Hello,

I have a problem with upstream in case of upstream error.

Here an exemple of my configuration:

upstream my_backend_list {
server 192.9.211.82:9001;
server 192.9.211.82:9002;
}

server {
location / {
proxy_pass http://my_backend_list;
}
}

My 2 backends are offline.
When I send a request, I have the expected result:
$status: 502
$upstream_addr: 192.9.211.82:9001, 192.9.211.82:9002
$upstream_status: 502, 502

But, if I try several requests at a time, some results (50%) are unexpected:
$status: 502
$upstream_addr: my_backend_list
$upstream_status: 502

First, I don't want "my_backend_list" string appears in my logs and my servers statistics.
Then, I'am afraid the backends are not contacted in this cases.

Here some variables values:
request->upstream_states->nelts == 1
request->upstream_states->elts[0].peer->data == "my_backend_list"

Is there a way to avoid this behaviour ?
Thank you in advance.

Traquila

Hello, I didn't found any solution for my problem.

Thank in advance.