[H-GEN] Cloning Linux System

David Jericho david.jericho at aarnet.edu.au
Sun Jun 5 20:58:59 EDT 2005


Robert Brockway wrote:
> On Sat, 4 Jun 2005, Kelvin Heng wrote:
> 
> 
>>I need to clone a Redhat Linux (7.2) server to a secondary location 
>>(e.g. Another Server or USB HDD). But I have some restriction, I am not 
>>able to have any downtime for the server and will be required to carry 
>>out this task while the system is operational.
> 
> 
> This is an interesting problem - not being able to have any downtime.  

Yes, I agree completely. You haven't given us any information regarding
what type and the level of disk IO.

The issue really is it takes a certain amount of time to copy data from
disk to disk. If a file previously copied changes further along in the
copy operation, it won't be copied in the updated state. This could (and
probably will) lead to data inconsistency.

If you were running a newer version of Linux, I'd suggest you use the
snapshot feature of LVM on your important disks. The other option is to
use a RAID-1 like solution and just take the disk off line when you want
a backup snapshot. Mind you, neither solution solves the data
consistency problem, which has to be solved with extra planning and
software.

> If a system is so integral that you need its services all the time then it 
> has to be in some sort of hot swap, cluster or fail over situation.  The 
> best a serious high availability environment can really offer is 6 
> sigma[2] and this costs a lot of money.  Mucho $$$$.

> [2] Failure is 6 std deviations above the mean.  IIRC this is 5 minutes 
> downtime per year.

Otherwise known as five-nines, 99.999%. Five-nines is quite a feat and
if you've achieved it without redundancy planning on commodity hardware,
you should buy a lottery ticket.

Service availability != uptime, even though the two can coincide.

-- 
David Jericho
Systems Administrator, AARNet
Phone:     +61 7 3864 8379
Mobile:    +61 4 2302 7185





More information about the General mailing list