Since some time ago DNS cache started periodically get 0.0.0.0 IP for local addresses. I do have a local DNS serving intranet zones on one of the servers and router refers to its DNS (non-forwarding one) among with provider’s DNS. At the same moment when router gets 0.0.0.0 I still can run normally nslookup <server_name> <local_DNS> and it return proper IP while requesting IP from router gives an error. Currently I do solve this by flushing DNS cache periodically but maybe there is proper solution? Or this is just a bug of 6.29.1 and will be solved in further versions?
Hardware RB1100AHx2 running RouterOS 6.29.1
Here is my DNS config
Screen Shot 2015-07-28 at 14.39.09.png
And here is what cache is getting
Screen Shot 2015-07-28 at 14.39.11.png
Hey @Ajax I can confirm this issue.
As of now the 7.20.x branch still has this or a similar issue.
In my test scenario we have more then 1k devices which are using upstream public dns servers ie:
8.8.8.8
1.1.1.1
208.67.222.222
And their pairs:
8.8.4.4
1.0.0.1
208.67.220.220
We have tested it on:
DSL
LTE
FIBER
The main domain which we tried verifying the issue against is: lambda.eu-west-1.amazonaws.com
which has a 60 seconds ttl.
We were watching the cache entries since the cache flush and we have seen for no reason a 2 hours ttl 0.0.0.0 entries for multiple domains.
Since the dns service is critical for this business environment we have decided to shutdown the DNS service on the RouterOS device and offload this function to a more reliable solution.
One solution would be to distribute the most stable and reliable public dns service using the DHCP network configuration.
From my experience dnsmasq, unbound and bind don’t suffer from this illness.
I understand that Mikrotik would like to provide more features in their dns service, but this one to my understanding from the fact is designed for home environment.