Problem Symptom
1. In FusionStorage 8.0.RC1, FusionStorage 8.0.RC2, FusionStorage 8.0.RC2.SPH001, FusionStorage 8.0.T12, FusionStorage 8.0.0, FusionStorage 8.0.T29, FusionStorage 8.0.0.1, or FusionStorage 8.0.T31, after the management node IP address is changed, DeployManager fails to be upgraded.
2. If only the management node IP address is changed, the nodes whose IP addresses are not changed fail to be upgraded during the upgrade of the first component.
Problem Diagnosis
1. Run the cat /var/log/deploy/scripts command to view the alarm information about the upgrade failure of the standby management node. Then, check whether the IP address configured in the /opt/servicetool/data/backup/main.yml file on the active and standby management nodes is the IP address currently used by the system. If the IP address information is incorrect, partial configuration information is missing during the IP address change in the earlier version.
2. Run the ps -ef|grep /opt/fusionstorage/deploymanager/clouda/bin/cloud_agent|grep -v grep command to check the process startup time of all nodes whose IP addresses are not changed and check whether the CloudA process is restarted after the IP address of the management node is changed.
Causes
1. In the earlier version, partial configuration information is missing when the IP address of the management node is changed. As a result, DeployManager cannot be uninstalled and fails to be upgraded.
2. After the management node IP address is changed, if the CloudA process on all nodes whose IP addresses are not changed are not restarted, an error message is displayed indicating that DeployManager fails to be upgraded.
Solution
1. Check before the upgrade: For the version that fails to be upgraded, check whether the IP addresses configured in the /opt/servicetool/data/backup/main.yml file on the active and standby management nodes are those used by the system. If the IP address information is incorrect, run the vim main.yml command and manually change the IP address in the main.yml file to the current IP address used by the system.
To prevent this problem, after the management IP address is changed, run the vim main.yml command and manually change the management IP address configured in the /opt/servicetool/data/backup/main.yml file to the new management IP address that is used by the system currently. Change the value of uninstall_action to 0 in the upgrade_step.ini file in the cd OceanStor_100D_deploymanager*/action directory. Then, perform the upgrade again.
2. Restart the CloudA process on all nodes whose IP addresses are not changed and try again.
Run the kill -9 $(ps -ef|grep /opt/fusionstorage/deploymanager/clouda/bin/cloud_agent|grep -v grep|awk '{print $2}') >>/dev/null 2>&1 command.
After the process is killed, the scheduled monitoring task automatically starts the CloudA process. Run the cd OceanStor_100D_deploymanager*/action command to go to the directory where the DeployManager upgrade package is stored, and run the sh upgrade.sh command to perform the upgrade again.
Check After Recovery
Check whether the upgrade can be performed successfully.
Suggestion and Summary
N/A
Applicable Versions
All
