Tomasz Chmielewski
2012-May-24 15:05 UTC
[Gluster-users] migrating from distribute to replicate - possible?
I have the following gluster setup (distribute): Type: Distribute Status: Started Number of Bricks: 10 Transport-type: tcp Bricks: Brick1: ca1-int:/data/glusterfs Brick2: ca2-int:/data/glusterfs Brick3: ca3-int:/data/glusterfs Brick4: ca4-int:/data/glusterfs Brick5: ca5-int:/data/glusterfs Brick6: ca6-int:/data/glusterfs Brick7: ca7-int:/data/glusterfs Brick8: ca8-int:/data/glusterfs Brick9: ca9-int:/data/glusterfs Brick10: ca10-int:/data/glusterfs What would be the easiest way to migrate it to "Replicate" type? What if I do the following: - remove all bricks/volume, - create it from scratch as "replicate" type, and point to the same paths, - balance / ls -lR. Am I right to think that it would just copy the missing files from one server to another, and after some time it would be migrated to "replicate", with no data loss, and almost no downtime (the downtime would be just what it takes to remove / readd volume/bricks)? -- Tomasz Chmielewski http://www.ptraveler.com