Issue/Introduction

A VM instance startup failure can have multiple symptoms and causes. Identify the applicable symptom and cause, and then try the corresponding resolution.

Environment

• Product: ZCF
• Component: Cloud
• Version: 3.x-5.5.30

Symptoms

A VM instance startup failure generally has one of the following four symptoms:

• Symptom 1: After you access a Linux VM instance, the error welcome to emergency mode! After logging in, type "journalctl -x b" to view system logs, "systemctl reboot”to reboot, "systemctl default"to try again to boot into default mode. Give root password for maintenance is displayed.
• Symptom 2: The VM instance fails to start and the error stderr:exec of\"/usr/local/zstack/dnsmasq\"failed:No such file or directory\n","location":" is displayed.
• Symptom 3: The VM console displays a black screen.
• Symptom 4: The VM instance fails to start and the error VG locked by other host is displayed.

Possible causes

Cause 1

Symptom 1 might occur for either of the following reasons:

• A data volume fails to be mounted automatically when the VM instance starts.
• An unexpected host power outage damages the file system.

Cause 2

Symptom 2 might occur because a host power outage causes the dnsmasq file to be lost.

Cause 3

Symptom 3 might occur because the SELinux configuration file is loaded incorrectly, preventing the system from starting normally.

Cause 4

Symptom 4 might occur because sanlock on a host in the cluster has not been released.

Resolution

Resolution for cause 1

• If a data volume fails to be mounted automatically when the VM instance starts, you can try the following procedure:

1. Sign in to the VM instance. In /etc/fstab, delete or comment out the command that mounts the data volume.
2. Edit /etc/rc.d/rc.local and use the volume UUID to configure the data volume to be mounted automatically at startup.

• If a host power outage damages the file system, you can try the following resolution:

Boot the system with a file system repair tool or a third-party live CD tool and repair the file system.

Resolution for cause 2

1. Use the scp command to copy /usr/local/zstack/dnsmasq from a healthy host to the affected host.
2. Reconnect the host.
3. Restart the VM instance.

Resolution for cause 3

• Enter single-user mode and check whether the SELinux configuration file contains an incorrectly written parameter. If it does, correct the parameter and restart the VM instance.

Resolution for cause 4

• On the VM Instances page, select the VM instance and click Actions > Advanced Policies > Start on Specified Host to start the VM instance on the host where it was last started.