Hi team, here's a new case.
Problem Symptom
If the compute node uses the Arm OS,
the VBS process cannot be started after the OS is upgraded and the VBS process exits.
The VBS process exits in the following scenarios:
(1) The OS is rolled back.
(2) The node is reset.
(3) The VBS Process is abnormal.
Problem Diagnosis
Log in to the node where the VBS process is abnormal as user root.
Run the following command to check whether the operating system is Arm and whether the returned result is aarch64. If yes, go to 3. Otherwise, this document is not applicable.
uname -m
Run the following command and check the command output. If value in the command output is greater than 1, the problem occurs. Otherwise, this document is not applicable.
cat /opt/dsware/vbs/conf/kernel_version_record | wc -l
Causes
After the OS of a compute node is upgraded, the kernel version changes. VBS selects the matching kernel ko file based on the upgraded kernel version. During the selection process, the system searches the /opt/dsware/vbs/conf/kernel_version_record file first. If the file contains two or more records, the matching rule treats multiple records as one record. As a result, the matching parsing is incorrect, the matching fails, and the VBS kernel module cannot be started.
The /opt/dsware/vbs/conf/kernel_version_record file contains two records, as shown in the following figure:
Solution
Log in to the faulty VBS node as user root and run the following command to clear the kernel matching records. After the command is executed, the VBS process automatically starts.
> /opt/dsware/vbs/conf/kernel_version_record
Check After Recovery
Run the following command to check whether the VBS process is started:
ps -ef | grep dsware_vbs | grep -v grep
Check the command output. If the command output is not empty, the VBS process is started properly.
