From mboxrd@z Thu Jan 1 00:00:00 1970 From: Loic Dachary Subject: Re: [Ceph-maintainers] Deprecating ext4 support Date: Tue, 12 Apr 2016 08:39:38 +0200 Message-ID: <570C982A.5090600@dachary.org> References: Mime-Version: 1.0 Content-Type: text/plain; charset=windows-1252 Content-Transfer-Encoding: QUOTED-PRINTABLE Return-path: Received: from smtp22.services.sfr.fr ([93.17.128.11]:15273 "EHLO smtp22.services.sfr.fr" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752087AbcDLGjq (ORCPT ); Tue, 12 Apr 2016 02:39:46 -0400 In-Reply-To: Sender: ceph-devel-owner@vger.kernel.org List-ID: To: Sage Weil , ceph-devel@vger.kernel.org, ceph-users@ceph.com, ceph-maintainers@ceph.com, ceph-announce@ceph.com Hi Sage, I suspect most people nowadays run tests and develop on ext4. Not suppo= rting ext4 in the future means we'll need to find a convenient way for = developers to run tests against the supported file systems. My 2cts :-) On 11/04/2016 23:39, Sage Weil wrote: > Hi, >=20 > ext4 has never been recommended, but we did test it. After Jewel is = out,=20 > we would like explicitly recommend *against* ext4 and stop testing it= =2E >=20 > Why: >=20 > Recently we discovered an issue with the long object name handling th= at is=20 > not fixable without rewriting a significant chunk of FileStores filen= ame=20 > handling. (There is a limit in the amount of xattr data ext4 can sto= re in=20 > the inode, which causes problems in LFNIndex.) >=20 > We *could* invest a ton of time rewriting this to fix, but it only af= fects=20 > ext4, which we never recommended, and we plan to deprecate FileStore = once=20 > BlueStore is stable anyway, so it seems like a waste of time that wou= ld be=20 > better spent elsewhere. >=20 > Also, by dropping ext4 test coverage in ceph-qa-suite, we can=20 > significantly improve time/coverage for FileStore on XFS and on BlueS= tore. >=20 > The long file name handling is problematic anytime someone is storing= =20 > rados objects with long names. The primary user that does this is RG= W,=20 > which means any RGW cluster using ext4 should recreate their OSDs to = use=20 > XFS. Other librados users could be affected too, though, like users=20 > with very long rbd image names (e.g., > 100 characters), or custom=20 > librados users. >=20 > How: >=20 > To make this change as visible as possible, the plan is to make ceph-= osd=20 > refuse to start if the backend is unable to support the configured ma= x=20 > object name (osd_max_object_name_len). The OSD will complain that ex= t4=20 > cannot store such an object and refuse to start. A user who is only = using=20 > RBD might decide they don't need long file names to work and can adju= st=20 > the osd_max_object_name_len setting to something small (say, 64) and = run=20 > successfully. They would be taking a risk, though, because we would = like=20 > to stop testing on ext4. >=20 > Is this reasonable? If there significant ext4 users that are unwilli= ng to=20 > recreate their OSDs, now would be the time to speak up. >=20 > Thanks! > sage >=20 > _______________________________________________ > Ceph-maintainers mailing list > Ceph-maintainers@lists.ceph.com > http://lists.ceph.com/listinfo.cgi/ceph-maintainers-ceph.com >=20 --=20 Lo=EFc Dachary, Artisan Logiciel Libre -- To unsubscribe from this list: send the line "unsubscribe ceph-devel" i= n the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html