Re: transfer OS from failing HD questions

[Ray Olszewski <ray@comarre.com>]

James Miller wrote:
> Here's a question I see as alot more newbie-oriented than memory 
> functions, (m)allocs and the like. I recently had strong indications 
> that my HD was failing in my Ubuntu (Debian variant: or is Debian now an 
> Ubuntu variant?) machine. Pondering over what to do about saving data 
> from that machine and having to do the minimal amount of reinstallation, 
> I came up with the following, seemingly quite precarious plan: I would 
> get a new HD, do a fresh install of Ubuntu to it, then stick the old 
> drive in as slave, boot from a rescue CD, and copy the entire contents 
> of the old drive to the new one, replacing any files on the new one that 
> have the same name as those on the old one. Both the new install and the 
> installation on the old HD were the same Ubuntu release, btw. Yes, it 
> looks like a real kludge--not pretty, and maybe not even effective. It 
> took a really long time. The new Ubuntu install was over in about 20 
> minutes--the smoothest part of the undertaking. Transferring the 70 GB 
> of data from the old drive to the new one took many hours. I was getting 
> pretty poor drive access rates for one thing, maybe due to the fact that 
> the old drive is on its way out. Then, I would occasionally get I/O 
> errors on the old drive where a file couldn't transfer. The more of 
> those I got, the more sure I was that this approach wouldn't work in the 
> end. But when I finally finished copying over everything, guess what? 
> The system booted right up. It looked and acted just like the system did 
> when the old HD was functioning properly. I have yet to run into any 
> system glitches, but I've only been running it for a couple of days now. 
> On to my questions.
> 
> 1) What sort of problems might I expect to crop up as I continue to use 
> this system? Doubtless some binaries or other types of system files were 
> among the ones that wouldn't transfer because of I/O errors. 2) What 
> other approach might I have taken to essentially preserve the existing 
> installation (both configuration and data) on a new HD?
> 
> Input will be appreciated.

This is pretty close to what I've done on my Debian (usually Sid, 
sometimes Sarge) systems. But I haven't had a failure of a root 
filesystem in ages, so I may be forgetting some of the details. 
(Transferring any filesystem other than root is, of course, trivial, 
unless you use a distro that does that convoluted partitioning into 
separate partitions for /usr, /tmp, and I-forget-what-all-else.)

A couple of tweaks that might improve the process:

1. Make sure DMA is on for both drives. Use hdparm to do this.

2. Installing the old drive as secondary master, instead of primary 
slave, should give you slightly better transfer rates.

3. You don't actually need a rescue disk to do this; you can boot from 
the new primary master. You just want to be sure not to cp over a few 
directories ... /proc, /tmp, /var/log, maybe a couple of others I'm not 
thinking of ... probably some odds and ends in /etc ... from the old 
drive. Maybe the rescue-disk approach is simpler after all.

My usual practice these days is to set up my primary drive with separate 
partitions for root (/), /boot, and /home. Then if the drive with my 
root partition does fail, I just do a fresh, complete reinstall to a new 
drive, and only transfer over /boot (unless I decide to compile &/or 
install a new kernel), /home, and the small number of files I've 
customized from the install in (for example) /etc.

-
To unsubscribe from this list: send the line "unsubscribe linux-newbie" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.linux-learn.org/faqs