[Problem Description]
When NAS services are used, the host performs operations similar to move a/b/b a/ on the file system, triggering controller self-healing. After the self-healing, the controller cannot be powered on.
[Symptom Description]
The controller performs self-healing and cannot be powered on, and NAS services of the file system cannot be used.
[Cause]
1. Perform the move a/b/b a/ operation on the file system. The V3 storage system locks the source and destination files (a/b/b, a/). If a file with the same name exists on the destination end, delete the file and move the file. Files a and b are locked twice before deletion. In this example, a and b are locked twice, causing a deadlock. As a result, storage controller B hangs I/Os. (The following uses the file system on controller B as an example.)
2. After controller B performs self-healing, services are switched to controller A. The host delivers the move command again, causing I/Os to be suspended on controller A. During the power-on process of controller B, the image cache of controller A is required. In V3R2, the hangIO operation is performed on all file systems in batches. (Suspend the I/Os delivered by the host and wait for the I/Os being executed to end.) Because I/Os have been mounted to the file system, all file systems on controller A are in the hangIO state and cannot be stopped. All NAS services are suspended. In addition, controller B cannot obtain the image of the peer controller. As a result, controller B fails to start.
[Location Method]
Search the message log for Heal print (pal_counter) massage begin: and check whether the statistics LOCK_SPC_DMNFOR_READ and LOCK_SPC_DMNFOR_WRITE have returned I/Os. If there are both, this is the problem.
[Emergency Measures]
Restart the two controllers at the same time. That is, run the rebootsys command in the minisystem to restart the two controllers.
[Solution]
Upgrade the device to V300R003C10SPC100 or a later version.
[Post-Recovery Check]
NAS services of the storage system are running properly.
[Appendix]
NA
