Administrator Guide

unmapping on hosts or volumes (depending on the host operating system) that are using replicated volumes. Refer to the following
sections for information about disabling and enabling unmapping in VMware ESX, Red Hat Enterprise Linux, or Windows Servers
2012 and 2016 operating systems.
Unmapping with Red Hat Enterprise Linux
By default, unmapping is disabled in Red Hat Enterprise Linux.
Linux le systems that support unmap operations, such as ext4 on Red Hat Enterprise Linux 6, must be mounted with the-o
discard option to enable free space recovery functionality. If the le system does not support the-o discard mount option, it
does not support free space recovery on PS Series storage array volumes.
When free space recovery using volume unmapping is enabled, larger les are deleted and their space is reclaimed. However, volume
unmapping operations can stop new writes to the volume and adversely aect performance. This scenario is most likely with
replicated volumes. Therefore, Dell recommends that you not use volume unmapping with replicated volumes. Also, running mkfs,
defragging, or deleting les on le systems mounted using the -o discard option can generate large numbers of unmap
operations, which are not recommended for replicated volumes.
Unmapping with Windows Servers 2012 and 2016
By default, unmapping is enabled in Windows Servers 2012 and 2016, which automatically detects a volume’s provisioning
capabilities, including whether or not the volume can process unmap operations. When data is deleted from a volume in Windows,
the corresponding space on the PS Series storage array volume is also freed automatically, thus maximizing the eciency of the
array.
To disable unmapping in Windows Servers 2012 and 2016, issue the following command:
fsutil behavior set disabledeletenotify 1
To reenable unmapping, issue the following command:
fsutil behavior set disabledeletenotify 0
To check the current setting of unmapping, issue the following command:
fsuitl behavior query disabledeletenotify
NOTE: The disabledeletenotify setting is a global operating system setting that not only disables unmap
operations from being sent to the PS Series storage arrays, but also disables TRIM to SSDs. For more information about
the fsutil utility, see: technet.microsoft.com/en-us/library/cc785435(v=ws.10).aspx
Automatic defrag can generate a large number of unmap operations, which are not recommended for replicated volumes. Therefore,
you should turn o automatic defrag for replicated volumes.
Set a Volume or Snapshot with Lost Blocks Online
In rare circumstances, a volume (or snapshot) might lose blocks. For example, loss can occur if the power fails and then a control
module cache battery fails. (If the control module cache battery is the only power source for a control module for more than 72
hours after a power failure occurs, the battery can fail.)
If a volume (or snapshot) loses blocks, the current status of the volume (or snapshot) is offline-lost-cached-blocks. In
addition, the group generates an event message.
You can choose to set the volume online but retain the lost blocks. If an application tries to read a lost block, an error occurs. If an
initiator writes new data to a lost block before it is read, the block is no longer lost. The members containing lost blocks have a status
of RAID lost blocks until initiators write to all the lost blocks.
About Volumes
155