The "script" forgot? What you mean to say, is that you ran shitty code in a production environment, very likely without proper testing beforehand.
Funny how when you do an OpenVZ migration, even manually, the only thing that needs to be updated in Solus is the nodeid. You see, OpenVZ stores information such as what IPs are assigned to a VM in that VM's conf file, right on the node. And when you perform a migration, one of the first things OpenVZ does is recreate said conf on the new node, ensuring that all of the same specs/settings/ips are intact.
So what does this mean for little Jon's story? Well, if you strip the excessive BS and just look at what he's actually saying, you get "In-Use IPs were marked as available in Solus, and re-assigned to new VMs". Well, we just covered that OpenVZ migrations don't require screwing with Solus' ipaddresses table, so just what was going on there?
In order for Solus to reassign those IPs, the ipaddresses.vserverid field would've had to be set back to 0/''. This doesn't happen on its own - either someone removed and readded a range in Solus, or this "in-house developed script" was designed by someone that has no CLUE what they're doing. And those are the folks administrating the nodes, folks. Pretty damn scary.
Just to grind a little bit more salt in, and show just what a BS excuse this is - even if this 'automated script' nonsense is true (why the hell would you want 'automated mass migration'... do you even know what that means?), and a bunch of clients had duplicate IPs... why would you renumber entire locations? Out of (boredom, mostly), I wrote a couple of quick bash and perl scripts. The first one queried _EVERY_ one of our OpenVZ nodes, took note of containers sharing an IP. The second script, had the first found any matches, would've given the newest VPS of the pair a new IP, removed the duplicate, then hit the Stallion API to email the client letting them know what had happened. That's problem fully solved _AND_ affected client notified, all in under half an hour. Now, based on estimated runtime for processing say... 15,000 VPSes with duplicate IPs - I figure that whole process would've taken 3-4 hours.
But, no. He only has 3 nodes affected, and his solution? Mass-renumber, inconveniencing EVERY client on the node, and causing even more damaging PR when he has to *struggle* to get even backhanded compliments.
Honestly, anyone that believes his stories deserves to be his "client" at this point.