On to. 25. okt. 2012 kl. 11.00 +0200, Brian Candler wrote: > On Thu, Oct 25, 2012 at 10:23:33AM +0200, Runar Ingebrigtsen wrote: >> Does that mean the bauer-power article [1] about how healing fails is >> inaccurate? > > AFAICS that page doesn't say anything about how gluster healing works. He > says he had data corruption, and I do not doubt that. > > However the page doesn't specify the version of gluster FS used, the > versions of Ubuntu and ZFS used, etc. So I can't conclude what caused the > corruption. > > Running Linux on ZFS and Gluster on ZFS in production would take someone > quite brave I think. Plus he seems to be using IET (iSCSI). That is, I > guess he's running a virtual SAN by serving disk image over iSCSI, where > those disk image files are stored on a replicated gluster filesystem on top > of ZFS? Very brave. Awesome answer, Brian. I already knew that ZFS is a risky choice, and that iSCSI is not as friendly towards failover as NFS, CIFS or GlusterFS. I is very educating to get your view on this. -- Best Regards Runar Ingebrigtsen
