Hello RIPE Community,
I am currently dealing with one specific networking problem affecting my website where visitors using IPv6 connectivity occasionally experience connection timeouts or very slow page loading, while the same website generally works normally when accessed over IPv4. The website is hosted on a server that has both IPv4 and IPv6 connectivity configured, and the domain has DNS records for both address families. Most visitors can access the website without noticing anything unusual, but a portion of requests appear to attempt an IPv6 connection and then fail to establish a reliable connection within the expected time. When the same hostname is accessed from an environment that uses IPv4, the website responds normally. The issue is intermittent from my perspective because I do not see a complete outage, but individual users can experience a page that hangs during the initial connection and eventually reports a timeout. I am trying to investigate this as one specific IPv6 reachability problem rather than several unrelated website or server issues.
I have verified that the website’s IPv6 address is published in the DNS configuration and that the address is intended to point to the same server hosting the website. The server itself has IPv6 enabled, and the web service is configured to accept connections for the website. When I test the website through IPv4, the expected response is returned consistently, while IPv6 testing can produce a different result depending on the network from which the request originates. I have been comparing DNS resolution and connection behaviour from multiple networks so I can determine whether the issue is limited to a particular upstream provider or whether there is a more general routing or reachability problem. The important clue is that the website remains available over IPv4 during the periods when IPv6 users report the timeout, so the application itself does not appear to be completely unavailable.
I have also started checking the network path to the server rather than focusing only on the web application. I am comparing IPv4 and IPv6 traceroute results and checking whether packets reach the server’s network as expected. In some tests, IPv6 appears to resolve correctly but the connection does not complete normally, which makes me suspect that there could be a routing, filtering, or connectivity issue somewhere between the visitor’s network and the server. I am particularly interested in understanding whether the advertised IPv6 prefix is being propagated correctly and whether there could be an issue affecting only certain networks or upstream paths. I have not made any assumptions about where the failure occurs because I want to distinguish between a local server configuration issue and a broader routing problem. I am collecting timestamps and test locations so that any comparison can be tied to an actual period when the website was experiencing the problem.
The problem is difficult to reproduce consistently because an IPv6 connection can work successfully from one network while timing out from another network around the same period. This makes a simple web-server configuration error less obvious, although I am still checking the server’s IPv6 listener, firewall rules, and interface configuration. I am also checking whether the server is receiving the connection attempts when an affected visitor reports a timeout. If the connection attempt never reaches the server, that would point toward a routing or upstream connectivity problem, whereas a connection reaching the server but failing to establish the HTTP session would suggest that I need to investigate the server-side configuration more closely. I am therefore trying to gather enough network-level evidence to identify where the IPv6 path breaks instead of simply disabling the AAAA record and forcing everyone to use IPv4.
I have started collecting basic diagnostic information from both successful and unsuccessful IPv6 tests, including DNS results, the resolved IPv6 address, connection timing, traceroute output, and the approximate time of each test. I am also comparing results from different networks because the behaviour appears to vary depending on the source network. The website’s IPv4 connectivity continues to work during these tests, which gives me a useful baseline for comparison. I would like to understand which RIPE tools or publicly available measurement resources could help determine whether my IPv6 prefix is being announced and routed correctly, and whether there are known ways to identify a path-specific IPv6 reachability problem. I can provide sanitized addressing and routing information if needed, but I want to avoid posting unnecessary sensitive configuration details publicly.
I would appreciate guidance from the RIPE community on how to systematically troubleshoot this specific problem where my dual-stack website is consistently reachable over IPv4 but can intermittently become unreachable or time out for some users connecting over IPv6. In particular, I would like advice on which BGP, routing, IPv6 reachability, traceroute, DNS, or server-side checks would be most useful for determining where the connection is failing. I would also like to know how to distinguish a problem with my server’s IPv6 configuration from an upstream routing or prefix-advertisement issue affecting only certain networks. My goal is to identify and correct the underlying IPv6 connectivity problem so that visitors using IPv6 receive the same reliable access to the website as visitors using IPv4, rather than simply removing IPv6 support as a workaround. Sorry for long post!