Hello, What happens when the first node fails (and is not coming back) when using the storage platform? We're using mirroring so our clients continue to have access and we have a backup of the configuration so we're not concerned about data loss. We are not seeing how to get the "first server / management server" back online though by either making another server the management server or installing a new server as the "first server" and then restoring the configuration. Any insight is appreciated.
First server is serving web UI and most of server management is done through it. If first server is down, running volumes keep continue to serve and depending on the volume type, some of the files served by first server may be missing on mount point. Regards, Bala Bo Molocznik wrote:> Hello, > > What happens when the first node fails (and is not coming back) when using the storage platform? > > We're using mirroring so our clients continue to have access and we have a backup of the configuration so we're not concerned about data loss. We are not seeing how to get the "first server / management server" back online though by either making another server the management server or installing a new server as the "first server" and then restoring the configuration. > > Any insight is appreciated. > _______________________________________________ > Gluster-users mailing list > Gluster-users at gluster.org > http://gluster.org/cgi-bin/mailman/listinfo/gluster-users
Hi,> > So right now if the first server (management server) goes away there is no way to get the management console back up and running? >Yes.> Also, what does that mean for node failure? If a node fails and is replaced with a different node what happens to the mirrored files? Is there are way to specify when adding a node that it is to replace another failed one? >Node failure is that the machine is not able to connect to storage network and serve data. Mirror type volume will keep serving full data from another server. For example, a mirror type volume involves six servers means three mirrored pairs and data is distributed across these pairs. If a server is down, data is served by its mirror server. If pair of server is down, data stored in those pair won't be available, but we get other data. Upcoming 3.1 version will have dynamic volume, in which we can specify a replacement node. In 3.0, there is no such option. Thanks, Regards, Bala