- Using custom TLS certificates for model providers (such as Azure, OpenAI, or a custom model server)
- Mounting internal certificate authorities (CAs) to enable TLS connections for databases and other external services.
Use custom TLS certificates for model providers
This feature is currently only available for the following model providers:
- Azure OpenAI
- OpenAI
- Custom (our custom model server). Refer to the custom model server documentation for more information.
LANGSMITH_PLAYGROUND_TLS_MODEL_PROVIDERS
: A comma-separated list of model providers that require custom TLS certificates. Note thatazure_openai
,openai
andcustom
are currently the only supported model provider that supports custom TLS certificates, but more providers will be supported in the future.LANGSMITH_PLAYGROUND_TLS_CA
: The custom certificate authority (CA) certificate in PEM format. This must be a file path (for a mounted volume).- [Optional]
LANGSMITH_PLAYGROUND_TLS_KEY
: The private key in PEM format. This must be a file path (for a mounted volume). This is usually only necessary for mutual TLS authentication. - [Optional]
LANGSMITH_PLAYGROUND_TLS_CERT
: The certificate in PEM format. This must be a file path (for a mounted volume). This is usually only necessary for mutual TLS authentication.
Mount internal CAs for TLS
- Create a file containing all CAs required for TLS with databases and external services. If your deployment is communicating directly to
beacon.langchain.com
without a proxy, make sure to include a public trusted CA. All certs should be concatenated in this file with an empty line in between.
- Create a Kubernetes secret with a key containing the contents of this file.
- If using custom CA for TLS with your databases and other external services, provide the following values to your LangSmith helm chart:
Helm
- Make sure to use TLS supported connection strings:
- Postgres: Add
?sslmode=verify-full&sslrootcert=system
to the end. - Redis: Use
rediss://
instead ofredis://
as the prefix.
- Postgres: Add