[Top] [All Lists]

Re: LBA to File?

To: Stephen Lord <lord@xxxxxxx>
Subject: Re: LBA to File?
From: Andy Skunza <askunza@xxxxxxxxxxxxxxx>
Date: Fri, 07 Mar 2003 09:30:21 -0500
Cc: linux-xfs@xxxxxxxxxxx
In-reply-to: <1046997413.1373.14.camel@localhost.localdomain>
References: <3E67BFA7.4080601@echostar.com> <20030306221631.GA6770@f00f.org> <3E67CB02.2030600@echostar.com> <1046990328.439.273.camel@stout.americas.sgi.com> <3E67D0CE.3080902@echostar.com> <1046997413.1373.14.camel@localhost.localdomain>
Sender: linux-xfs-bounce@xxxxxxxxxxx
User-agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.3a) Gecko/20021212
Stephen Lord wrote:

On Thu, 2003-03-06 at 16:50, Buzbee, James wrote:

The scary thing about what we are seeing is that the "grown defect list" on the disk is showing a lot of sectors in these two areas. It's almost as if we are wearing out the sectors in these areas by writing them so often. Strange but true ;-(


Ugh, yes these will be continually written to, the log in particular, I like Eric's idea, but the scope of that is fairly drastic and it will only postpone the inevitable - you can probably do the math to work out how long the drive will run before it runs out of bad blocks to remap to. All running with alternating logs would do is delay the drive becoming scrap metal. The reliability of these devices may not be up to the load you are placing on them. Not all ide drives are this bad, I have been running xfs on the same ide drive for about 3 years now - but then I am not running a PVR. I suspect also you may have the write cache turned off to avoid errors after a crash.

Possibly the way your app runs is making things worse. There is a script
in  the cmd/misc directory called xfsstats.pl you can watch log activity
with this. In particular xs_log_writes and xs_log_blocks. If log_blocks
is not a reasonable multiple of log_writes then you are doing a lot of
very small writes.


What is a "reasonable multiple" of log_writes to log_blocks? Would 22 be unreasonable? And, how to use xfs_stats.pl per xfs partion?


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