You are on page 1of 4

ADMINSTRATE MIRROR VOLUMES

Provides redundancy.
Improves concurent read perfomance
Requires plex synchronization (in background)
Only volumes with concatenated or striped plexes
By default, mirror created with the same layout as original plex
Each mirror on separated disks
Alls disks in the same disk group
MIGRATE DATA TO A NEW ARRAY
1-Set up LUNS on the new array.
2-Get the OS to detect the LUNs.
3-Get VxVM to recognize the LUNS.
4-Initialize the new LUNs.
5-Add the LUNs to the same disk group as the original volume.
6-Mirror the volume using the new LUNs.
7-Wait for the synchronization to complete.
8-Set volume read policy to the plex on the new LUNs
9-If satisfied, remove old plex.
10- Remove old LUNs from the disk group
ADDING / REMOVING
Add a mirror:
vxassist -g diskgroup mirror volume_name layout=layout_type dm_name
vxassist -g appdg mirror appvol
Removing a mirror:
vxassit -g diskgroup remove mirror volume_name [!dm_name]
vxplex -g diskgroup -o rm dis plex_name
vxassist -g appdg remove mirror appvol
vxplex -g appdg -o rm dis appvol-02
DIRTY REGION LOGGING (solo para volumenes mirror)
(despues de un crash organiza todas las plexes que queden identicas
al sincronizar no ocupa sincronizar todo porque Dirty Region loggin sabe
en donde se estaba grabando y cual se debe sincronizar)
Only in mirrored volumes
For quick recovery in sistem crash
OFF by default
Multiple logs for redundancy
1-The DRL updates dirty region bits
2-If a system crashes during I/O
3-Volume recovery is performed for those regions only
ADDING AND REMOVE A DIRTY REGION LOG (DRL) TO A VOLUME
vxassist -g diskgroup addlog volume_name \
[logtype=drl] [nlog=n] [attributes[
vxassist -g appdg addlog appvol logtype=drl
vxassist -g appdg removelog appvol
puede especifcarse el tipo de log y cuantas replicas
VOLUME READ POLICIES
-Round robin= if al the plexes are the same

-Preferred plex= explicy use one plex, all read from one single plex
-Selected plex= Voumen manager select where to read
-Site read=Cluster or conjuntion o sites
Read local site for example or selecet array
Select any number of plex
SET THE VOLUME READ POLICY
vxvol -g diskgroup rdpol policy volume_name [plex]
policy= round | select | prefer | siteread
vxvol -g appdg rdpol prefer appvol appvol-02
vxlist vol appvol | grep -i policy
Read Policy: PREFER
RESIZING A VOLUME AND A FILE SYSTEM
Performed online
Increase or decrease volume size
-Disk group have available space
-VxVM rules on space allocation apply
Volume must have available space
Order of volume and file system resize is important
METHODS RESIZING A VOLUME AND FILE SYSTEM
-VOM: Both
-VXRESIZE: Both
-VXASSIST: Volume only
-FSADM: VxFS only
vxresize [-bxs] [fstype] -g diskgroup \
volume-name [+|-] new_length
vxassist -g diskgroup maxgrow volume_name
vxresize -g mydg myvol 5g ->set size
vxresize -g mydg myvol +1g
vxresize -g mydg myvol -1g
RESIZE A VOLUME MANAGER DISK \
TO MATCH A RESIZE LUN
Intented for LUNS that are part of an imported disk
Requires if the LUN is resized in the hardware
vxdisk [-f] -g diskgroup resize dm_name
vxdisk -g appdg resize appdg01
MOVING DATA BETWEEN SYSTEM
-Deporting a disk group:
Stop application
Umount file system
vxdiskadm: remove access to (deport) a disk group
vxdg [-n new_dg_name] [-h hostname] deport diskgroup
Optionally:
Specify a new host

Rename the disk group


After deporting
The disk gruop is unavailbale
The disk are still visible to VxVm
-Importing a disk group:
vxdiskadm: Enable access t0 (import) a disk group
vxdg [-ftC] [-n new_name] import diskgroup
vxvol -g diskgroup start all (pre-5.1SP1)
-t opcion es temporalmente importado, si se reinicia queda en estado deportado p
ara importar en donde sea
After:
Start volumes (if necesary)
Mount file system
Start application
You can:
Specify a new disk gruop
Clear host locks
Import as temporary
Force an import
RENAMING VXVM OBJECTS
Online operation
Object names must be unique withing the DG
Related objects must also be renamed
Posible impact on FS or App
vxedit -g diskgroup rename old_name new_name
vxedit -g appdg rename appdg01 appdf03
RENAME A DISK GROUP
vxdg -n new_dg_name deport old_dg_name
vxdg import new_dg_name
or
vxdg deport old_dg_name
vxdg -n new_dg_name import old_dg_name
DISK GROUP VERSIONS AND FORMATS
Each DG has a version related to software
The private region is different en each version
SF release
5.0 and 5.0 MPx
5.1
5.1 SP1
6.0

Disk group ver.


140
150
160
170

vxdg [-T version] upgrade diskgroup


CDS DISK GROUPS
CDS attribute: cds=on
DG version =110 (or higher)

Supported DG version
20-140
20-150
20-160
20-170

Used for seamless transfer of data between different platforms


Created by default starting VxVM 4.x
Cannot contain non-CDS disks
The disk group must be in good condition
Disk groups can be converted while online or offline
vxcdsconvert [-A] -g diskgroup [-o novolstop] group
vxcdsconvert -g appdg -o novolstop group

You might also like