I've been pulling my hair out over this! We recently migrated our company website to a new host, and over a week ago, I updated the A records and CNAME at our registrar to point to the new server IP. Despite global propagation checkers showing everything is set correctly, around 2% of our clients are getting a 'Page not found' error. When I check services like whatsmydns.net or DNSChecker, every location indicates that the new IP address is properly set, and they're showing all green checks.
Here's what I've tried so far to troubleshoot:
1. I asked clients to clear their browser cache (Ctrl+F5) but it didn't help.
2. One client ran nslookup and was still getting the old IP.
3. I lowered the TTL (Time To Live) to 300 seconds before the switch specifically to avoid issues like this.
4. The old host has been completely shut down, so accessing it leads to a dead end.
Is it possible that their local ISP DNS is caching the records for more than a week? That seems wild. What can I do to resolve this issue now, and how can I prevent this problem in the future?
4 Answers
It sounds like there might be a local DNS issue on the client's end, or perhaps they’re hitting an outdated DNS server. You could ask them to run an nslookup to see which DNS server they're using. If they’re using something like Rogers in Canada, they might be stuck with old cache. It could also be worth tracing the DNS to see if the old host is still in play somewhere.
Make sure to check what the TTL was set at before you lowered it. If it was something like a week, it might take that long for everyone to see the new IP. In the future, try not to switch off the old host until you're certain the new settings are being picked up everywhere.
Yeah, it’s usually smart to leave the old site running a bit longer just to avoid these issues. Waiting out the original TTL makes a lot of sense.
It's also possible there's an internal DNS conflict or a host file override on their devices, especially if they’re part of a corporate network. Have you checked that they don't have any custom DNS settings causing the issue?
Good point! Sometimes companies set up local DNS servers that can mess with resolution if not properly configured.
Have you tried flushing the DNS cache on the client machines? They can do this using the command `ipconfig /flushdns`. Sometimes local caches can cause issues even after changes are made.
I suggested that to one client and it actually worked for them! It’s a quick fix for some, but definitely not universal.
Right, and if they discover that the clients are using an outdated DNS, they might want to switch to a public DNS, like Google's (8.8.8.8), which should help.