3 ms·

as much as i like cloudflare (and i like them a lot), it's kind of absurd that this kind of thing can happen. a lot of red flags that, if true, would mean that

as much as i like cloudflare (and i like them a lot), it's kind of absurd that this kind of thing can happen. a lot of red flags that, if true, would mean that their infrastructure require a lot more care (127.0.0.1 as the source of an audit event? no email when DNS records are deleted? no 1-to-1 message due to this happening?).

ocdtrekkie 7y ago
At the very least, this sort of lack of good process is definitely what happens when Google decides to cut you off (and another person just commented a similar experience with Amazon), but I suspect it's likely the case for a much larger number of companies and services than people realize. It's fundamental internet architecture, and often little more thought goes into account termination than what you'd do to ban someone from your mid-2000s phpBB forum. So much business focus goes into the onboarding experience, and since you assume all of the people your service terminates are "probably bad people anyways", not a lot of thought goes into offboarding, or ideally, appeals.
- use-net 7y ago
  just wait until MS revokes certain certs and all Win machines with TPM stop booting LOL!
thedanbob 7y ago
I had an issue with them recently where a SRV record pointing to “.” (meaning “service unavailable”) was being rewritten to the string “false”. It didn’t take them too long to fix it, but it made me wonder how they managed to push a bug like that to production without some sort of automated test catching it.
- ocdtrekkie 7y ago
  IIRC, if you're on a free plan you get exposed to code changes a little faster than their paying customers.
  - Operyl 7y ago
    Correct, if I recall correctly they outlined this in their SEC filings.
  - thedanbob 7y ago
    Which is fair, I'd rather be a guinea pig than look at ads in exchange for a free service. I was just surprised that the thing they broke was as well defined and testable as DNS validation.
- johnklos 7y ago
  Simple. They don't give a damn about doing what we've all been doing properly for a quarter of a century. Apparently these large companies are above owning O'Reilly books.

ocdtrekkie 7y ago

At the very least, this sort of lack of good process is definitely what happens when Google decides to cut you off (and another person just commented a similar experience with Amazon), but I suspect it's likely the case for a much larger number of companies and services than people realize. It's fundamental internet architecture, and often little more thought goes into account termination than what you'd do to ban someone from your mid-2000s phpBB forum. So much business focus goes into the onboarding experience, and since you assume all of the people your service terminates are "probably bad people anyways", not a lot of thought goes into offboarding, or ideally, appeals.

use-net 7y ago
just wait until MS revokes certain certs and all Win machines with TPM stop booting LOL!

thedanbob 7y ago

I had an issue with them recently where a SRV record pointing to “.” (meaning “service unavailable”) was being rewritten to the string “false”. It didn’t take them too long to fix it, but it made me wonder how they managed to push a bug like that to production without some sort of automated test catching it.

ocdtrekkie 7y ago
IIRC, if you're on a free plan you get exposed to code changes a little faster than their paying customers.
- Operyl 7y ago
  Correct, if I recall correctly they outlined this in their SEC filings.
- thedanbob 7y ago
  Which is fair, I'd rather be a guinea pig than look at ads in exchange for a free service. I was just surprised that the thing they broke was as well defined and testable as DNS validation.
johnklos 7y ago
Simple. They don't give a damn about doing what we've all been doing properly for a quarter of a century. Apparently these large companies are above owning O'Reilly books.