CODEWITHSUNDEEP
openshift
Michael Courcy
Michael Courcy

Reputation: 637

How are resolved *.svc address in openshift cluster

When I create an app in openshift it expose a service in *.svc address.

oc project
Using project "coco" on server "https://master.lab.example.com:8443"
oc get svc -o wide
NAME      CLUSTER-IP      EXTERNAL-IP   PORT(S)    AGE       SELECTOR
ruby-ex   172.30.18.144   <none>        8080/TCP   9h        app=ruby-ex,deploymentconfig=ruby-ex

From a node of the cluster I can curl it 

 curl -vvv ruby-ex.coco.svc:8080
 * About to connect() to ruby-ex.coco.svc port 8080 (#0)
 *   Trying 172.30.18.144...
 * Connected to ruby-ex.coco.svc (172.30.18.144) port 8080 (#0)
 ....

Thus it work as expected, but I can't find who's solving this domain name. Still from the same node 

 dig +short ruby-ex.coco.svc

Output nothing. Thus I wonder how curl find the right ip ?

EDIT 1: Following the answer of Graham my understanding progress

The file /etc/resolv.conf has this content :

search  lab.example.com cluster.local
nameserver 10.0.2.15

I can see that 10.0.2.15 belong to the node and dnsmask listen on port 53 

netstat -tulnp | grep 10.0.2.15
tcp        0      0 10.0.2.15:53            0.0.0.0:*               LISTEN      1274/dnsmasq
udp        0      0 10.0.2.15:53            0.0.0.0:*                           1274/dnsmasq

Going to the conf of dnsmask 

cat /etc/dnsmasq.d/node-dnsmasq.conf
server=/in-addr.arpa/127.0.0.1
server=/cluster.local/127.0.0.1

And 

netstat -tulnp | grep 127.0.0.1:53
tcp        0      0 127.0.0.1:53            0.0.0.0:*               LISTEN      2128/openshift
udp        0      0 127.0.0.1:53            0.0.0.0:*                           2128/openshift

Show that the process openshift is listening to 127.0.0.1:53.

When I dig using this server the connection timeout 

dig +short  @127.0.0.1 ruby-ex.coco.svc
;; connection timed out; no servers could be reached

But for a known domain in the cluster it answers properly

dig +short  @127.0.0.1 node1.lab.example.com
172.25.250.11

I just can't make the link between cluster.local and an *.svc address and I can't also explain why curl resolve the address but not dig.

Edit 2

I finally found the the answer, due to my bad understanding of /etc/resolv.conf. With the search directive in resolv.conf curl try ruby-ex.coco.svc and then ruby-ex.coco.svc.cluster.local and so on trying each domain. And indeed 

dig +short @127.0.0.1 ruby-ex.coco.svc.cluster.local
172.30.18.144

works as expected, Thanks for your help.

Upvotes: 3

Views: 2084

Answers (1)

Graham Dumpleton
Graham Dumpleton

Reputation: 58523

There is a DNS internal to OpenShift which handles the request. Look at /etc/resolv.conf and it will have something like:

nameserver 172.31.39.82
search myproject.svc.cluster.local svc.cluster.local cluster.local ap-southeast-2.compute.internal
options ndots:5

A hostname corresponding to the name of a service, when created, is added to the internal DNS.

Upvotes: 4

Related Questions