rancher: rancher-dns does not honor upstream TTL, static TTL for all requests

Rancher versions: rancher/server: 1.5.9 rancher/agent: 1.2.2

Infrastructure Stack versions: network-services: metadata 0.9.1, network-manager 0.6.6

Docker version: (docker version,docker info preferred)

Operating system and kernel: (cat /etc/os-release, uname -r preferred)

RancherOS 1.0

Type/provider of hosts: (VirtualBox/Bare-metal/AWS/GCE/DO)

Parallels VM

Setup details: (single node rancher vs. HA rancher, internal DB vs. external DB)

HA with external DB

Environment Template: (Cattle/Kubernetes/Swarm/Mesos)

Cattle

Steps to Reproduce:

I have consul dns for service discovery. When I start my services dns lookup seems to be ok:

me@container$ dig db.cachet.service.x
...
...
db.cachet.service.x. 0	IN	A	10.42.143.80

After that I try to upgrade db and got the same thing, but on host(outside of container) dns lookup works fine:

me@host$ dig db.cachet.service.x
...
db.cachet.service.herzen. 0	IN	A	10.42.8.85

me@container$ dig db.cachet.service.x
...
...
db.cachet.service.x. 0	IN	A	10.42.143.80

Is there any way to reduce TTL for internal dns cache? Because, there is no way to wait five or more minutes for dns cache invalidation.

Guys, please, do something with it. It’s really awful.