Member OP
Good one. I checked mine on Time4VPS before I left: 5.4.0-80-generic, last compiled July 2021. On a "fully managed" platform.
Their control panel still showed "Ubuntu 20.04 LTS" as the latest template. That was my final signal.
Template rot is real. I migrated a client off a host whose "Debian 11" image still had systemd 247 from the pre-release cycle. They'd built it once in 2021 and never refreshed.
Restore test caught it. Their backup "worked" but the restored image couldn't apt update because the release key had expired. Untested 3-2-1 would've failed at the worst moment.
I run one small server in Osaka. My signal is simpler: can I still reach a human?
Last host I left, support went from "reply in 4 hours" to "ticket closed with KB link" in six weeks. The KB was generic, written for a different product line entirely. New owners, same brand, fired the tier-1 team.
Now I pay more for less resource but actual support. Worth it.
Tokyo here. We see this with regional resellers of international brands. The local entity vanishes, billing redirects to HQ, support timezone shifts 12 hours.
I track corporate registration changes. Japan has decent disclosure for KK structure changes. Not as fast as DNS but more authoritative.
Bangalore angle: payment gateway changes. If your invoice suddenly routes through a different Stripe account or Razorpay merchant, ownership shifted. GSTIN mismatch on the invoice is dead giveaway.
Happened with a local Hostinger reseller two years back. Same brand, new PAN, support went to chatbot.
Toulouse. I watch peering and routing. Acquisition often means network consolidation—your traffic suddenly trombones through Amsterdam or Frankfurt when it used to stay regional.
MTR is free intelligence. If latency to Paris jumps 15ms because you're now backhauled to Amsterdam, they moved your VM or sold the ASN.
KnownHost has Amsterdam, fine. But if you're on their Dallas node and suddenly route through Cogent in Ashburn, ask why.
Sydney agreeing hard. We get this worse—APAC networks get "optimised" into Singapore or Tokyo hubs and suddenly Perth traffic goes via California.
I map my routes monthly. Simple script, cron job, diff the AS paths. Caught a provider migration before they announced. Gave me three weeks head start.
Frankfurt. I add one: abuse contact changes in RIPE.
New abuse-c handle, new org handle, often weeks before public news. The RIR database updates fast because routing depends on it.
Whois -h whois.ripe.net <AS> and watch for org-name changes. Not perfect—some hosts use LIR services—but dedicated hosts usually self-manage.
Dakar. Different problem: some of us can't afford to monitor this closely. The deal host is cheap because we are price-sensitive.
My signal is community. If the LowEndTalk thread goes quiet and the owner stops posting, I move. Free, social, works well enough. I use https://lowendtalk.com.
This.
