Subject: Re: OT: Was: Re: How to capture all file system writes
To: Jason Thorpe <thorpej@wasabisystems.com>
From: Greywolf <greywolf@starwolf.com>
List: tech-kern
Date: 10/22/2003 10:33:50
Thus spake Jason Thorpe ("JT> ") sometime Today...

JT> On Wednesday, October 22, 2003, at 08:35  AM, Greywolf wrote:
JT>
JT> > Hm.  Judging by that, perhaps PBL is NOT the wisest move (for the
JT> > uninitiated:  jfs and xfs are more stable than either of ext3 or
JT> > reiserfs).
JT>
JT> Physical block logging vs. logical operation logging don't really have
JT> much to do with the stability of the file system.  That is, unless you
JT> count code complexity; physical block logging is much less complex when
JT> it comes time to replay the journal (i.e. you don't have to reconstruct
JT> what the actual physical block operations were).  In any case, there
JT> are plenty of other reasons why a particular file system implementation
JT> could suck, none of which really have anything to do with the
JT> journaling strategy chosen.

Sorry, then, for the generalisation, but as ext3 and reiserfs have had
huge reliability problems and both depended on PBL, with that as the data
point presented, it leads very easily to my incorrect conclusion. :-)

				--*greywolf;
--
NetBSD: the free unix for the rest of us.