What if you did the 'peer probe' by name but each node's local name resolves to 127.0.0.1 via /etc/hosts? Jeff White - Linux/Unix Systems Engineer University of Pittsburgh - CSSD On 03/21/2012 02:13 PM, Brian Cipriano wrote: > Hey all - > > Anyone running Gluster on AWS EC2 instances out there? > > I'm having some issues using Gluster on EC2 instances. I saw an old > thread titled "ec2 peers" but didn't see a solution posted as part of > that thread. > > Here's my problem - > > If I stop/start a single EC2 instance, usually its internal DNS name and > IP address change. This is a problem for my Gluster volume, which is > using the instances' internal DNS names - all of a sudden a bunch of > bricks in the volume aren't there - though they do still exist, they're > EBS volumes and they're attached to the same instance. It's just that > the instance has a different DNS name/ IP. > > Any EC2+Gluster users out there who have found a solution to this? What > do you do when the hostname of one of your servers changes? > > I can't use replace-brick because the old instance isn't available at > its old hostname. > > I've tried assigning an Elastic IP and using that instead of internal > DNS, but my first server is unable to see itself as a peer when > identified by IP. I think Elastic IP is only accessible from outside the > AWS networks. > > I would really appreciate any thoughts on this. Thanks! > > - brian > _______________________________________________ > Gluster-users mailing list > Gluster-users at gluster.org > http://gluster.org/cgi-bin/mailman/listinfo/gluster-users
