VxRack: storcli commands not effective

Using storcli commands to set VD attributes. This can happen after a disk failure, or replacing a disk.

Symptoms

Some storcli commands to set VD attributes would return successfully, but upon checking, the VD attributes are not updated.

Sometimes when there is a failed disk, the attributes of other VDs might change and storcli commands cannot set them.

Impact

Reduced performance

This problem can occur in two situations:

• RAID controller firmware has been upgraded, but the host has not been rebooted yet
• There is preservedcache for the failed disk, which prevents the storcli commands to be effective

Use "storcli /c0 show all" to check what is the version of currently running firmware - if it is not, reboot the host in order to load the correct firmware.

If the firmware is already running at its expected version, use the following command to check preservedcache:

storcli /c0 show preservedcache

If there is preservedcache, this command shows it, for example:

Controller = 0
Status = Success
Description = None
-----------
VD State
-----------
4 Missing

And the VD number in the output will most likely not be present in "storcli /c0/vall show," as it is usually a failed disk.

Use the following command to delete preservedcache (Replace '#' with the number in "VD" column in above command's output. Based on the above example, use number '4').

storcli /c0/v# delete preservedcache [force]

After that use storcli commands to set VD attributes again.

Impacted versions

This is not a ScaleIO software issue and can affect all servers with LSI raid controllers and all versions of ScaleIO.

Fixed in version

N/A This is not a ScaleIO issue.