A Management Node upgrade fails. How do I resolve this?

Symptom

In a dual-Management Node deployment, one Management Node fails to upgrade.

Cause

Database synchronization is out of sync, causing one Management Node to fail the upgrade.

Solution

1. Run the following command on each Management Node to back up the database:

   [root@localhost ~]# zstack-ctl dump_mysql

2. Run the following command on the failed Management Node to upgrade it individually:

   [root@localhost ~]# bash zstack-install.bin -u

3. Run the following command on the other Management Node to reinstall the dual-Management Node service:

   [root@localhost ~]# ./zsha2 install-ha -config zs-install.conf

4. The installation succeeds and the environment is restored.

   Do not perform any other operations on the system during a platform upgrade to avoid upgrade failures.