IPv6 prefix changed, took 20 minutes to update everything

Provider: KnownHost
Old prefix: 2001:db8:aaaa::/48
New prefix: 2001:db8:bbbb::/48
Notification: 7 days advance
Actual change: 03:00 UTC

Went down 0 minutes. Prefix delegation updated via ansible. DNS secondary updated in 4 minutes. Reverse DNS took 16 minutes. Total impact: 20 minutes of stale PTR records only.

Checklist used:

• Prefix list in git repo
• Terraform state has no hardcoded addresses
• Monitoring uses hostnames not addresses
• Certificates not bound to specific prefix

Automation saved me. Posting in case useful.

Wah, 20 minutes is very fast lah! I already did prefix change last year with RackNerd, took 3 hours because I not prepare good. Can or not you share the ansible playbook? I want to see how you handle the delegation dong.

My setup still has some hardcoded address in firewall rule, need to fix already before next change. Your checklist is very helpful, brother!

This is insane value preparation!!

I run Ampere on HostHatch and did similar renumbering last month. Took 45 minutes because I had one legacy IP dependency hiding in monitoring config. Runs circles around my old x86 setup for DNS propagation though!

Your git repo approach is smart. I keep prefix list in ansible vault but not versioned well. Stealing this!

What tool you use for reverse DNS mass update? I had to script against KnownHost API directly — https://www.knownhost.com — their panel is slow for bulk PTR.