A useful clue in the original case
A report involving AIO and Caddy includes connect: connection refused on internal port 11000 at a loopback address. Q1 Official AIO guidance requires the proxy destination and APACHE_PORT to match the deployment. It distinguishes a proxy running directly on the host from one running inside a container. Q2
From a separate proxy container, localhost refers to that container's own environment. Correct public DNS therefore does not prove that the proxy can reach the intended AIO upstream.
Inspect each stage of the route
- Document the deployment model. Record where AIO and the proxy run, which container networks they use and the intended Apache port. Follow the AIO documentation variant matching that arrangement.
- Compare domain resolution. Check A and, where present, AAAA responses from a suitable external network. Do they identify the intended ingress? An outdated IPv6 address may affect a route different from a successful IPv4 test.
- Inspect public ingress. Does HTTPS for this exact domain reach the intended proxy with the expected certificate? Record hostname, response status and log timestamp. An unrelated proxy landing page is not the same route.
- Test from the proxy's perspective. In the actual proxy environment, check destination name, address and port. For connection refused, establish whether the expected service listens there and whether the network path is correct. Compare the destination with
APACHE_PORTand its configured binding. - Observe the AIO check again. After a justified correction, repeat the supported validation and correlate proxy and AIO logs. If it still fails, the next unsuccessful stage should now be more clearly identified.
Avoid a blanket bypass
Disabling domain validation does not repair an unreachable upstream. Special AIO deployment modes have their own documented requirements; do not copy an exception from a different setup. Expose internal ports only through the intended proxy path instead of publishing them as a general troubleshooting measure.
Retain the final network mapping and a successful external HTTPS check. They provide a useful reference when containers or proxy configuration are later replaced.
Frequently asked questions
Do I need to change DNS records?
Only if resolution points to the wrong ingress. A connection failure between the proxy container and AIO is not explained by assigning a different public IP.
What helps with a handover?
Provide AIO and proxy versions, an anonymised network diagram, DNS results, APACHE_PORT, upstream destination and correlated logs. Exclude secrets from environment variables and certificate files.
Help with this case
We can help isolate the affected process and prepare a controlled next step. Discuss case A0291 with KAN KAI WORKS.
Prepare a findings note
Summarise your observations for your enquiry. The note is created in your browser and is not saved or sent here. You can then copy it into the contact form.
Unknown is fine. Do not enter passwords, customer data or complete confidential logs.
Support with this issue
We investigate Nextcloud using your error message, versions and a narrowly scoped example.
Sources
Sources checked on 5 November 2026. Individual reports do not establish general prevalence.
