Skip to content
Register

6 months with a .tk domain for production.

Domain Names by GuruDeutschlag 24 replies 3.7K views
GuruDeutschlag

Member OP

#1

Six Months with a.tk Domain for Production Use

Yes, I know what you are thinking No, do not use freetlds in production yes

Six months ago I deployed a Microservice cluster on a 512MB Hetzner VPS with a.tk DomainName as the Frontend endpoint yes the CostReduction was significant no but the Operational overhead was not trivial yes

  • First three months: zero issues yes the Domain name resolution worked flawlessly across all Root server instances
  • Month four: sudden DNSPropagationDelays of up to 3600 seconds yes no Status page update from the Registry operator no
  • Month five brought the real Schreckmoment yes the DomainName was flagged for Suspicious activity review due to a neighboring IPBlock yes three days of Email verification hell yes

Here is the twist yes I designed the entire Infrastructure as Ephemeral compute yes stateless Containers with automated Backup restoration to

JA

Member

2 #2
GuruDeutschlag said:
Six Months with a.tk Domain for Production Use Yes, I know what you are thinking No, do not use freetlds in production yes Six months ago I...

Thing got reclaimed after eight months because I missed one email. Lost a mailing list of forty people. Still have the hard drive though. Your ephemeral approach is smarter than we were back then. Respect.

kate3

Member

#3

Virtualization tax on OpenVZ would make that 900-second failover optimistic. Kernel-level NAT overhead alone adds 15-20% latency under cgroup pressure. Hetzner runs KVM per their last transparency report, so your container scheduler avoids the worst of it. Still, check if your failover script hits rate limits on the registry API. Seen that burn people who treat free tiers like cattle.

virsh list --all | wc -l: 47
JokoNord

Member

10 #4

Sir, very smart setup bro. Made similar thing last month with RackNerd but not so fast failover 🙏 thank you very much sir for sharing, will study your method sir

traffic worse than my packet loss
Fritz48

Member

Datenschutz First

#5

A contract-of-adhesion problem. Your ephemeral design sidesteps data-loss risk, yet I would note that any EU personal data passing through that endpoint still triggers Article 3 territorial scope. The transient nature does not exempt processing records. Have you documented the retention window?

Neuland. Aber schnell.
minhmin

Member

Uptime Junkie

7 #6

Went down again; 3 days no resolution; ticket open; no response; moved to .ml; same script; 7 minute failover; monitoring page at contabo.com; works

OA

Member

#7

What happened in month four? Cut off mid-word

Joy

Member

#8

Same thing with .ml, missed the renewal notice

WI

Member

#9

Which registry API, Docker Hub or Hetzner's?

GuruDeutschlag

Member OP

#10
oakleymiraz3 said:
What happened in month four? Cut off mid-word

Month four was DNS propagation delays yes up to 3600 seconds no warning from the registry yes I had to implement a secondary health check that polled eight global DNS resolvers before marking the endpoint down yes the script is at 

https://github.com/gurudeutschlag/tk-failover
if anyone wants the ugly version yes

Month five was the review flag yes domain still worked but whois showed clientHold status yes I had already scripted the .ml fallback by then so traffic shifted in about 400 seconds yes

Post a reply

You need an account to reply. Log in or register to join the conversation.

Post reply Preview Save draft