02-02-2011
Inherited VIO server an LPARs
Lucky me, someone has installed a server and got it running with the best intentions, but leaving me a headache.
We have a simple p520 with 4 disks. 2x145Gb & 2x300Gb. The smaller disk pair have been built into a VIO mirrored rootvg, and quite right too.
The other two disks form a volume group each (VIO storage pools). The plan was to ensure that disk allocated to the LPARs would be made in pairs, one from each user-volume group so that we would be confident at the LPAR that the hdisk definitions presented would be from different real media. The problem I now have is that we are adding more real disk (another 16x300Gb in a separate CEC) and add more LPARs.
So, should I create more volume groups on the VIO server (storage pools) on a one-disk-one-pool rule, or should I approach this another way. What would be nice would be a way to present the LPAR with a single device made from a mirror at the VIO server and enforcing strictness on the VIO server to make sure that the mirrors are on different real disks.
I feel that if/when we have a disk failure, I'm going to have a huge mess of failed hdisks in the various LPARs we will have and a huge overhead in logging and maintaining disk separation for all these LPAR hdisks and mirrors making sure that LPAR mirrors do not inadvertently end up on the same real disk as the primary copy, albeit a different hdisk that the LPAR sees.
Big question/headache I know. Any suggestions?
Robin
Liverpool/Blackburn
UK
10 More Discussions You Might Find Interesting
1. SCO
I have inherited admin duties on an old server running SCO Open Server 5. No changes have been made to this thing for a very long time. The server has started randomly disconnecting itself from the network. I have to reboot the server to get it to find the network again. It happens about once a... (3 Replies)
Discussion started by: jmickens
3 Replies
2. AIX
Hi,
I am little confused about the virtual Ethernet configuration on VIO and Client Partitions. There is alot of info on the internet but it gets more confusing....
If I have LHEA, it is very simple. Just assign LHEA (logical host ethernet adapter) to client partition -> run smitty tcpip and... (10 Replies)
Discussion started by: filosophizer
10 Replies
3. AIX
Hi,
I want to know wheather partition size for installation of vio client can be specified on vio server
example
If I am installing vio server on blade with 2*300gb hard disk,after that I want to create 2 vio client (AIX Operating system) wheather I can specify hard disk size while... (1 Reply)
Discussion started by: manoj.solaris
1 Replies
4. AIX
Hi
My vio client is getting its virtual Ethernet services from dual vio servers. What command if I execute on the vio
client will get me the names of the vio servers ? (2 Replies)
Discussion started by: samsungsamsung
2 Replies
5. AIX
trying to put VIO server software into this p520, firmware upgraded to the latest and greatest..SF240_415_318 , I think
a few questions, on this practice AIX machine on AIX 7.1
VIO can only be installed if ASMI or HMC is running ???
if so, perhaps ASMI is much simpler, since I will need to... (2 Replies)
Discussion started by: ppchu99
2 Replies
6. AIX
Hi
In the vio server when I do # lsattr -El hdisk*, I get a PVID. The same PVID is also seen when I put the lspv command on the vio client partition. This way Im able to confirm the lun using the PVID.
Similarly how does the vio client partition gets the virtual ethernet scsi client adapter... (1 Reply)
Discussion started by: newtoaixos
1 Replies
7. AIX
Hi,
I would like to know installing vio server on local disk and mirroring rootvg,
if I am creating AIX VIO CLIENT(lpar),
and any of single local hard disk failuare.
will it affect lpars? will lpars able to boot.
what needs to be done? (1 Reply)
Discussion started by: manoj.solaris
1 Replies
8. AIX
Hi,
I am facing an issue in vio server. When I run
bosboot -ad /dev/hdisk0
I am getting an error
trustchk: Verification of attributes failed: /usr/sbin/bootinfo
: accessauths
regards,
vjm
Please use code tags next time for your code and data. (8 Replies)
Discussion started by: vjm
8 Replies
9. UNIX for Advanced & Expert Users
Hello,
I would like to identify the blade or/and bladecenter from the hosted VIO server.
I prefer the "command line" solution. May be kdb.
Like, I want to ask the child about his father.
Thank you (1 Reply)
Discussion started by: x41
1 Replies
10. AIX
I have a broken PV in a VIO VG that's used to support client LPARs using LVs. On the client LPAR, I reduced all PVs from the relevant client VG and thus deleted it. I.e. there is no client LPAR using the VIO VG. Yet when I try to reducevg the VIO VG, it complains that the LV hosted on the PV is... (2 Replies)
Discussion started by: maraixadm
2 Replies
LEARN ABOUT HPUX
vxdiskadd
vxdiskadd(1M) vxdiskadd(1M)
NAME
vxdiskadd - add one or more disks for use with Veritas Volume Manager
SYNOPSIS
vxdiskadd disk_address_pattern_list
DESCRIPTION
The vxdiskadd utility configures disks for use by Veritas Volume Manager (VxVM). Disks must already have been formatted. vxdiskadd
prompts the user for a disk group name and disk name for the disks. If no disk group name specified, the disks will be left as unassigned
replacement disks for future use. A new disk group may be created that will become the disks' disk group. If a disk group is specified for
the disks, the user is prompted for whether the disks should be designated as spares for the disk group. For new disks, all space on the
disk becomes free space in the disk's disk group. vxdiskadd interactively allows initialization to be done for all the disks specified or
allows the user to ask to make the decision for each disk individually.
One or more disks may be specified with a disk_address_pattern_list. The basic form of a disk address is c#t#d#. Any leading portion of
this address may be specified to indicate that all disks that match that leading portion should be used. For example c2t0 may be used to
specify all disks on controller 2, target 0. More than one disk address or address pattern may be specified on the command line. In case
of enclosure-based names, the disk address will be of the form enclosurename_diskno. The enclosurename indicates that all disks in the
specified enclosure are used. For example, emc1_ is used to specify all disks in the enclosure emc1. The word all may be used to specify
all disks on the system. Disk address names relate directly to device node names in the /dev/dmp and /dev/rdmp directories. Here are more
examples:
all: all disks
c3 c4t2: all disks on controller 3 and controller 4, target 2
c3t4d2: a single disk (in the c#t#d# naming scheme)
xyz_0 : a single disk (in the enclosure based naming scheme)
xyz_ : all disks on the enclosure whose name is xyz
The files /etc/vx/cntrls.exclude, /etc/vx/disks.exclude and /etc/vx/enclr.exclude may be used to exclude controllers or disks from use by
vxdiskadd. Each line of cntrls.exclude specifies the address of a controller to exclude, for example, c2. Each line of disks.exclude spec-
ifies a disk to exclude, for example, c0t3d0. Each line of enclr.exclude specifies the name of an enclosure to exclude, for example, enc0.
All courses of action that do not involve clear failure conditions require prompts from the user, with defaults supplied as appropriate.
Help is made available at every prompt. Entering ? in response to a prompt displays a context-sensitive help message.
OPERATIONS
Reconnecting a drive that was temporarily inaccessible
This situation is detected automatically by noting that the specified drive has a disk ID that matches a disk media record with no
currently associated physical disk. After reconnection, any stale plexes referring the disk are reattached, and any stopped volumes
referring the disk are restarted. This reattach action is performed by calling the vxrecover utility.
Initialization of a disk with reserved regions
Initialization of the disk is performed by calling vxdisksetup.
Adding a disk to an existing disk group
This operation can be performed independently of the initialization of the disk drive. The operation adds the disk to the group so
that its storage is added to the free space pool in that disk group. The vxassist utility can subsequently allocate from that free
space.
Creation of new disk groups in which to import a new disk
If disks are added that are required to be put into a disk group that does not exist, the option of creating the disk group is
offered.
Hardware-Specific Note
Some environments provide guidelines to optimize VxVM's interaction with intelligent storage systems. If these guidelines are present, VxVM
follows the guidelines when adding disks to disk groups. If the operation fails due to these guideline(s), you are prompted to use the
force option. The force option ignores any such storage-specific guidelines.
FILES
/etc/vx/cntrls.exclude Specifies the address of controllers to exclude from vxdiskadd operations.
/etc/vx/disks.exclude Specifies the address of disks to exclude from vxdiskadd operations.
/etc/vx/enclr.exclude Specifies the address of enclosures to exclude from vxdiskadd operations.
NOTES
EFI disks cannot be added to a disk group that is compatible with the Cross-platform Data Sharing (CDS) feature, nor can they be formatted
as CDS disks.
SEE ALSO
vxassist(1M), vxdisk(1M), vxdiskadm(1M), vxdisksetup(1M), vxintro(1M), vxrecover(1M)
VxVM 5.0.31.1 24 Mar 2008 vxdiskadd(1M)