[HPADM] SUMMARY2: How to Move Some Free PE's to Another VG

From: Dave T. (davidlt77_at_hotmail.com)
Date: 05/21/03

  • Next message: Mc Nellis, Dennis J (Dennis): "[HPADM] LUN using FC60"
    To: blhconsulting@mindspring.com, hpux-admin@dutchworks.nl
    Date: Wed, 21 May 2003 08:11:33 -0400
    
    

    After I sent my previous summary, Bill Hassell provided some enlightening
    information. I've included it below.

    It turns out that the VG wasn't originally configured to handle more disks,
    so we can't add any more. We are migrating to a VA, so we will allocate
    more space to the VG before we begin migration.

    Thanks for your help, Bill!

    Dave

    ----Original Message Follows----
    From: Bill Hassell <blhconsulting@mindspring.com>
    To: "Dave T." <davidlt77@hotmail.com>
    Subject: Re: RE: [HPADM] How to Move Some Free PE's to Another VG
    Date: Tue, 20 May 2003 13:03:15 0200 (UTC)

    Hi,

       The document is accurate and works very well. However, it
       states clearly at the beginning that this is used to move
       extents around in a single volume group. For instance,
       to improve performance, you might take a busy lvol and move
       it to another PV (physical disk). But the disk must already
       be part of the same volume group. pvmove is an atomic
       operation which means that the system can be very busy and
       the move will always work OK, albeit slowly since each move
       is carefully monitored to ensure that the data is completely
       intact and not changed during the move.

    What I'm considering is moving PV's from vgpd to vgpd1. Although we have
    a
    lot of free PV's on vgpd, all those PV's currently have extents in use.
    However, I want to try to do the following:

    1. Move the data from some of the PV's to other PV's within vgpd using
    pvmove.
    2. Remove the free PV's from vgpd.
    3. Add the free PV's to vgpd1.

       The problem is that if you look at the pvmove command, there
       is no option to specify a disk in another VG. It would make
       no sense to do this since splitting \extents among different
       VGs would mean that soime central location keeps track of all
       the extents for all the disks and the purpose of a VG is no
       longer valid.

    Following are the contents of a document from the ITRC:

    How to move logical volumes from one disk to another using pvmove DocId:
    KBRC00003756 Updated: 9/29/00 5:26:00 AM

    PROBLEM
    This document explains how to move all the data on a LVM disk to a
    different
    disk in the same volume group.

       Notice the "same volume group"?

       So it isn't risky, it's impossible. Buy another disk or two.

    NOTE: If the original VG was not created with enough extents
    to handle additional disks, adding the disk will not give you
    all the space. You are limited to 65k extents total. When you
    first create a VG, the extent size is adjusted automatically
    (larger than 4 megs) to accomodate all the disk that you add
    to the VG. But if you add more disks later, the extent size
    may be too small to count the new space and you *must* rebuild
    the VG from scratch, which means backup the data, remove the VG,
    then rebuild the VG and specify an override for the extent size
    large enough to handle more disk in the future (ie, 16 or 32 meg
    extents if the VG will grow to terabyte size).

    Bill

    --
    Best regards,
    Bill Hassell (blhconsulting@mindspring.com)
    _________________________________________________________________
    Tired of spam? Get advanced junk mail protection with MSN 8. 
    http://join.msn.com/?page=features/junkmail
    --
                 ---> Please post QUESTIONS and SUMMARIES only!! <---
            To subscribe/unsubscribe to this list, contact majordomo@dutchworks.nl
           Name: hpux-admin@dutchworks.nl     Owner: owner-hpux-admin@dutchworks.nl
     
     Archives:  ftp.dutchworks.nl:/pub/digests/hpux-admin       (FTP, browse only)
                http://www.dutchworks.nl/htbin/hpsysadmin   (Web, browse & search)
    

  • Next message: Mc Nellis, Dennis J (Dennis): "[HPADM] LUN using FC60"

    Relevant Pages

    • Re: Do extents impact speed?
      ... My point was that - and I'm just making surmises here based on what very ... We only have one or two mirrored disk pairs so ... > Subject: Re: Do extents impact speed? ... > at all the above applies to cached SANs). ...
      (comp.databases.informix)
    • RE: Do extents impact speed?
      ... It is exactly because of the advances in technology,memory and disk size, ... Subject: Re: Do extents impact speed? ... > and forward, increasing the lag time. ... at all the above applies to cached SANs). ...
      (comp.databases.informix)
    • Re: Are multiple extents still a problem with LMT?
      ... degraded performance in queries that have to go to disk. ... thousand extents. ... Will LMT eliminate most of the effects of extent ... extent sizes (whicih you've done already in a DMT configuration so I ...
      (comp.databases.oracle.server)
    • RE: Do extents impact speed?
      ... this many extents will almost certainly impact speed for the worst. ... there would be a significant increase in disk seek time. ... Even the Performance Guide says: "Performance suffers when disk seeks for a table must span more than one extent, ... Kansas City Informix Users Group ...
      (comp.databases.informix)
    • Re: Slow Filesystem I/O
      ... > Bill Todd wrote: ... >>it onto the disk as compensation, ... Spiralog did not change *default* file system behavior in any ... when RMS specified that its internal buffer should be ...
      (comp.os.vms)