1

[other] MDADM RAID rebuilds incorrectly on reboot

view story
linux-howto

http://ubuntuforums.org – So my computer has four 1.5 TB drives installed. Up until recently, these drives existed in two RAID1 configurations of 3 TB. Recently I decided to create one large RAID5 device out of them instead. I did this remotely, and had no spare drive to dump data to - so I disassembled one of the RAIDs (it was just a backup of the other from a few weeks back) and since the RAID contained under 1.5 TB of data - I created a partition on one of the disks and stored it there. So I created the RAID5 in a degraded mode with only 3 disks and copied the data back on to the new 4.5 TB RAID. After that, I (Hardware)

colm's picture
Submitted by colm on 09/07/2012

Stories similar to [other] MDADM RAID rebuilds incorrectly on reboot

I previously had a raid5 with mdadm using four 2TB drives.

HP DL 380G7 Raid swap drives 38 weeks 6 days ago

disks 0 and 1 are raid redundant (OS) and the remaining 6 drives are in raid 5 I believe. i would like to pull out disks 0 and 1 and install new drives to build a new OS. I need to be able to reinsert the old drives and reboot back to the original OS. (swap) I had a serious problem in the past attempting this.

I have 6 1tb drives in a RAID5. 1 drive went down. On the RAID was 2 virtual machines that I really need back up and running. The spare drive I have to put in the server is a 1.5tb drive, which exceeds the physical per drive limit of the 2020SA. The drive is found in the disk utility, but it is not found in the array management section. I cannot add the drive to the array to have it rebuild.

i recently start working in a new (small) company having an "old" file server with windows server 2003 and an hw raid 5 with four disks. Last week the server is dead. We decided to build up a newer system with windows server 2008 using a PC with this hardware : CPU: Core i3 3,2 GHz RAM: 2GB Motherboard supports Raid.

I created a raid 1+0 via mdadm by the following...

mdadm -v --create /dev/md2 --level=raid1 --raid-devices=2 /dev/sde1 /dev/sdf1mdadm -v --create /dev/md3 --level=raid1 --raid-devices=2 /dev/sdg1 /dev/sdh1

mdadm -v --create /dev/md1 --level=raid0 --raid-devices=2 /dev/md2 /dev/md3 ...then finished it off by adding the output of mdadm --detail --scan to /etc/mdadm/mdadm.conf and did a mkfs.xfs on

A friend's machine running Windows XP refused to boot recently which is running 3 SATA disks on RAID 5 (which was previously upgraded from RAID 1 not by me). I have determined there to be a disk failure. The disks have been replaced many times in the past few years. I wish to backup the RAID5 partition before I try anything to fix it.

When I go to bring up the array with mdadm, it complains I don't have enough components to start it. This makes sense, since RAID5 normally takes a minimum of three disks.. but I don't want to buy another 2TB drive with prices what they are at the moment..

So I have a system (a large instance running ubuntu-11.10 on Amazon EC2) where I've done something like

mdadm --create bigdata --level 0 --raid-devices=2 /dev/xvdb /dev/xvdc

I get a RAID device of two disks (ephemeral instance stores) at /dev/md/bigdata. Yay. reboot. I now have a RAID device of two disks at ip-10-811-498-737:bigdata. (ip anonymized).

This is kind of obnoxious.

So here is what I have currently:

8 SAS Drives total 6 Drives setup as RAID 5 as Virtual 1. 1 Drive setup as Hot Spare. 1 Drive in Ready state that is not part of the RAID and is not set as a hot spare.

Basically, everything on this server is setup on the RAID 5 as one virtual disk (OS, data, etc) and partitioned out. I want to separate things out a bit.