Comment 0 for bug 1976440

Revision history for this message
Yi Feng (fengyi765) wrote :

After the Yoga version was released, the API and the MgmtDriver were verified and two MgmtDrivers were found to be faulty.

kubernetes_mgmt.py
  1. When checking whether the master node has been installed, the value of result.get_stderr() is handled incorrectly, so that the master node will not be installed.
  2. When checking whether the master node has been installed, if you directly call RemoteCommandExecutor to send commands, since the VM has not been started, the connection will fail, resulting in installation failure.
  3. When calling the encapsulated RemoteCommandExecutor function _init_commander_and_send_install_scripts(), the exceptions caught are not enough.
  4. Since the installation script install_k8s_cluster.sh does not specify the version of kubeadm, the master version is installed, and the master version is often unstable and the installation fails.
  5. https://review.opendev.org/c/openstack/tacker/+/830942 Modified the input parameter name of MgmtDriver, so that MgmtDriver cannot obtain the parameter value.

cnf_nodeport_mgmt.py
  1. When implementing the container update patch, it forgot to add a function to cnf_nodeport_mgmt.py, which resulted in an error when running python.
  2. If the names of the services are similar, the content of haproxy.cfg will be processed and modified incorrectly.