xfs
[Top] [All Lists]

Re: XFS and EVMS - Limited Success

To: "Karl M. Hegbloom" <karlheg@xxxxxxxxxxxxxx>
Subject: Re: XFS and EVMS - Limited Success
From: Eric Sandeen <sandeen@xxxxxxx>
Date: 23 Apr 2002 14:12:43 -0500
Cc: linux-xfs@xxxxxxxxxxx
In-reply-to: <1019588340.27446.23.camel@xxxxxxxxxxxxxxxxxxxxxxxxxxxx>
References: <20510000.1019570836@camillo> <Pine.LNX.4.44.0204230929590.4155-100000@stout .americas.sgi.com> <20020423165414.202b89e2.pegasus@xxxxxxxxxxxx> <3CC57AC5.4010308@xxxxxxx> <1019588340.27446.23.camel@xxxxxxxxxxxxxxxxxxxxxxxxxxxx>
Sender: owner-linux-xfs@xxxxxxxxxxx
On Tue, 2002-04-23 at 13:59, Karl M. Hegbloom wrote:
> On Tue, 2002-04-23 at 08:16, Stephen Lord wrote:
> 
> > We had some ioctl problem reports from mkfs which should now be fixed in
> > EVMS 1.0. Someone else reported it was working. I have not tried myself.
> 
> I am using EVMS, and see:
> 
> root@juniper:~
> # mkfs -t xfs /dev/evms/test 
> mkfs.xfs: warning - cannot set blocksize on block device /dev/evms/test:
> Invalid argument

That's not really a problem.  It just means that the BLKSETSZ ioctl is
not supported by evms; for now, anyway, xfs can live without it.

> It would be a lot easier for me
> to see what version of XFS the kernel has built into it if you would
> cause the "dmesg" (printk) output at startup to give a version number.

Well, there are 2 flavors of XFS - releases, and cvs snapshots (this
includes any patch you find on oss.sgi.com that's NOT under a
"Release-X" directory.)  It would have been a good idea to hard code a
release version into the releases, yes.  Sorry about that.

> I've given up on finding what version it is after 5 minutes of code
> grepping, sorry.  Ah... dpkg --status kernel-patch-xfs tells me that
> it's version "1.0.2+20020304-2".  So do I need a newer one to make XFS
> work right with EVMS?

Sounds like you got it from debian, then - and I guess they got it from
release 1.0.2, and maybe tweaked it a bit?

I would suggest getting patches + userspace from under the Release-1.1,
if you're looking for a set of things that are known to work well
together.  However, there isn't anything inherently wrong with the
combination you have now.

-Eric

-- 
Eric Sandeen      XFS for Linux     http://oss.sgi.com/projects/xfs
sandeen@xxxxxxx   SGI, Inc.


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