IPv6 in Shared Hosting Logs

Over the last several days, I’ve noticed IPv6 addresses in the daily access log for an IPv4 website on Shared Hosting.

I’ve thought I would only see IPv6 visits logged if my own site was condig’d for IPv6, like a dedicated IP address.

So I am assuming these IPv6 visits are from ranges registered post IPv4 capacity.

This tells me that if a website is IPv4, DH Shared Hosting server config will first check to see if a visitor has an IPv4 address, if not the IPv6 address will be logged.

HTTP servers don’t haggle with the client about preferred IP versions* – they just ACK the originating address and log the associated IP.

* That would be bigoted. Or something.

Think you missed my point.

I didn’t intend to imply the server negociated a preferred IP address, only a reason for which version is logged… since many clients come from servers which have both IPv4 and IPv6 assignments and some from servers with just IPv6.

The ball is in the client’s court as it decides which protocol to use when initiating the connection before sending a SYN packet. Our server has no say in the matter as it is not the initiator. One thing of note is that although we call them “versions” they are actually entirely different protocols and are not interchangeable on-the-fly.

When I send a GET request from my AWS account to my DH site, it logs as IPv4.

When I send the same request to my site at MIT, it logs as IPv6. The MIT IP address is IPv6.

A site here at DH I worked on a few months ago uses a dedicated IP address, and the logs show most all activity in IPv6, because the server is IPv6.

My shared server account here uses an IPv4 address, and almost all activity is logged as IPv4.

So the question remains why a few requests were recently logged as IPv6. The only reason seems to be that these came from newly registered addresses that do not also have an IPv4.

It would be because your AWS account is capable of communicating over both IPv4 and IPv6 protocols. When your GET request is sent out your AWS account is pre-selecting IPv4 to talk to DreamHost and IPv6 to talk to MIT. The AWS account is pre-selecting which protocol to use depending on the response received from the initial DNS lookup (v4 for DH, v6 for MIT).

Traffic appearing over IPv6 on a server that is also capable of responding of IPv4 suggests that the server is IPv6 aware (so-called Dual-Stack) but the logged client is sending requests using the IPv6 protocol. This might be due to the client not being IPv4 capable or that it is otherwise set up to prefer the IPv6 protocol.

For example, you can force your local Windows machine to connect over IPv6 only by opening your Network Settings and altering the Properties of your connection – toggling off the Internet Protocol Version 4 switch.

This topic was automatically closed 60 minutes after the last reply. New replies are no longer allowed.