Khalid Schofield wrote:
> On 1 Sep 2008, at 22:41, johan beisser wrote:
> 
>> On Sep 1, 2008, at 11:44 AM, Khalid Schofield wrote:
>>
>>> Hi,
>>> I'm running openbsd 4.0 (yeh old I know but it's a vital system  
>>> that I'm replacing but it processes data that makes a lot of money).
>>
>> Better replace the disk tomorrow, then. Or, implement the software  
>> on a new system, and take the hit on some downtime while it's being  
>> replaced.
>>
> 
> Thanks for the tip. Just bought one.
> 
>> Those are signs of odd errors on the physical media itself. OpenBSD  
>> can (and may) crash due to bad sectors and failed writes. I did  
>> allow a system to limp along on a bad drive for nearly a year while  
>> I tried to source a very old (no longer available) drive.
>>
> 
> The old disk is a 40Gb IDE disk and the new one one is a 120Gb disk.  
> If I want to clone the disk can I just cat /dev/sd0 > /dev/sd1 if I  
> boot off the install cd? I will rebuild this system on another box but  
> to make sure the disk doesn't die instantly I want to clone it asap  
> since it makes cash.

if it is making money for you, you need to have a much better
maintenance plan, which includes maintenance and upgrade plans.
Otherwise, your "money making" plans are temporary.

IF you can read the whole disk:
   dd if=/dev/rwd0c of=/dev/rwd1c bs=1M

will image the disk to another disk.
Replace "/dev/rwd1c" with whatever is appropriate -- another disk,
a file, whatever.  You can often do this "live", at the cost of an
fsck on the new disk (which is often not fatal...depending on the
app, of course).

Odds are, however, you are going to have a fatal read error, so
you will probably need to copy files rather than the entire disk
image.

Personally, I'd just toss together another machine, put OpenBSD 4.3
or 4.4 on it (ok, I'd 4.4 on it, you would put 4.3 on it :), and
migrate your apps over to it.

> I've got another system on the boil with a 36Gb 15k scsi disk and  
> decent hardware but I want to keep this mac mini server going just  
> long enough to role the new server out.

I'm not impressed.  It is not the bling of your hardware that's a
problem, it is the maintenance plan.  By pointing out the specs of
your hardware, I don't think you have learned your lesson.

You will do far better with TWO cheap machines than one expensive
machine.  This way, you can actually test your upgrade process on
a regular basis, and keep this "money maker" running securely, and
deal with hardware failures trivially.

Getting a system running is not the sign of good system
administration.  The trick is having a plan in place to KEEP it
running and maintained even when the expected happens.  Replacement
of the OS, the hardware, the apps all need to be part of the
initial plan.

Nick.

Reply via email to