<br><div><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">Thanks. GOOD POINT. Just after I read all the propaganda on the LVM howTo<br>about why I want LVM too.
<br><br>But you do have the option of keeping each LV on a separate PV. It's your<br>choice when a filesystem grows bigger than one PV -- either enlarge the LV<br>across PV's or make a new LV and filesystem for each new PV.
<br><br>Did I just lie? If you have ten PV's each with it's own LV and filesystem<br>and you lose one PV, is LVM robust enough to continue accessing the<br>remaining LV's and allow you to replace the failed LV?<br></blockquote>
</div><br>Well, if you really care about data integrity, build your LVM up over RAID-1 mirrors (which is what I've done). Yes, it's more expensive, but if a drive fails, you can:<br><br>1) Buy a new drive, add it to the mirror, and deactivate the old one.
<br>2) Buy two new drives, create a new RAID-1, add it to the LV and move all the data off the old mirror.<br><br>Personally, I'd opt for option two (since you'll likely end up with bigger drives), but either way, it's the best way to go for a robust, scalable storage solution, IMHO.
<br><br>Otherwise, yes, if you lose a PV in an LVM, you're in trouble... it's no different than RAID-0, in that sense, AFAIK.<br><br>Brett.<br>