<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
<head>
<meta content="text/html;charset=ISO-8859-1" http-equiv="Content-Type">
</head>
<body bgcolor="#ffffff" text="#000000">
This might make it more clear:<br>
<br>
MB SATA<br>
\> sda (sdb is the same)<br>
\>1 -> md0 (boot)<br>
\>2 -> md1 (VolGroup00 - LogVol01)<br>
\>3 -> md2 (swap)<br>
\> sdc (sdd sde are same)<br>
\>1 ->md3 (LogVolRD5) <br>
\> sdf (sdg is same)<br>
\>1 ->md4 (raid level 0 for temp backup space)<br>
[problem] \>2 ->was raid0 md5, now two separate physical
volumes of VolGroup00 with LogVol01 snapshots striping to sdg2 also<br>
\>3 ->md6 (RAID0 physical volumes for snapshots
of LogVolRD5)<br>
<br>
PCIe SATA<br>
\> sdg (same layout as sdf)<br>
<br>
So the problem is that the kernel gets all bitchy that it can't
properly find the device with UUID of either the md5 device(when it was
configured) or the sdg2 physical volume when using LVM striping. I'm
actually just noticing that this might be the sdg device.<br>
<br>
When I boot off of the Fedora 9 rescue media, it sees everything
perfectly. I can manipulate the drives and volumes however I want.
But it won't boot if sdg2 is part of something that is part of
VolGroup00.<br>
<br>
So is it possible that the PCIe card isn't loading soon enough for
VolGroup00? <br>
<br>
<br>
Greg Freemyer wrote:
<blockquote
cite="mid:87f94c370806020549p3babd7e6la16a26055fd70644@mail.gmail.com"
type="cite">
<pre wrap="">On Mon, Jun 2, 2008 at 3:42 AM, Brian W. Neu <a class="moz-txt-link-rfc2396E" href="mailto:ale@advancedopen.com"><ale@advancedopen.com></a> wrote:
</pre>
<blockquote type="cite">
<pre wrap="">to store snapshot data, I put a physical volume on top of a raid 0 and
extended the volume group onto it.
the problem is that as far as boot sequences go, it looks like only raid
1 mirrors are started before lvm -- and since a physical volume is
missing (since the raid0 hasn't been initialized), the volume group
doesn't start -- hence the machine doesn't boot.
Does anyone know a way to get the kernel to somehow do this, or is it
truly impossible?
</pre>
</blockquote>
<pre wrap=""><!---->
I don't know the answer, but why are you doing it in the first place?
Seems like you're loosing most of the management advantage of LVM by
putting raid 0 below it.
If speed is not critical, you could use LVM to add the drives
individually. Not sure if you can setup stripping with lvm or not.
If not, maybe you should request that as a feature enhancement from
the lvm team.
Greg
</pre>
</blockquote>
</body>
</html>