[prev in list] [next in list] [prev in thread] [next in thread] 

List:       evms-devel
Subject:    Re: [Evms-devel] trying to understand EVMS application
From:       Kevin Corry <kevcorry () us ! ibm ! com>
Date:       2004-10-23 20:56:31
Message-ID: 200410231556.31016.kevcorry () us ! ibm ! com
[Download RAW message or body]

Hi Eric,

On Friday 22 October 2004 19:38, Eric S. Johansson wrote:
> disk drives: sda: 9 GB SCSI; sdb; 18 GB SCSI; hde/g: 80 GB IDE
>
> desired layout:
>
> hde/g = raid 1 array.  To be treated as single contiguous storage with
> bad block handling
> partition: /1-raid1

Simple enough. In evmsn or evmsgui, create a BBR segment on each of the IDE 
disks. You'll wind up with hde_bbr and hdg_bbr. Then create a RAID-1 region 
using these two BBR segments, and you'll wind up with md/md0. You can make 
this RAID-1 region directly into a volume, or if you want to be able to 
subdivide the RAID-1, you'll want to create an LVM container using the RAID-1 
region. After that, create one or more LVM regions from the container, and 
make each LVM region into a volume. Once you have your volume(s), create a 
filesystem on each volume.

> sda+sdb: effectively one contiguous linear array with the capability of
> adding and removing disks if necessary (i.e. drive starts going bad, add
> new drive, transfer data to new drive, remove bad drive).
> additional goals also bad block handling and ability to resize
> partitions if I guess wrong on partitions size.  Resizing does not need
> to be live but it would be nice.
>
> partitions: /boot, /, /usr, /var, /var/log

Do you want all of these filesystems on the SCSI disks? Or will some be on the 
IDE disks?

First of all, you'll need a normal DOS partition (no BBR or LVM) for /boot, in 
order to be able to use Grub or LILO. So you'll first want to assign the DOS 
segment manager to sda. Then create one 100MB segment (or however big you 
want /boot) from the freespace on sda to get sda1. Create an EVMS volume from 
sda1 and add a filesystem to that volume. 

Now for the rest of the volumes. Create a second segment on sda that takes up 
the remaining freespace to get sda2. Next, create BBR segments using sda2 and 
sdb, and you'll get sda2_bbr and sdb_bbr. Next you'll want to create an LVM 
container using sda2_bbr and sdb_bbr. Then you can create four LVM regions in 
that container (one each for /, /usr, /var, and /var/log). Make each of these 
regions into EVMS volumes and add filesystems.

The LVM container lets you add additional disks in the future to expand the 
freespace, remove disks that aren't in use by any of the regions, and easily 
resize the LVM regions. See http://evms.sf.net/user_guide/#appxlvm for all 
the details about the LVM plugin.

Whether you can resize online depends on the filesystem you're using. Ext3 can 
only expand and shrink when unmounted. ReiserFS can expand when mounted or 
unmounted, but can only shrink when unmounted. XFS and JFS can expand only 
when mounted, but neither can shrink at all.

> in my previous attempt, I was having trouble specifying partitions so
> that they would boot.  But we will deal with that later.

In order to have your root fs on an EVMS volume, you need an init-ramdisk. See 
http://evms.sf.net/install/root.html for all the gory details. You'll also 
want to look at http://evms.sf.net/install/boot.html, especially if you're 
going to be using LILO.

Let me know if you have any more questions.
-- 
Kevin Corry
kevcorry@us.ibm.com
http://evms.sourceforge.net



-------------------------------------------------------
This SF.net email is sponsored by: IT Product Guide on ITManagersJournal
Use IT products in your business? Tell us what you think of them. Give us
Your Opinions, Get Free ThinkGeek Gift Certificates! Click to find out more
http://productguide.itmanagersjournal.com/guidepromo.tmpl
_______________________________________________
Evms-devel mailing list
Evms-devel@lists.sourceforge.net
To subscribe/unsubscribe, please visit:
https://lists.sourceforge.net/lists/listinfo/evms-devel
[prev in list] [next in list] [prev in thread] [next in thread]