2GB of Waste? How can it be? -- some JFS are worse than non-JFS

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



"IT3 Stuart Blake Tener, USNR-R" wrote:
> In terms of enterprise reliability, I understand, however, having an
> "office recommended journaling filesystem",

Some journaling filesystems can be _worse_ than non-journaled.  If
the recovery mechanism of the JFS is to "aggressively" go to the
journal, journal mis-reads can _toast_ a filesystem.  I'll take a
full fsck that _always_ works to a jounaling filesystem that
aggressively goes to the journal, or has unproven recovery tools.

I have experienced this on NTFS many times.  To the point where I
_force_ a CHKDSK anyway, on any improper shutdown (effectively
negating the benefit of a JFS).  NTFS is based on IBM's OS/2 HPFS,
so this isn't just a Microsoft design mentality (as HPFS follows
some of the same logic).  3rd party sources (i.e. could be heresy)
says IBM's JFS follows this "aggressive" journaling attitude -- so
combined with the lack of various kernel/utility support in JFS for
Linux right now, I'm not using it either.

I'll admit, ReiserFS doesn't seem to be overly "aggressive" in
journal reads.  But there is something to be said about the Namesys'
"focus" on the all-important "recovery" tools (probably because they
don't get updated immediately when structures do).  Case in point,
I've had people with just "unclean" ReiserFS partitions _toast_ them
when running a full fsck.  On the other hand, I've had physical disk
errors before, but fsck.ext2/3 comes through everytime with a
recovered fs.

When some Windows bigot gets on me about Ext2 not being a
"journaling filesystem," there is no arguing with the proven
reliability of an Ext2 fsck.

> and allowing people to install ReiserFS as the root filesystem
> notwithstanding that recommendation, is two different things.
> I do hope that in the next release of RH Linux, the ability to install
> into ReiserFS root filesystems is inclusive to the distribution. Many
> vendors allow customers to install software in unsupported configurations.
> Allowing customers to install that way, does not mean that you have to
> support it.

Unfortunately, it seems RedHat's customers do not see it that way.

Mandrake supports Ext3, JFS, ReiserFS and XFS "out-of-the-box."  It
has a nice installer and configuration suite that is sweet.  And if
you have non-commodity hardware or configurations, it breaks all
over the freak'n place.  I know.  After reporting things time and
time again, they never get fixed.  I fixed a few things, but gave up
when it came to submission.

I also see regular kernel panics with Mandrake kernels.

RedHat may not have everything in its installer, tools and support,
but what it does it does well with few bugs (that _never_ go
"unfixed" more than a month).  This is what RedHat does.

> I am not operating in an enterprise environment (my laptop),
> thus my concerns are different. Therefore, I suppose I do not see why
> the option to do that (perhaps albeit with a warning message) is not
> there.

Because RedHat doesn't want to support it!  Period.  And there are
endless reasons why.  The filesystem is _everything_, and RedHat
would have to ask, "what filesystem are you using?" everytime a
support call comes in.

And trust me, there are a _lot_ of issues that are
ReiserFS-specific.

> I was not aware that devfs could be mounted from fstab! I will
> look into that.

It can also be removed as a boot-time option.

-- Bryan

-- 
Bryan J. Smith, Engineer        mailto:b.j.smith@ieee.org   
AbsoluteValue Systems, Inc.     http://www.linux-wlan.org
SmithConcepts, Inc.          http://www.SmithConcepts.com
---------------------------------------------------------
1999 IRS Data:  The top 1% of income earners pay over 36%
of the taxes, but have less than 20% of the total income.





[Index of Archives]         [Linux RAID]     [Kernel Development]     [Red Hat Install]     [Video 4 Linux]     [Postgresql]     [Fedora]     [Gimp]     [Yosemite News]

  Powered by Linux