504 Gateway Time-out

[mortalys]

Hi everyone,

I been exploring APISIX and finally was able to get it working with SSL and domain gateway.api.domain.com - I replace the domain just for the privacy -regarding the APISIX setup is the base example for version 3.7 using Docker ( Docker Compose )

I have a somehow complex setup, since APISIX is installed in a internal network, so there is a domain resolver before reach it, but was able to made.

Now I was trying to setup a route and the example that is common ( httpbin.org ) works just fine. I keep gateway.api.domain.com and access the content.

When I try to change the Upstream target, which is locally inside of the internal network with the domain api.domain.com and I get an error "504 Gateway Time-out"

I been trying different solutions that came to my mind, for example, maybe APISIX required connection on ports which are blocked on the router (443) and only allowed for the other particular domain resolver machine, also added different ports 9443, 8443, 443 between domain resolver, apisix.

With this I added the host directly in /etc/hosts and when I try to connect curl/ping it does it so.

ping api.domain.com
pingapi.domain.com (192.168.1.101) 56(84) bytes of data.
64 bytes from *.domain.com (192.168.1.101): icmp_seq=1 ttl=64 time=0.308 ms
64 bytes from *.domain.com (192.168.1.101): icmp_seq=2 ttl=64 time=0.279 ms

curl https://api.domain.com:443 -k -vvv

• Trying 192.168.1.101:443...
• Connected to api.domain.com (192.168.1.101) port 443 (#0)

---

checking APISIX logs I also have this error ( doesn't matter if it's the local network ip or public ip ) client is the the original domain resolver, so there is a loop , which is:

• 2.82.28.xxx is public ip
• 192.168.1.101 then domain resolver which has the api.domain.com (upstream that I want to access from apisix)
• 192.168.1.140 apisix requests to 2.82.28.xxx is public ip

2023/12/09 11:20:39 [error] 50#50: *6081 upstream timed out (110: Connection timed out) while reading response header from upstream, client: 192.168.1.101, server: _, request: "GET / HTTP/1.0", upstream: "https://2.82.28.xxx:443/", host: "gateway.api.domain.com"

Any ideas? Thank you!

[mortalys]

I have an update on the case

When I change to Upstream type to Use the domain or IP from Node List now I reach the document root of https://gateway.api.domain.com/ not the api.domain.com

Maybe is some nginx configuration on the domain resolver?