Scenario:
1.The storage array was directly connected to the host through two redundant paths. A LUN was mapped to the host.
2.The LUN was successfully partitioned and was allocated with drive letters. An NTFS file system was created on the formatted LUN.
3.The host performed read/write operations on the LUN and the LUN failed. In such a case, the following two situations may occur: VXVM could not update LUN status in a timely manner. VXVM could not update the partitions created on the LUN in a timely manner.
4.Symptoms are: Files could still be copied to the partitions created on the failed LUN.
5.These files were stored on the host cache. However, if the host powered off unexpectedly, the file data was lost.
Cluster software version: native version 6.0The service data was written to the cache of the Storage Foundation (SF) management, not to the storage array directly.
The operating system did not report the error upon the LUN failure, and VXVM did not cope with the LUN status. Therefore, the data was still copied to the partitions created on the failed LUN.
The data was not written to the storage array until the memory allocated to SF was used out. The storage array then detected the LUN failure and reported it to the operating system which reported it to VXVM. VXVM then updated the status of the LUN and its partitions.
If it is absolutely necessary to use Huawei Symantec SF for managing disks on Windows, users must be notified of the risks described in this case.
SF is an aggregate of the storage management software provided by Huawei Symantec to offer a complete solution of managing heterogeneous storage arrays online. SF includes VXVM and VXFS.
END