On Wed, Dec 9, 2009 at 4:14 AM, Anand Babu Periasamy
<address@hidden> wrote:
Brian Hirt wrote:
I have a question about upgrading, but it's not really answered in http://gluster.com/community/documentation/index.php/GlusterFS_General_FAQ#Will_GlusterFS_be_upward_compatible.3F
I have a 4 node system, with AFR/Distribute. Can I shut down the servers one at a time, upgrade from 2.0.6 to 2.0.8 and restart the node. After doing this process on all four servers will everything be running okay?
Also, a bit longer term, migrating from 2.x to the soon to be released 3.x branch, will this method work as well? Once all servers have been upgraded, will they automatically renegotiate to the most recent protocol or does the entire cluster need to be shut down and restarted.
You can do that between 2.0.6 to 2.0.8. All 2.0.x will be compatible to each other. Though I recommend upgrading all servers at once. To upgrade to 3, you need to upgrade all your servers and clients. Backend data layout is same. After upgrading the installation, you restart GlusterFS servers and remount the clients.
Dynamic volume management (to add, remove, suspend) servers without requiring re-mount is going to be the key focus of 3.1 release.