[H-GEN] any ideas for cluster filesystem solution ?

[Martin Pool]

On 10 Nov 2000, Craig ARMOUR <c.armour at uq.net.au> wrote:

> > RAID 5 will help with read performance (if that's a problem), and let
> > you cope with drive failures without needing to do server failover,
> > which on any system is a risk best avoided.
> 
> server failover is a risk?

> Single servers because you have raid5 is all
> good and well, but there is reasons for doing server failover, such as
> hardware and software upgrades and maitenance.  It's all good and well
> haveing rocksolid storage when your network card fails! 

Of course there's a place for server failover, but I would still
suggest it should not be a first recourse:

 * it's more complex than RAID and therefore more likely to cause
   problems 

 * you almost certainly drop the currently active requests, database
   connections, etc.

 * presumably the machines have two network cards anyhow

In any case, Byron has a more important point: the biggest risk is
having the two machines sitting next to each other in the same
building and network.

> I believe I had suggest something very similar 
> 
> ( ssh user at box1 ' cat httpd.log ' ; ssh user at box2 ' cat httpd.log ) | sort
> > newhttpd.log ??? 

The basic idea is OK, but there are two drawbacks.  Firstly, it's
astoundingly inefficient, given that the two logs are basically
already sorted already.  Secondly, it won't work, as the standard http
log format begins with an IP address, not a date.  So you should merge
rather than sort and use a comparator that understands dates.

-- 
Martin Pool, Linuxcare, Inc.
+61 2 6262 8990
mbp at linuxcare.com, http://www.linuxcare.com/
Linuxcare. Support for the revolution.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 232 bytes
Desc: not available
URL: <http://lists.humbug.org.au/pipermail/general/attachments/20001110/4bea4136/attachment.sig>