xfs
[Top] [All Lists]

Re: xfs_repair on large fs: out of memory

To: Linux XFS <linux-xfs@xxxxxxxxxxx>
Subject: Re: xfs_repair on large fs: out of memory
From: pg_xfs@xxxxxxxxxxxxxxxxxx (Peter Grandi)
Date: Thu, 18 May 2006 00:54:47 +0100
In-reply-to: <20060517222353.GA32668@xxxxxxxxxxxxxxxxxxxxx>
References: <20060517193606.GO32378@xxxxxxxxxx> <20060517222353.GA32668@xxxxxxxxxxxxxxxxxxxxx>
Sender: linux-xfs-bounce@xxxxxxxxxxx
>>> On Wed, 17 May 2006 15:23:53 -0700, Chris Wedgwood
>>> <cw@xxxxxxxx> said:

cw> On Wed, May 17, 2006 at 03:36:06PM -0400, Ed L. Cashin
cw> wrote:

>> Hi.  In trying to run xfs_repair on a 13 TB filesystem,

I think I am not alone on this list in being very curious how
long does the check+repair take.

ed> the program either runs out of memory ... [ ... ] This
ed> machine has a gigabyte of memory.  It's running an x86_64
ed> 2.6.16.13 kernel.

Also, how do you backup that filesystem? How long does it take?

cw> I forget the ratio/math for working this out (Steve Lord
cw> posted a summary of this some time ago) but 1GB isn't nearly
cw> enough to run xfs_repair on a moderately sized filesystem.

The summary is this one probably, from a little while ago:

  http://OSS.SGI.com/archives/linux-xfs/2005-08/msg00045.html

It suggests that a 13TB filesystem might need from 10 to 20
gigabytes of RAM, after all only less than 0.2% of its size.

As a point of order I still think that perhaps one (I could
volunteer) should do a mailing list FAQ with a URL to which put
at the bottom of every message. But then most people who want
help on this list want their answer without wasting their time
searching the archive, so probably they would not pointlessly
read any FAQ either...


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