[TriLUG] ext3 surprise

[Jeremy Portzer]

On Thu, 2004-03-25 at 13:12, Brian Daniels wrote:

> My understanding of ext3 says this is a "can't happen" situation unless the 
> underlying hardware fails.  But the underlying hardware is a SMART-capable 
> IDE hd, and it insists all is well.
> 
> Isn't the point of a journaling file system to guarantee that fsck's are 
> unnecessary?  Any ext3 guru's on the list?
> 


This is a common misconception about ext3, at least for the default
mode.  In most cases, the journal is set up in "ordered data" mode.

See this whitepaper from Red Hat for a discussion:
http://www.redhat.com/support/wpapers/redhat/ext3/

It says that "data=ordered (the default mode), guarantees that the data
is consistent with the file system; recently-written files will never
show up with garbage contents after a crash."

However, this doesn't mean that you're guaranteed that all writes are
successful.  You would need to change the option to "data=journal"
(which is slower) in order to get the protection you are expecting.

In other words, ext3 gave you the advantage that you knew for sure that
/sbin/install-info was toast.  You were guaranteed of knowing which
files were inconsistent and which were fine.  Without any journaling,
there might have been files that passed an fsck fine, but actually had
garbage data in them.

Hope this helps,

Jeremy


/---------------------------------------------------------------------\
| Jeremy Portzer        jeremyp at pobox.com      trilug.org/~jeremy     |
| GPG Fingerprint: 712D 77C7 AB2D 2130 989F  E135 6F9F F7BC CC1A 7B92 |
\---------------------------------------------------------------------/
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 189 bytes
Desc: This is a digitally signed message part
URL: <http://www.trilug.org/pipermail/trilug/attachments/20040325/78a03e7b/attachment.pgp>