On Sun, Jul 31, 2016 at 6:53 PM, Goncalo Borges <goncalo.borges@xxxxxxxxxxxxx> wrote: > Dear cephers. > > I would like to request some clarification on migrating from legacy to > optimal (jewel) tunables. > > We have recently migrated from infernalis to Jewel. However, we are still > using legacy tunables. > > All our ceph infrastructure (mons. odss and mdss) are running 10.2.2 in > Centos 7.2.1511 and kernel 3.10.0-327.22.2.el7.x86_64. However, we are > providing 10.2.2 clients (basically ceph-fuse) in SL6 and kernel > 2.6.32-642.el6.x86_64. We do this because of the nature of the research > performed by our scientists where there is (still) the requirement to > provide cephfs under sl6. To be able to accomplish that, we have > successfully compiled ceph in SL6 with gcc 4.8.4, python 2.7.6, boost > 1.53.0-25 and fuse 2.9.7 which we enable via environment modules. > > The tunables documentation [1] states the following: > > WHICH CLIENT VERSIONS SUPPORT CRUSH_TUNABLES5 > - v10.0.2 (jewel) or later > - Linux kernel version v4.5 or later (for the file system and RBD kernel > clients) > > Can you please confirm that the two previous requirements are ORed and not > ANDed? I.e, can i migrate to Jewel tunables using ceph-fuse 10.2.2 in a SL6 > box (running a 2.6 kernel) compiled as I explained above? Right, the code you're using as the Ceph client needs to be that version. So in your case the clients are userspace, and you need Jewel code in userspace. If you have any kernel mounts, they'd need to be on kernel 4.5, but I don't think that's in your systems. -Greg _______________________________________________ ceph-users mailing list ceph-users@xxxxxxxxxxxxxx http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
 |