Continue connection to a local server fails in https

[YanickNoblanc]

Hi all

I have setup a local llm on a local server, it is running a vllm (0.11.0) with several models available, an open-webui ( v0.6.40) and a nginx proxy (1.29.3).

I can acces properly my llm through open-webui chat interface at https://server or http://server
When configuring Continue VSCode plugin (on Windows) to acces my server through http, it works fine in agent or chat mode and answers my questions.

But when trying to access server in https, I systematicaly get a connection error, Developpers Tools logs:
[@continuedev] error: Connection error. {"context":"llm_stream_chat","model":"/models/Llama-3.1-8B-Instruct","provider":"openai","useOpenAIAdapter":true,"streamEnabled":true,"templateMessages":false}
I checked the server configuration and certificates are correct. Certificate is signed by our internal CA.
Here is Continue config.yaml I am using:

name: Local Config
version: 1.0.0
schema: v1
models:
  - name: Llama-3.1-8B-Instruct
    provider: openai
    model: /models/Llama-3.1-8B-Instruct
    env:
      useLegacyCompletionsEndpoint: false
    apiBase: http://server:8443/v1
    ##TODO https does not work !
    apiKey: AAAAAAAAAAAAAAAAA
    roles:
      - chat
      - edit
      - apply

What should I do for Continue to work with https ?

[continue]

Related issues about HTTPS/SSL certificate problems with local servers:

Self-signed certificate issues:

• VScode SSL self signed issue  #1474 - VScode SSL self signed issue (verifySsl not working with self-signed certs)
• caBundlePath and verifySsl ignored for openai model #889 - caBundlePath and verifySsl ignored for openai model
• need to disable ssl verification #4310 - need to disable ssl verification (Ollama with nginx proxy)
• Trusted self-signed certificate works for chatting but not for auto-completion #5115 - Trusted self-signed certificate works for chatting but not for auto-completion
• @Jira context: error "self signed certificate in certificate chain" - lost of global requestOptions used in fetch with YAML config #5484 - @jira context: error "self signed certificate in certificate chain" - lost of global requestOptions used in fetch with YAML config

Certificate verification issues:

• [CON-144] OpenAIConnectionError probably caused by failed SSL certificates #218 - OpenAIConnectionError probably caused by failed SSL certificates
• Unable to get local issuer certificate failure #5578 - Unable to get local issuer certificate failure

Related to open-webui/vllm with HTTPS:

• Can't connect to open-webui API #3957 - Can't connect to open-webui API (testing with nginx server and localhost)

These issues all relate to SSL/TLS certificate verification problems when connecting to local or self-hosted servers with HTTPS, particularly with self-signed or internal CA certificates.

[YanickNoblanc]

Thank you very much for all the information.

[YanickNoblanc]

Unfortunatly does not give me a clear answar to my problem. The certificate on my server is not self signed. I tried adding requestOptions verifySsl: false and caBundlePath:... : it does not solve my issue. Still getting same communication error.

[github-actions]

This issue hasn't been updated in 90 days and will be closed after an additional 10 days without activity. If it's still important, please leave a comment and share any new information that would help us address the issue.