api-us-east-1-cell-1.domotz.com.

[johnpoz 2]

Does this really need a TTL of only 60 seconds?

 

It generates a HUGE amount of queries with such a short ttl.. I don't want to really override it locally... But such sort ttl just seems unwarranted unless you were about to change it out.  Wouldn't something like 5 min or 10 or even higher not be viable?

[Giancarlo 26]

Hi Johnpoz,

 

thanks for highlighting this.

 

Very recently we moved our DNS from one provider to another. In order to prevent issues, and having the possibility to rollback to the previous provider in the shortest time, we had setup that really short TTL (1 minute).

 

We have now reverted back to its original value (10 minutes). We will evaluate in the next few days whether it is safe on (our side) to make it at a even higher value.

 

Please, let us know if this is fine for the moment. Stay tuned, and I'll let you know if we will be able to set it up to 1h (or so).

 

Thanks,

Giancarlo 

[johnpoz 2]

Yeah that looks better at 600 for number of queries being seen

 

Thanks!  I would think that should save you some money too.. Pretty sure awsdns charges per number of queries...

[johnpoz 2]

Well its back to 60 seconds again...

 

;; QUESTION SECTION:
;api-us-east-1-cell-1.domotz.com. IN    A

;; ANSWER SECTION:
api-us-east-1-cell-1.domotz.com. 60 IN  A       34.193.116.17

;; AUTHORITY SECTION:
domotz.com.             172800  IN      NS      ns-1501.awsdns-59.org.
domotz.com.             172800  IN      NS      ns-1761.awsdns-28.co.uk.
domotz.com.             172800  IN      NS      ns-415.awsdns-51.com.
domotz.com.             172800  IN      NS      ns-940.awsdns-53.net.

 

Are you moving DNS again?  Looks to be the same awsdns to me..

[johnpoz 2]

Come on really... Why is this still 60 freaking seconds?

[Giancarlo 26]

1 hour ago, johnpoz said:

Come on really... Why is this still 60 freaking seconds?

 

It has been changed back to 10 minutes now. Note: due to scalability, we might temporary change back to 60 seconds a few times in the next few weeks (we want to avoid loosing connections for longer during migration of services).