Here's a case that alarms in a controller of a storage array cannot be cleared after the host is restarted in a scenario where UltraPath V100R006 is used.
Symptom
UltraPath V100R006 is installed on a Linux host.
A storage array is connected to the host using an optical fiber.
A single-controller alarm is generated on the storage array due to path disconnection.
The host is restarted when the single-controller alarm persists.
After the path is restored, the single-controller alarm still persists.
Fault Diagnosis
After the single-controller alarm is detected by the user mode of UltraPath for Linux, the alarm is sent to the kernel.
After receiving the alarm, the kernel adds the alarm to a cyclic list and constantly tries to send the alarm to the storage array.
After the path is restored, the user mode sends an alarm clearance command to the kernel.
The command is also added to a cyclic list and retries constantly.
In addition, the alarm is transferred from the current alarm file to the historical alarm file.
Due to the frequent intermittent path disconnection, the kernel of UltraPath cannot send the alarm clearance command to the storage array because path selection fails.
After the host is restarted, the retry queue maintained in the kernel has been deleted.
The user mode does not maintain the previous alarm information anymore.
Therefore, the single-controller alarm clearance command is no longer sent to the storage array.
As a result, the single-controller alarm cannot be cleared.
Solution
This problem does not affect services.
Manually clear the alarm from the storage array.
Hope this will help you solve similar cases.
