[Problem Description]
R version upgrade process (The source version is converged storage V3R1/V3R2, and the target version is converged storage V3R3 or later.) During the upgrade, the system control board is reset unexpectedly. As a result, the upgrade process is suspended and services on the NE are interrupted. The upgrade fails in the dual-to-single-to-single-to-reset process. The failure step is StartRssService.
[Symptom Description]
After the upgrade fails, check whether a process fails (duo-to-single-to-single-to-single-to-single-to-single-to-single-to-single-to-single-failure) on the main control board.
[Alarm Information]
NA
[Cause Description]
(1) Why is controller A reset abnormally and the upgrade process suspended?
--After controller A is reset abnormally, controller B fails to process the dual-to-single-to-single-to-single-to-single-to-single-to-single-to-single-to-single-to-single-to-single-to-single-to-single-to-single-to-single-to-single-over upgrade fails. As a result, the upgrade fails and controller A fails to be powered on.
(2) Why does the dual-to-single switchover of controller A fail?
--If the dual-to-single-to-single-to-single-to-single-to-single-to-single-to-single-to-switch LUN switchover fails, value-added services fail to be started at the destination end (controller B) of the LUN switchover.
(3) Why does controller B fail to start value-added services?
--Before dual-to-single LUN switchover, the switchover of controller B is being processed. (Controller B is powered on after the upgrade is complete and restarted.) Two LUN switchover processes are concurrent. As a result, value-added services fail to be initiated. (Concurrent replication is not supported.)
(4) Why does the system control switch over LUNs concurrently?
--In the R version upgrade, in the dual-to-single-to-single-to-single-to-dual process, the OBJ messages sent by the faulty node are reclaimed (waiting for the completion of the LUN switchover upon power-on). Due to the change of the reclaiming mechanism in the R version, the reclaiming is incomplete. As a result, value-added services are restarted, and the process fails.
[Problem Location]
The log of the main control board contains RSF_INFO:other sys event is handling, current callback fail.][RSF][RSF_DoStartLunWork.
[Solution]
1. Patch installation is not supported for problems occurring during the upgrade.
2. Emergency on the live network:
1). Enter the developer mode.
2). Run the minisystem command to enter the minisystem system.
3. Run the rebootsys command to restart the controller.
4) Repeat steps 1 to 3 to restart the remaining controllers.
5) After the operation is complete, wait until the system is powered on again.
[Applicability]
Emergency commands are applicable to all versions.
