如何在Kubernetes集群中使用脚本语言连接数据库

Question

Using Google Kubernetes Engine, I have set up a mysql database deployment & service in kubernetes, much like the wordpress example . I want to access it via R with R which I have in another deployment & service in my kubernetes cluster (r-user-app, w/2 containers). This worked with networks in docker.

> kubectl get services
NAME         TYPE           CLUSTER-IP      EXTERNAL-IP      PORT(S)                         AGE
kubernetes   ClusterIP      10.15.240.1     <none>           443/TCP                         8d
mysql        LoadBalancer   10.15.248.129   35.227.123.232   3306:31410/TCP                  3m
r-user-app   LoadBalancer   10.15.255.135   35.227.33.206    3838:31003/TCP,8787:32395/TCP   1d

Exec into the rstudio container inside the r-user-app deployment, I see the mysql service:

root@r-user-app-1788932656-vgczh:/# getent hosts mysql
10.15.248.129   mysql.default.svc.cluster.local

However, when I try to get it via R, I have no such luck:

root@r-user-app-1788932656-vgczh:/# R
R version 3.4.2 (2017-09-28) -- "Short Summer"
-----
> require("RMySQL")
Loading required package: RMySQL
Loading required package: DBI
> mydb<-dbConnect(MySQL(), user="root", password="mypassword", host="10.15.248.129")
Error in .local(drv, ...) :
  Failed to connect to database: Error: Can't connect to MySQL server on '10.15.248.129' (107)

Using the EXTERNAL-IP doesn't work either, and putting port=3306 doesn't help.

Same thing in python:

> db=MySQLdb.connect("10.15.248.129", "root", "mypassword")
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/usr/local/lib/python2.7/dist-packages/MySQLdb/__init__.py", line 81, in Connect
    return Connection(*args, **kwargs)
  File "/usr/local/lib/python2.7/dist-packages/MySQLdb/connections.py", line 193, in __init__
    super(Connection, self).__init__(*args, **kwargs2)
_mysql_exceptions.OperationalError: (2003, 'Can\'t connect to MySQL server on \'10.15.248.129\' (111 "Connection refused")')

Maybe I don't even know the right question to ask! Please help! #kubenube

Answer 1

You don't need to use the IP. You can access the servers directly by name. You even don't need to use the full DNS name with "mysql.default.svc.cluster.local", just "mysql" is enough.

You should also check the mysql pod, if it is deployed properly. Using Type=LoadBalancer is not a good idea for databases, as it opens them to the outer network.

Answer 2

Answering my own question: This was an issue with the yaml. Needed to be sure to match the Service

  selector:
    app: my-app
    tier: mysql

and the Deployment

spec:
  selector:
    matchLabels:
      app: my-app
      tier: mysql

Now we can access the database using "mysql" instead of a host IP address.