Networking Issues

Symptom Troubleshooting/Resolution
Unable to load vagent on one container, but it works on another container.

This may be caused by an inconsistent iptables setting across all the Worker hosts.

On the affected Worker host, edit /etc/bluedata/bluedata.conf and adjust the iptables setting to match that of the other Worker host(s) in the Container Platform deployment. Then restart the bds-controller service by executing the following commands:

stop bds-controller
                        start bds-controller
The container hostname cannot be resolved, but the container can be accessed via its IP address.

Execute the command # ps -ax | grep dnsmasq. If this is not pointing to /etc/bluedata/bds-dnsmasq.conf, then adjust this by executing the following commands:

sudo service dnsmasq stop
                        sudo chkconfig -del dnsmasq
                        sudo /usr/sbin/dnsmasq --conf-file=/etc/bluedata/bds-dnsmasq.conf
The Controller cannot access or ping a container (virtual node).

Validate the following network settings:

  • Floating IP range
  • CIDR info
  • IP next hop address

Collect the web interface screenshot of these settings.

Verify that you can ping between the physical hosts. While running the ping test, collect the packet trace on both the Controller and the affected virtual node. This will help narrow down the network failure. The Gateway may be improperly configured, or the wrong CIDR may be in use. Use the tcpdump tool, such as:

tcpdump -i bond0 icmp or arp

Check this at the following locations:

  • Host: bond0/eth0
  • Internal gateway: bd_public

  • Inside the container: Verify that the tcpdump RPM package is installed, and then execute the following:

    route -n
                                tracepath -n <destination>
                                traceroute <destination>