Skip site navigation (1)Skip section navigation (2)
Date:      Wed, 28 Jan 1998 19:52:01 -0600
From:      "Jeffrey J. Mountin" <mountin.man@mixcom.com>
To:        James Wyatt <jwyatt@rwsystr.RWSystems.net>
Cc:        freebsd-isp@FreeBSD.ORG
Subject:   Re: Sendmail - low on space
Message-ID:  <3.0.3.32.19980128195201.0071d610@198.137.186.100>
In-Reply-To: <Pine.LNX.3.91.980127224514.13089B-100000@rwsystr.RWSystems .net>
References:  <Pine.BSF.3.96.980127165831.21902I-100000@guardian.fortress.org>

next in thread | previous in thread | raw e-mail | index | archive | help
At 11:14 PM 1/27/98 -0600, James Wyatt wrote:
>On Tue, 27 Jan 1998, Andrew Webster wrote:
>> On Wed, 28 Jan 1998, Norman Hoy wrote:
>> Of course what would be REALLY nice would be a virtual disk driver like
>> AIX's JFS where you can just keep adding disks when you run out of space
>> on the logical volume!
>> 
>> Oh well, we can at least dream...
>
>	[ much else deleted, cause I use a tty, not a window... 8{) ]

<sympathy>
</sympathy> ;)

>I *love* this feature more than I can tell. Your OS installs with the
>needed partition sizes, leaving the rest of the disk 'uncommitted'. The
>install process expands them by the amount they need for each package. 
>When you get paged on a low disk space alarm, you can just sprinkle some
>more space on whichever filesystem is low. It adjusts the mirroring and
>striping drives' partitions too! If you do not realize what you are doing,
>it can easily lead to hopelessly fregmented drives, so don't grow by small
>amounts. 
>
>You can grow them while users are going at things full blast, but the new
>mirror can take a few minutes to catch-up. You can only grow, not shrink.
>It is *very* easy to get used to this power tool!  I use several of
>different unicies, but AIX is the only one I've used with this. Doesn't
>someone else like HPUX or OSF/1 have something like this? 

Novell does this.  Need space, drop in a drive.

>Another nice feature is SSA (Serial SCSI Array?) drives allow multiple
>machines to access the same sets of drives to improve process-takeover in
>fail-over systems. If a box goes, it's sibling grabs it's drives, ARPs
>for it, and starts whatever processes it needs. Users see a few seconds of
>pause and then it just works again - while you fix the hardware in peace!
>We have lost systems and not realized it for hours.
>
>A take-over scheme like this could work on FreeBSD if you could have two 
>machines share a SCSI-UW bus with some drives. Target-mode SCSI sould 
>allow the machines to exchange info on who had what drives. Any takers? (^_^)
>
>Of course, when we tried to use DCE to make some nice client-server 
>systems, we found the software didn't scale as well as the hardware... Oh 
>well, at least DCE had decent RPCs and DNS still works!

This would certainly solve the backup server problem.  What about if the SSA fails and 2 or more servers die?

Novell can use a fiber link between 2 servers to mirror them.  So no matter what fails on the primary, there isn't a hitch in stride.

Guess we have some things on our FBSD wish list then. :O


Jeff Mountin - Unix Systems TCP/IP networking
mountin.man@mixcom.com




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?3.0.3.32.19980128195201.0071d610>