I use linux bridge mode to set up a intranet. But always crash

asked 2018-04-23 09:20:43 -0500

Rich gravatar image

The current status of our architecture: 1 X controller node 1 X network node 2 X compute node: compute1, compute2 1 X block node We are using this cloud to handle high traffic of video stream. There are two protocol we support, rtmp and http. In each node except controller node. We create virtual machine to handle traffic. Each node use 10gb switch to link each other. Every virtual instance can communicate each other through vlan. The network (10Gb) and controller node(1gb) have the ability to communicate outside, but not the others.

In application level: The network node have one instance with haproxy inside to handle load balance. The two compute node have two instance inside each to handle http (nginx) and rtmp(adobe media server) request. The block node have one instance with nfs server insde, which provide share file system for application to use.

The problem is that: for some reason, the whole intranet crash. When the crash happen. We can not visit each node using intranet. In some case we have to use IP console to restart one of the compute node: compute1 in most case. And then the intranet will come back. Finally, we restart the other virtual instance. The whole cloud will come back.

Does any go through this situation? Please tell the way to debug the intranet.

Regards

edit retag flag offensive close merge delete