From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from len.romanrm.net ([176.31.121.172]:35262 "EHLO len.romanrm.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1760790Ab3DBREZ (ORCPT ); Tue, 2 Apr 2013 13:04:25 -0400 Date: Tue, 2 Apr 2013 23:04:21 +0600 From: Roman Mamedov To: =?UTF-8?B?U3fDom1p?= Petaramesh Cc: "BTRFS, Linux" Subject: Re: Moving... Message-ID: <20130402230421.016afe58@natsu> In-Reply-To: <515A95C7.5090607@petaramesh.org> References: <515A95C7.5090607@petaramesh.org> Mime-Version: 1.0 Content-Type: multipart/signed; micalg=PGP-SHA1; boundary="Sig_/KifHl6Z_tfnJevexKDG7lnr"; protocol="application/pgp-signature" Sender: linux-btrfs-owner@vger.kernel.org List-ID: --Sig_/KifHl6Z_tfnJevexKDG7lnr Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: quoted-printable On Tue, 02 Apr 2013 10:24:39 +0200 Sw=C3=A2mi Petaramesh wrote: > Goodbye BTRFS, hello ZFS :-) >=20 > I'm finally making the move, I couldn't stand the terrible BTRFS > performance anymore, and spending 2 long minutes waiting for the HD LED > to come off everytime I clicked anywhere. >=20 > Did what I could, got the latest kernels, defragged, removed BTRFS > snapshots, to no avail... >=20 > So I'm in process of moving the 3 main laptops in here from BTRFS to ZFS.= .. Whatever works for you -- but at this point I trust my data to BTRFS more, than I would trust ZFS. (of course having current backups etc). The following may not sound like a compliment to BTRFS; but on my filesyste= ms with heavy snapshotting, recently I am trialling a practice to run major ap= ps like a mail client and a browser via "eatmydata", which overrides and NOOPs fsync. Since in any case I have "sync; btrfs sub snap...." on those machines set to every 30 minutes in crontab, the potential of data loss due to "eatmydata" is never too great. I noticed the general experience is a lot snappier this way. You could try = at least to verify whether or not your HD LED issues (:D) are indeed caused by slow fsyncs, or by something else. I should note that I currently use a 3.7 series kernel, and haven't upgraded to 3.8 and 3.9 yet, AFAIK those had some fsync improvements which might ren= der this trick useless. But this still doesn't change the fact that apps in general seem to over-use fsync. --=20 With respect, Roman --Sig_/KifHl6Z_tfnJevexKDG7lnr Content-Type: application/pgp-signature; name=signature.asc Content-Disposition: attachment; filename=signature.asc -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.12 (GNU/Linux) iEYEARECAAYFAlFbD5UACgkQTLKSvz+PZwjFSACfbP1ZiWzsOoB8s2Fkm7vj7ffw XQEAn3cLoUMLhUF6UV6WZTe2Nha5CYbT =959x -----END PGP SIGNATURE----- --Sig_/KifHl6Z_tfnJevexKDG7lnr--