Hacker News Reader

Comment by nmc | Hacker News Reader

> DynDNS replacement

I would disagree. This may look like a replacement from the end-user's point of view, but it does not address the real issues of DNS management: space to store all zone data, bandwidth to support requests flow, resources to handle the load.

Either you are just proxying someone else's service, or you are running your own on (supposedly) limited resources. Am I wrong?

Moreover, I heard from DynDNS engineers that the main reason that led them to stop their free hostname offer was: it was abused for spamming too often. How do you intend to cope with that?

View on HN

> it does not address the real issues of DNS management: space to store all zone data, bandwidth to support requests flow, resources to handle the load.

You're absolutely right! However, as you said, it solves the user's problem from their point of view.

The main reason I started this project was that other dyndns alternatives (whether they be proxies or "fully-fledged" operations) have rather ugly domain names, in my opinion. 'zzzz' popped into my head, I saw that 'zzzz.io' was free, and decided to get it. A few weeks later, I finally got around to throwing a webapp together so others could benefit.

Not such a horrible thing to do, or?

And yes, the service is backed up by Amazon's Route 53. I did learn a thing or two about BIND and DNS from the experience though, which is also a plus.

I entirely agree, and I am glad you got my point.

Not sure about the beauty of 'zzzz', but it certainly is a good thing to learn by doing. Nice job.

You did not answer my question about spam. Are you taking measures to prevent your domain to being flagged as a spammer?

Nice resource. Just last month I set up Bind9 on a spare VM for the same reasons (albeit just for personal use).

Is this project something you'd consider releasing the source code for (obviously I'm happy to write my own code if needs be but there's no point in me reinventing the wheel if you already have this sitting on a public git repo)

You could've used FreeDNS (freedns.afraid.org) with your own domain. They do this for years.

Doesn't FreeDNS deactivate your account if you don't log into their website for a couple of months, just like DynDNS did?

No, it's Dyn's and No-IP's thing.

From FreeDNS news page[1]:

"2012-03-29 18:20:33, 2 years ago: Free accounts not accessed at least once every 6 months will be considered dormant and unloaded from memory."

What does "dormant" mean here? Do they delete your DNS records? It might not be as draconian as what DynDNS was doing, but certainly a barrier you should consider.

[1]: http://freedns.afraid.org/news/

Here's an email I got from FreeDNS last year:

Your account at freedns.afraid.org has not been visited in at least 5 1/2 months.

User: Xxxxx Xxxxx (xxxxxxx) Last visited: 2013-09-21 (169 days ago)

Unless you visit the site anytime in the next 2 weeks, your account will be considered dormant. "Dormant" consists of unloading any stale DNS records from memory which you may have set up in the past.

This stale entries optimization will free up several gigabytes of memory, making it available for active users. This will allow a DNS server to do a cold configuration boot, and load in new zones in a fraction of the time and space.

Users supporting freedns.afraid.org with a premium plan of any kind (even the smallest) will not be affected.

What does "not accessed" mean? One could write a script that would log in every month or so, click around (very carefully of course ;) ) and logout. How would they know the difference?

As a matter of fact, that's what I was doing to bypass DynDNS limitations: running a PhantomJS script via cron once every week to login, click around and logout. But it is a tiny hassle anyway. Much better to stick around with a service that does not have these annoyances.

And that's just what I did with https://github.com/kopf/dyndns-autologin . A few months later, they discontinued the free service altogether.

Somewhat sadly, it was my github project that seemed at times to garner the most interest from outside, even though it was just a shoddy python script thrown together in a 10 minute cigarette break at work.

Deleted comment

security vulns are not in and of itself (themselves?) indicative of the usability/quality of the software.

If someone were to finger all the nameservers in the world, I would bet that BIND has a large user share.

A good alternative is NSD from the team that also makes unbound.

Reply on news.ycombinator.com