vulnerability during partition-size-growth procedures
From: Brendan Hide <hidden>
Date: 2012-07-23 05:45:47
Hi all In the wiki (https://raid.wiki.kernel.org/index.php/Growing#Expanding_existing_partitions) there's mention of increased vulnerability to failure when growing an array because disks have to be failed before additional disks can come into play. Are there plans to allow a disk migration/clone feature in future? If not having this feature built-in, would dd (or GNU ddrescue) perhaps be used to do this manually? The only down side I can see to doing it manually is that it would likely require downtime. See below. 2-disk RAID0 (Simplest for illustration purposes, also impossible to do with current procedure): 02468AC <- disk0 13579BD <- disk1 Introduce a new larger disk intended to replace disk1 and sync the data across: 02468AC <- disk0 13579BD <- disk1 13579BD------- <- disk2 Fail/remove disk1: 02468AC <- disk0 13579BD------- <- disk2 The current procedure has us degrade the array before we put any data onto the larger disk2. -- Brendan Hide +27 83 448 3867 http://swiftspirit.co.za/