From mboxrd@z Thu Jan 1 00:00:00 1970 From: Gavin McCullagh Subject: Re: LINEAR RAID, little help Date: Sat, 7 Apr 2007 18:33:52 +0100 Message-ID: <20070407173352.GA23645@gmail.com> References: <46177DD8.9080707@pcfusion.co.uk> Reply-To: Gavin McCullagh Mime-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: QUOTED-PRINTABLE Return-path: Content-Disposition: inline In-Reply-To: <46177DD8.9080707@pcfusion.co.uk> Sender: linux-raid-owner@vger.kernel.org To: Rich Cc: linux-raid@vger.kernel.org List-Id: linux-raid.ids Hi, The main reason I'm posting (given others can answer these questions better) is to ask a further question:=20 Why would anyone use RAID-linear? If RAID-0 gives better performance f= or the same (reduced) reliability, what's the point of using Linear? Do y= ou get slightly more space out of it? I presume it's for stuff where you don't place much value no your data. On Sat, 07 Apr 2007, Rich wrote: > I currently have a linear RAID setup via mdadm with is made up of 3=20 > drives. I just have a few questions that I can't seem to find searchi= ng=20 > around on Google, etc. I must admit I've never used linear raid. May I ask what made you choo= se it over say raid-0? > First question, what happens if one drive fails (I know I will loose = the=20 > data on that drive) but how, if at all can I recover the data on the=20 > other drives can I plug them in (on their own) as though they were in= =20 > fact an individual drives? Do I need to execute a rebuild command in=20 > mdadm at all to rebuild the array? http://tldp.org/LDP/EVMSUG/html/characraidlvls.html I'm sure someone else will give you a better answer but it would appear from these docs that your situation is similar to that of a person usin= g RAID-0 who has a drive failure. Basically, you end up with a big porti= on of your filesystem data missing so you may lose everything. I guess yo= u might be able to recover some data using some filesystem recovery tools= but it's likely to be a pretty unreliable process and likely to vary from o= ne filesystem to the next. It's probably a little more reliable than RAID= -0 in that what's left is probably contiguous but I would certainly not wa= nt to rely on such a process for my data. The faq says similar "RAID-linear is a simple concatenation of partitio= ns to create a larger virtual partition. It is handy if you have a number small drives, and wish to create a single, large partition. This concatenation offers no redundancy, and in fact decreases the overall reliability: if any one disk fails, the combined partition will fail." http://tldp.org/HOWTO/Software-RAID-0.4x-HOWTO-2.html > Second question, how can I go about adding a drive to my linear RAID,= I > wish to add two new 500GB drives but I'm unsure how. I have found how= to's > for RAID 5 but I just wanted to check it was a similar process for > Linear? Also this won't effect any data currently on the drive will i= t? You don't say what you're doing on this array, but before modifying it,= I'd be seriously inclined to question whether RAID-linear is really the rig= ht thing to be using at all. Anyway, the mdadm manpage says=20 "Currently the only support available is to =B7 change the "size" attribute for RAID1, RAID5 and RAID6. =B7 increase the "raid-disks" attribute of RAID1 and RAID5. =B7 add a write-intent bitmap to any array which support these bitmaps= , or remove a write-intent bitmap from such an array." which suggests you can't. That might be wrong though as it sounds (to = me anyway) like linear would be one of the easier ones to implement grow f= or. I'm not sure. Gavin - To unsubscribe from this list: send the line "unsubscribe linux-raid" i= n the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html