Gandalf Corvotempesta [2016-05-27 13:18 +0200] :> and the recommended way is to shut everything down as the rolling > update states: "feel adventurous"I just read that... Anyway, I hope that it's possible. I'm with you since I don't see a point in a clustered file system if you can't take a node down by plan. Not only to upgrade GlusterFS, but the entire system. Niklaas -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 819 bytes Desc: not available URL: <http://www.gluster.org/pipermail/gluster-users/attachments/20160527/08084526/attachment.sig>
On 27/05/2016 9:22 PM, Niklaas Baudet von Gersdorff wrote:> I just read that... Anyway, I hope that it's possible. I'm with you > since I don't see a point in a clustered file system if you can't take > a node down by plan. Not only to upgrade GlusterFS, but the entire > system.I've done rolling upgrades of point releases without issues (replica 3 volume), so long as you're slow and methodical I don't think its an issue. Major upgrades of any system tend to involve pain - I know the ceph upgrade from Hammer to Jewel (9.x - 10.x) was a non trivial procedure that involved taking the cluster down. -- Lindsay Mathieson -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://www.gluster.org/pipermail/gluster-users/attachments/20160527/4e85c639/attachment.html>
Il 27/05/2016 13:22, Niklaas Baudet von Gersdorff ha scritto:> I just read that... Anyway, I hope that it's possible. I'm with you > since I don't see a point in a clustered file system if you can't take > a node down by plan. Not only to upgrade GlusterFS, but the entire > system. NiklaasExactly what I mean.