https://community.hpe.com/t5/operating-system-hp-ux/lvm-issue/m-p/4407111#M352540 Hello Ravi,<BR /><BR />Here is what you can follow:<BR /><BR />View the current consistency recovery policy<BR /># lvdisplay /dev/vg01/data<BR />Consistency Recovery MWC<BR />Disable the MWC<BR /># umount /data<BR /># lvchange â M n â c y /dev/vg01/data<BR /># mount /data<BR />Verify that the change succeeded<BR /># lvdisplay /dev/vg01/data<BR />Consistency Recovery NOMWC<BR /><BR /><BR />*The MWC policy introduces some runtime overhead, but guarantees fast consistency<BR />recovery after a system crash, even for very large logical volumes. The MWC policy is<BR />appropriate for most data logical volumes, especially for very large logical volumes that<BR />would take a long time to resynchronize after a crash without the MWC/MCR.<BR />â ¢ The NOMWC policy requires no runtime overhead, but may take longer to recover mirror<BR />consistency after a system crash. The NOMWC policy is appropriate when runtime<BR />performance is critical, and recovery time is less of concern.<BR /><BR />NOMWC<BR />If you choose to use the NOMWC policy, and a system crash occurs, LVM will spawn<BR />nomwcsyncd daemons at activation time to resynchronize each mirrored logical volume that<BR />was open at the time of the crash.<BR /><BR />nomwcsyncd doesnâ t just synchronize portions of the logical volume that had pending writes<BR />at the time of the crash; it synchronizes the entire logical volume from top to bottom. For<BR />large logical volumes this may take a long time. Fortunately, the daemon runs in the<BR />background so system activity may proceed while LVM does the synchronization.<BR /><BR />â ¢ The NONE policy requires no runtime overhead, but leaves your mirrors in an<BR />inconsistent state after a system crash. This policy is often used for swap logical<BR />volumes.<BR /><BR />Extracts taken from LVM student's guide.<BR /><BR />Hope this helps !<BR />R.K. Fri, 24 Apr 2009 04:52:09 GMT R.K. # 2009-04-24T04:52:09Z https://community.hpe.com/t5/operating-system-hp-ux/lvm-issue/m-p/4407110#M352539 Hi All,<BR /><BR />I have old lvs which have Consistency Recovery as NOMWC. I have created a new vg and lvols in that vgs. However i am unable to set the Consistency Recovery as NOMWC . It is showing as MWC. This is a 2 node cluster.<BR /><BR />Should i use vgchange or lvchange.<BR />I did this:<BR />Turn the mirror write cache off on a logical volume:<BR /><BR />lvchange -M n /dev/vg01/lvol1<BR /><BR />Can some one help me on this with complete syntax.<BR /><BR />Also, please explain MWC NOMWC NONE options.<BR /><BR />~Ravi Fri, 24 Apr 2009 02:21:56 GMT https://community.hpe.com/t5/operating-system-hp-ux/lvm-issue/m-p/4407110#M352539 Nath_3 2009-04-24T02:21:56Z https://community.hpe.com/t5/operating-system-hp-ux/lvm-issue/m-p/4407111#M352540 Hello Ravi,<BR /><BR />Here is what you can follow:<BR /><BR />View the current consistency recovery policy<BR /># lvdisplay /dev/vg01/data<BR />Consistency Recovery MWC<BR />Disable the MWC<BR /># umount /data<BR /># lvchange â M n â c y /dev/vg01/data<BR /># mount /data<BR />Verify that the change succeeded<BR /># lvdisplay /dev/vg01/data<BR />Consistency Recovery NOMWC<BR /><BR /><BR />*The MWC policy introduces some runtime overhead, but guarantees fast consistency<BR />recovery after a system crash, even for very large logical volumes. The MWC policy is<BR />appropriate for most data logical volumes, especially for very large logical volumes that<BR />would take a long time to resynchronize after a crash without the MWC/MCR.<BR />â ¢ The NOMWC policy requires no runtime overhead, but may take longer to recover mirror<BR />consistency after a system crash. The NOMWC policy is appropriate when runtime<BR />performance is critical, and recovery time is less of concern.<BR /><BR />NOMWC<BR />If you choose to use the NOMWC policy, and a system crash occurs, LVM will spawn<BR />nomwcsyncd daemons at activation time to resynchronize each mirrored logical volume that<BR />was open at the time of the crash.<BR /><BR />nomwcsyncd doesnâ t just synchronize portions of the logical volume that had pending writes<BR />at the time of the crash; it synchronizes the entire logical volume from top to bottom. For<BR />large logical volumes this may take a long time. Fortunately, the daemon runs in the<BR />background so system activity may proceed while LVM does the synchronization.<BR /><BR />â ¢ The NONE policy requires no runtime overhead, but leaves your mirrors in an<BR />inconsistent state after a system crash. This policy is often used for swap logical<BR />volumes.<BR /><BR />Extracts taken from LVM student's guide.<BR /><BR />Hope this helps !<BR />R.K. Fri, 24 Apr 2009 04:52:09 GMT https://community.hpe.com/t5/operating-system-hp-ux/lvm-issue/m-p/4407111#M352540 R.K. # 2009-04-24T04:52:09Z