After VMTools is installed and running normally on a virtual machine, the cluster periodically checks the state of the virtual machine's operating system. If the system detects that the operating system is unresponsive or has restarted abnormally, a critical alert is triggered. Specific scenarios include:
HA not enabled for the virtual machine operating system: If the operating system is unresponsive, a critical alert is triggered.
HA enabled for the virtual machine operating system: If the virtual machine restarts multiple times within a short period but fails to recover, a critical alert is triggered.
The virtual machine { .labels.vm_name } has an operating system failure.
The virtual machine { .labels.vm_name } has experienced operating system unresponsiveness { .labels.os_ha_count } times within { .labels.os_ha_duration } hours.
The virtual machine operating system may have frozen, crashed, or restarted abnormally, rendering the virtual machine unable to provide services externally. This may result in service interruption or slow response.
The virtual machine operating system is unresponsive or has experienced repeated failures.
Try logging in to the virtual machine using SSH, VNC, or other methods. After the login, check the system load, processes, and logs to identify the cause of the anomaly.
If you cannot log in to the virtual machine, or cannot perform any operations after logging in, log in to the node where the virtual machine resides and run the following command to export the memory dump file of the virtual machine:
virsh dump --memory-only --live $vm_uuid $vm_uuid.dump
Note:
The dump file size is typically close to the allocated memory size of the virtual machine. For large virtual machines, ensure that sufficient physical disk space is available. It is recommended to save the file to an NFS mount directory (such as
/usr/share/arcfra/images/tmp/).
After exporting the memory dump file, you can try performing a force restart on the virtual machine in AOC to reboot the system for recovery. If the virtual machine still cannot recover, contact an Arcfra after-sales engineer and submit the memory dump file.
