zhaofeng0019 / nginx-upstream-dynamic-resolve-servers Goto Github PK

Imagine this simple configuration:

upstream upstream {
    server example.org resolve;
    server example.com resolve;
    server example.org resolve;
}

When just first upstream contains resolve param, everything works fine. But when multiple upstream servers has resolve param, nginx worker process fail with error in log like this:

socket() failed (97: Address family not supported by protocol) while connecting to upstream, client: 127.0.0.1, server: localhost, request: "GET / HTTP/1.1", upstream: "https://J?./", host: "localhost"

followed by worker crash. I will try to debug

Hi,
We're using dynamic_resolve directive from tengine, but are analyzing other options, since that directive resolves DNS in the request path and it should be better to have it asynchronously update, not blocking any request.

I have some questions regarding this module:

• Why is ssl session resumption disabled, when tengine's directive has it enabled?
• According to this if dns has just refreshed and a request is being processed the request would be failed, is that correct? Isn't there a high chance of this happening if you have a lot of traffic and DNS refresh quickly (60s)?
• Is there any way this module is different that tengine's?

nginx的stream块下也有upstream的设置，能否对其进行适配呢

Hi,
Is there any limitation to use this module?
I have tested if there are a lot of domain name use "resolve", the server will return "(52) Empty reply from server" or "(35) LibreSSL SSL_connect: SSL_ERROR_SYSCALL in connection to xxx" in many requests.