Why Compute node tries to connect another IP address of controller?

asked 2017-09-11 04:16:26 -0600

Jianliang gravatar image

Hi All,

I am using packstack to install OpenStack on a 2 nodes cluster. Both 2 nodes are CentOS VMs. One is set as Controller (with 3 network interfaces, Host-only 192.168.99.106, NAT 192.168.1.4, and Host-only 192.168.99.110), the other is set as Compute. After generating the answers.txt for packstack, I modified the AMQP_Server to the controller IP 192.168.99.106. When I launched packstack to install, it was stuck with installing compute node at stage of "testing if puppet apply is finished". When I checked the log from compute node, the error (errno=113) showed that it kept trying to connect to 192.168.1.4 at port 5672. The port was defined as open in the iptables on the controller. The IP 192.168.1.4 can be pinged from the compute node. I don't why. Any help is appreciated. Thank you!

Best regard, Jianliang

edit retag flag offensive close merge delete