From mboxrd@z Thu Jan 1 00:00:00 1970 From: Bernd Schubert Subject: Re: reiserfs logging patches udpated Date: Thu, 25 Mar 2004 01:47:42 +0100 Sender: bernd-schubert@web.de Message-ID: <200403250147.54246.bernd-schubert@web.de> References: <1080149399.14737.77.camel@watt.suse.com> <20040324143516.3162cbfe.akpm@osdl.org> <1080168379.14743.124.camel@watt.suse.com> Mime-Version: 1.0 Content-Transfer-Encoding: quoted-printable Return-path: list-help: list-unsubscribe: list-post: Errors-To: flx@namesys.com In-Reply-To: <1080168379.14743.124.camel@watt.suse.com> Content-Disposition: inline List-Id: Content-Type: Text/Plain; charset="us-ascii" To: reiserfs-list@namesys.com Cc: Chris Mason , Andrew Morton =2D----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 > It does, but the code has been in testing -suse and on the reiserfs > list. This dooesn't mean data=3Dordered is perfect, but it's not quite > day one either. I can switch the default back, but I'd rather have a > trial by fire ;-) Oh please not, just have a look at german newsgroups, there are real flamew= ars=20 about the stability of reiserfs. Everytime someone has a disk problem and a= =20 reiserfs partion is affected by it, then there are more than a dozen answer= s=20 that its the fault of reiserfs. PLEASE, don't prove them right. The last example is less than 2 weeks old (see subject "Crash --> ReiserFS= =20 screwed up"), since Tim told on de.comp.os.unix.linux.misc that badblocks=20 reported that his harddisk is ok, I told him to ask here. By his post here,= =20 one could see, that his harddisk or ide connection was not so fine as he=20 previously stated, so we continued to discuss in on=20 de.comp.os.unix.linux.misc and finally could solve the problem by setting a= =20 lower udma speed. However, besides of the proper discussion about the ide problem, there was = a=20 'sub' thread with about 50 messages about the instability of reiserfs!!! IMHO, almost all of the reported reiserfs problems are due to hardware erro= rs,=20 but some are also due to patches which are enabled by default by=20 distributions like Suse or RedHat.=20 Just another personal example: Before my first post on this list, I didn't= =20 know much about reiserfs, only that it handles system crashes better then=20 ext2 and that I can have file >2GB on 2.4.x. So I went ahead and replaced=20 every ext2 partition by reiserfs on the next Suse-7.0 installation. Then th= e=20 problems arrived, since I also exported everything via nfs and since the Su= se=20 installer had choosen the 3.5 format without warning me about it, I run int= o=20 serious trouble and reverted everything that was exported by nfs back to ex= t2=20 for the next 2 years. Even until Suse-7.3, the 3.5 format was the default from the Suse patched=20 mkreiserfs, without warning the user about it. I think -mm kernel are experimental and can include experimental stuff, but= =20 later in the main line everything that is experimental for the 3.6 format=20 should be disabled by default. I wouldn't mind if it becomed included, but= =20 marked as experimental, so that one can easily wouldn't include it in a=20 server kernel. Sorry for my long (slightly off topic) message, Bernd =2D----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.4 (GNU/Linux) iD8DBQFAYiw0C8BUnAF+ydYRApYMAJ9KW83lYY9TJ7oIwiLmHwlub/cf3ACeIaSL dPX8EuqpIH2kvMh+JPTrRGA=3D =3Dm5ez =2D----END PGP SIGNATURE-----