[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: High Availability... again



On Wed, 11 Oct 2000, Russell Coker wrote:

> On 2000-10-11 05:49, Roger Abrahamsson wrote:
> >First of all, thanx for all the answers, I've gotten a bit further now.
> >One thing I'm thinking of.. Does anyone know any hardware or software for
> >Linux that allows an existing RAID system to be grown/shrunken later on?
> >Would be a shame having this nice system up and running and then stop when
> >we run out of disk space..
> 
> LVM does growing and shrinking.  I believe that it has (or is destined to 
> have) mirroring support (which is far from full RAID support).  I doubt that 
> LVM is what you want to use for RAID type functionality.
> 
> For changing the size of a software RAID partition in Linux you would want to 
> have all your storage as a single software RAID, run LVM on top of that, and 
> then use LVM partitions for all file systems.  This allows you to dynamically 
> grow and shrink file systems within the same storage pool.
> 
> However if you want to add extra hard drives to add extra space then I think 
> you currently can't do it.  Let me know if this is what you require and I'll 
> investigate it further.
> 

Well, was not thinking of software raid here. However LVM seems to be what
I need, as it apparently seems to work well with hardware raid. I can grow
and shrink filesystems residing on several different raid systems as they
will appear as only one disk each.. Now if the tcp/ip stack in 2.4.x
kernels will be reentrant perhaps I could handle some heavy load on this
too.

As I understand this I will be able to add new things as I go, as one
machine always can be shut down, added more raid cards to and then change
the LVM settings on it when it has become the active box. Then repeat
that for the other... 

/Roger


-------------------------------------------------
Roger Abrahamsson, Sys/Net Admin, Obbit AB
Radhusespl.17D, S-90328 Umea, Sweden
-------------------------------------------------



Reply to: