xfs
[Top] [All Lists]

Re: logdev on IDE

To: linux-xfs <linux-xfs@xxxxxxxxxxx>
Subject: Re: logdev on IDE
From: Florin Andrei <florin@xxxxxxx>
Date: 24 Jul 2002 12:40:59 -0700
In-reply-to: <1027535678.14036.28.camel@xxxxxxxxxxxxxxxxxxxxxx>
References: <1027535026.1648.5.camel@xxxxxxxxxxxxxxxxxxx> <1027535678.14036.28.camel@xxxxxxxxxxxxxxxxxxxxxx>
Sender: owner-linux-xfs@xxxxxxxxxxx
Indeed, that would be a problem.
Anyway, some generous souls just gave me an external SCSI drive, so the
case for the IDE is closed. ;-)

But, for future refference, is there a way to tell if an IDE drive is
doing bad things? I mean, other than pressing Reset repeatedly...
Or perhaps someone put up a list of good/bad IDE drives, in regard to
the caching problem...
I guess this should be interesting for anyone using a journalised FS.

On Wed, 2002-07-24 at 11:34, Eric Sandeen wrote:
> Hm, if that IDE speed is coming from the write-cache on the drive, you
> may as well just put your log on a ramdisk.  :)  Of course, log
> integrity might suffer on a crash.
> 
> Seriously, unless you -really- trust that write cache, be careful!
> 
> -Eric
> 
> On Wed, 2002-07-24 at 13:23, Florin Andrei wrote:
> > >From a pure performance perspective, i wonder if anyone tried to put the
> > logdev on IDE while using SCSI for the actual XFS disks/arrays.
> > 
> > Theoretically, the new IDE drives should be quite fast, but anyway, if
> > anyone tried that, i'd like to hear whatever good/bad things happened,
> > performance-wise.
> > 
> > -- 
> > Florin Andrei
> > 
> > "Still wondering why they called it 'United Linux'. Firstly it sounds
> > like a football team and secondly wouldn't 'Turbo Susiva' have sounded
> > much better?" - Alan Cox
> -- 
> Eric Sandeen      XFS for Linux     http://oss.sgi.com/projects/xfs
> sandeen@xxxxxxx   SGI, Inc.         651-683-3102
> 
-- 
Florin Andrei

"Still wondering why they called it 'United Linux'. Firstly it sounds
like a football team and secondly wouldn't 'Turbo Susiva' have sounded
much better?" - Alan Cox


<Prev in Thread] Current Thread [Next in Thread>