I just recently finished working on a big migration project recently where we are switching from one DNS provider to another. Since it involved a large number of URLs, I needed to make sure once the DNS is switched, all URLs would still work as expected.

To ensure the migration went smoothly, I implemented a few tests, including a crawl script to check the HTTP status codes of the most visited URLs.

In cURL, this is achieved by specifying the --resolve flag in the format of <host>:<port>:<ip>. In PHP, you can achieve the same thing by setting the CURLOPT_RESOLVE option via curl_setopt, using the same format. In my script, I am looping a set of URLs by letting cURL to resolve it automatically to check the current HTTP status code, then do another check by overriding the host to a different IP.

Only gotcha that I found when writing the script was that even if you instantiate different cURL sessions via curl_init, once they are added to the same cURL multi handle created with curl_multi_init, the host resolution doesn’t always work as expected.

Maybe I did something wrong, but I already added a few extra precautions just to be safe like this:

<?php

$ch = curl_init();

curl_setopt( $ch, CURLOPT_DNS_CACHE_TIMEOUT, 0 );
curl_setopt( $ch, CURLOPT_FORBID_REUSE, true );
curl_setopt( $ch, CURLOPT_FRESH_CONNECT, true );