connections it routes to DMS.
With an NGINX ingress controller, add the following to the TCP services config map (as described here):
25: "mailserver/mailserver:25::PROXY"
465: "mailserver/mailserver:465::PROXY"
587: "mailserver/mailserver:587::PROXY"
993: "mailserver/mailserver:993::PROXY"
Adjust DMS config for Dovecot + Postfix
Only ingress should connect to DMS with PROXY protocol
While Dovecot will restrict connections via PROXY protocol to only clients trusted configured via haproxy_trusted_networks
, Postfix does not have an equivalent setting. Public clients should always route through ingress to establish a PROXY protocol connection.
You are responsible for properly managing traffic inside your cluster and to ensure that only trustworthy entities can connect to the designated PROXY protocol ports.
With Kubernetes, this is usually the task of the CNI (container network interface).
Advised approach
The "Separate PROXY protocol ports" tab below introduces a little more complexity, but provides better compatibility for internal connections to DMS.
Connections to DMS within the internal cluster will be rejected
The services for these ports can only enable PROXY protocol support by mandating the protocol on all connections for these ports.
This can be problematic when you also need to support internal cluster traffic directly to DMS (instead of routing indirectly through the ingress controller).
Here is an example configuration for Postfix, Dovecot, and the required adjustments for the Deployment
manifest. The port names are adjusted here only to convey the additional context described earlier.
kind: ConfigMap
apiVersion: v1
metadata:
name: mailserver-extra-config
labels:
app: mailserver
data:
postfix-main.cf: |
postscreen_upstream_proxy_protocol = haproxy
postfix-master.cf: |
smtp/inet/postscreen_upstream_proxy_protocol=haproxy
submission/inet/smtpd_upstream_proxy_protocol=haproxy
submissions/inet/smtpd_upstream_proxy_protocol=haproxy
dovecot.cf: |
haproxy_trusted_networks = <YOUR POD CIDR>
service imap-login {
inet_listener imap {
haproxy = yes
}
inet_listener imaps {
haproxy = yes
}
}
# ...
---
kind: Deployment
apiVersion: apps/v1
metadata:
name: mailserver
spec:
template:
spec:
containers:
- name: docker-mailserver
# ...
ports:
- name: smtp-proxy
containerPort: 25
protocol: TCP
- name: imap-proxy
containerPort: 143
protocol: TCP
- name: subs-proxy
containerPort: 465
protocol: TCP
- name: sub-proxy
containerPort: 587
protocol: TCP
- name: imaps-proxy
containerPort: 993
protocol: TCP
# ...
volumeMounts:
- name: config
subPath: postfix-main.cf
mountPath: /tmp/docker-mailserver/postfix-main.cf
readOnly: true
- name: config
subPath: postfix-master.cf
mountPath: /tmp/docker-mailserver/postfix-master.cf
readOnly: true
- name: config
subPath: dovecot.cf
mountPath: /tmp/docker-mailserver/dovecot.cf
readOnly: true
Info
Supporting internal cluster connections to DMS without using PROXY protocol requires both Postfix and Dovecot to be configured with alternative ports for each service port (which only differ by enforcing PROXY protocol connections).
- The ingress controller will route public connections to the internal alternative ports for DMS (
*-proxy
variants). - Internal cluster connections will instead use the original ports configured for the DMS container directly (which are private to the cluster network).
In this example we'll create a copy of the original service ports with PROXY protocol enabled, and increment the port number assigned by 10000
.
Create a user-patches.sh
file to apply these config changes during container startup:
#!/bin/bash
# Duplicate the config for the submission(s) service ports (587 / 465) with adjustments for the PROXY ports (10587 / 10465) and `syslog_name` setting:
postconf -Mf submission/inet | sed -e s/^submission/10587/ -e 's/submission/submission-proxyprotocol/' >> /etc/postfix/master.cf
postconf -Mf submissions/inet | sed -e s/^submissions/10465/ -e 's/submissions/submissions-proxyprotocol/' >> /etc/postfix/master.cf
# Enable PROXY Protocol support for these new service variants:
postconf -P 10587/inet/smtpd_upstream_proxy_protocol=haproxy
postconf -P 10465/inet/smtpd_upstream_proxy_protocol=haproxy
# Create a variant for port 25 too (NOTE: Port 10025 is already assigned in DMS to Amavis):
postconf -Mf smtp/inet | sed -e s/^smtp/12525/ >> /etc/postfix/master.cf
# Enable PROXY Protocol support (different setting as port 25 is handled via postscreen), optionally configure a `syslog_name` to distinguish in logs:
postconf -P 12525/inet/postscreen_upstream_proxy_protocol=haproxy 12525/inet/syslog_name=smtp-proxyprotocol
For Dovecot, you can configure dovecot.cf
to look like this:
haproxy_trusted_networks = <YOUR POD CIDR>
service imap-login {
inet_listener imap-proxied {
haproxy = yes
port = 10143
}
inet_listener imaps-proxied {
haproxy = yes
port = 10993
ssl = yes
}
}
Update the Deployment
manifest ports
section by appending these new ports:
- name: smtp-proxy
# not 10025 in this example due to a possible clash with Amavis
containerPort: 12525
protocol: TCP
- name: imap-proxy
containerPort: 10143
protocol: TCP
- name: subs-proxy
containerPort: 10465
protocol: TCP
- name: sub-proxy
containerPort: 10587
protocol: TCP
- name: imaps-proxy
containerPort: 10993
protocol: TCP
Note
If you use other Dovecot ports (110, 995, 4190), you may want to configure those similar to above. The dovecot.cf
config for these ports is documented here (in the equivalent section of that page).