Using Proxies

Solace Cloud supports deployments using http, https, and SOCKS5 proxies. You can use proxies to regulate, or exclude your event broker service's access to the internet, or you may require proxies to meet security requirements. Solace supports proxies in most situations and works with you to ensure that your security requirements are met with the proxy.

Solace can configure your deployment to use proxies for all of your connectivity requirements; however, there are some limitations you must consider when using proxies for Operational Connectivity between the Home Cloud and Mission Control Agent, which can have up to five main connections as outlined in Connectivity Model for Kubernetes Deployments.

In environments where you want to restrict access to the internet, you can configure a proxy with the http_proxy, https_proxy, or all_proxy (for SOCKS5) properties of your Kubernetes cluster.

In simple terms, using proxies maintains security while providing convenience. Using a proxy allows you to:

  • control your event broker service's traffic.
  • hide the IP addresses of your event broker service and achieve online anonymity.
  • work with the network restrictions of your environment.

See the Use Cases for some examples of why you may want to use a proxy.

If you choose to use a proxy, you must let Solace know at the time of deployment. You must provide the host, port, protocol, username, and password information for your proxy server. We use this information to configure the <datacenter-name>-values.yaml file with the appropriate parameters for the Helm Chart to deploy the Mission Control Agent. If you intend to use distributed tracing, you can supply a list of proxy exclusions to selectively bypass proxy routing for specific destinations like a self-hosted observability platform.

The example below shows components that may be required for a proxy:

proxy:
  ivmr:
    enabled: false
  distributedTracing:
    noProxy:
      enabled: false
      hosts: []
  host: ""         # must be provided
  protocol: "http" # options are http, https, or socks5
  port: ""
  username: ""     # must be provided
  password: ""     # must be provided

As an example, a customer whose proxy is at proxyserver.customer.com:3341 with a username user1234 and password 3444password3232 would have the following entry in the values.yaml file.

proxy:
  ivmr:
    enabled: false
  distributedTracing:
    noProxy:
      enabled: false
      hosts: []
  host: "proxyserver.customer.com:3341"
  protocol: "http" 
  port: ""
  username: "user1234"
  password: "3444password3232"

Considerations When Using a Proxy for Operational Connectivity Between Mission Control Agent and Home Cloud

If your security requires, Solace can configure the Operational Connectivity between the Mission Control Agent and Solace Home Cloud to run through a proxy. Operational Connectivity carries the communications required to install the Mission Control Agent and to install, operate, and monitor event broker services. For more information, see Connectivity Model for Kubernetes Deployments. The proxy settings are also used for connections through Solace Cloud between event broker services and Event Portal. For more information, see Connecting Event Broker Services to Event Portal Through Solace Cloud.

We support authenticated and non-authenticated HTTP and SOCKS5 proxies. While we tested HTTP proxy support for Operational Connectivity using Squid and Zscaler, any proxy that meets the following requirements should work:

  • For HTTP/HTTPS proxies, your proxy must support:

    • HTTP CONNECT

    • proxying to port 55443 (default) or port 443 (when using WebSocket transport)

  • For SOCKS5 proxies, your proxy must support the SOCKS5 protocol for TCP tunneling to port 55443 (default) or port 443 (when using WebSocket transport).

  • If your proxy performs TLS inspection (also known as SSL inspection), you can either disable inspection for the Mission Control Agent to Solace Home Cloud connection, or use WebSocket transport on port 443, which is compatible with TLS-inspecting proxies. For more information, see Using WebSocket Transport for TLS-Inspecting Proxies.

Using WebSocket Transport for TLS-Inspecting Proxies

The Mission Control Agent can connect to Solace Home Cloud using SMF over secure WebSocket on port 443, as an alternative to using TCP on port 55443. This enables deployments behind TLS-inspecting proxies (such as Squid, Zscaler, or Netskope) to operate without requiring a security-policy exception for port 55443 or exempting the connection from inspection. TCPS on port 55443 remains the default transport.

Enabling WebSocket Transport

To enable WebSocket transport, Solace configures the following settings in your private region configuration:

  • IVMR Transport — WebSocket over TLS (port 443). The IVMR URL changes to use the WebSocket host, for example wss:production-ivmr-ws.messaging.solace.cloud:443. For a list of URLs for Solace Home Cloud regional sites, see Connection Requirements for Solace Cloud Components

  • CA Secret name — (Optional) The name of a Kubernetes Secret containing your proxy's CA certificate. This is only required when your proxy re-signs TLS traffic.

The downloaded Helm values contain the following configuration:

vmr:
  host: wss:<env>-ivmr-ws.messaging.solace.cloud:443
  transport: websocket
  caBundle:
    secretName: <secret-name>   # only present if configured

Creating the CA Secret for TLS-Inspecting Proxies

If your proxy re-signs your TLS traffic, you must create a Kubernetes secret containing your proxy's Certificate Authority (CA) before installing or upgrading the Mission Control Agent. You must place the secret in the Mission Control Agent namespace.

To create the CA secret, perform these steps:

  1. Obtain your proxy's CA certificate in PEM format. A bundle containing multiple certificates is supported.

  2. Create the secret in the solace-cloud namespace using the following command:

    kubectl -n solace-cloud create secret generic <secret-name> --from-file=ca.pem
  3. Install or upgrade the Mission Control Agent using Helm values downloaded from the Private Region tab of the Cloud Console.

Network Requirements for WebSocket Transport

When using WebSocket transport, your network must meet the following requirements:

  • Outbound HTTPS (port 443) to <env>-ivmr-ws.messaging.solace.cloud (or a regional equivalent), either directly or through the configured proxy.

  • The proxy must allow WebSocket upgrades (Upgrade: websocket) to Solace Home Cloud.

Some proxies are case-sensitive when matching the WebSocket Upgrade header. If the proxy expects an exact match, ensure it is configured to accept websocket (in lowercase).

WebSocket Transport Behaviors

The following table describes key behaviors when using WebSocket transport.

Behavior Description

CA trust

Public CAs remain trusted. The proxy CA certificate is added to the trust store. It is not used instead of public CAs.

CA rotation

After updating the CA secret, you must restart the Mission Control Agent pod for the changes to take effect.

Mismatched settings

The installation fails if the transport mode and host URL do not match. For example, specifying websocket transport with a tcps: host URL causes an error.

Missing CA Secret

If the configured CA secret does not exist, the Mission Control Agent pod remains in the ContainerCreating state.

Troubleshooting WebSocket Transport Connections

If the Mission Control Agent cannot connect to Solace Home Cloud at startup, it logs an error message identifying the cause:

IVMR connection failed: <CATEGORY> - <hint>

The following table describes the error categories and their meanings.

Category Meaning

AUTHENTICATION / VPN_NOT_ALLOWED

The credentials or the Message VPN were rejected.

TLS_UNTRUSTED_CERTIFICATE

The proxy CA certificate that should be contained in the secret, is either missing or incorrect.

TLS_HOSTNAME_MISMATCH

The host is not covered by the certificate.

PROXY_REJECTED

The proxy refused the connection or requires credentials.

DNS / NETWORK_TIMEOUT / NETWORK_REFUSED

There is a name resolution, firewall, or routing problem.

CLOSED_DURING_CONNECT / TLS_HANDSHAKE

The TLS negotiation was blocked or closed by a proxy or firewall.

On a successful connection, the Mission Control Agent logs the following message:

IVMR connected to <host> over wss

Configuring Proxy Exclusions for Distributed Tracing

Distributed tracing supports only HTTP or HTTPS proxies.

If you are using distributed tracing, and an HTTP or HTTPS proxy, Solace can configure a NO_PROXY exclusion list to selectively bypass proxy routing for specific destinations while maintaining proxy routing for all other traffic. Proxy exclusions are useful when you need to send traces to internal or on-premises systems that should not route through your corporate proxy, while still routing external services through a proxy for security compliance.

For more information, see Using Distributed Tracing with a Proxy in a Customer-Controlled Cluster.

Use Cases

There are numerous reasons you may decide to use a proxy. For deployments inside a private region, using a proxy is a simple method of controlling your services access to regions outside of your private region. The following use cases provide some examples where using a proxy can help manage event broker service access to the public internet:

Secure Access to Public Internet

Perhaps you decide to permit access to the public internet for the event broker services for outbound traffic. A proxy might be useful for outbound connections, such as RDPs (REST Destination Points). The event broker service can be configured to originate from a static IP address. The proxy makes it easier for applications outside of your deployment to whitelist. Proxies are useful in situations where you don't use a NAT or where you don't have a static IP address.

Managing Connections to Outside Services

For deployments inside Customer-Controlled Clusters, you can use a proxy as a method of controlling and managing access to services outside the Customer-Controlled Cluster with hostnames instead of IP addresses. For example, you may have a requirement for the Mission Control Agent to contact Datadog, which handles the monitoring data for event broker services. Because Datadog doesn't use static IP addresses, you can use a proxy to streamline the maintenance of connecting to Datadog and have the proxy determine whether a connection between the Mission Control Agent and Datadog is allowed using a hostname instead of IP addresses.