Hi Mathew,
If you use FUSE mount, the client is connecting to all servers in the volume and thus it works like MPIO.
For example - 3 replica volume has the same data in all 3 gluster servers and thus the cluent has 3 'paths' to the volume.
If one fails, a timeout will occur and thr client will stop trying to update that server's bricks.
Of course, you can have multiple IPs defined per gluster server and thus the client will try to reach the server's bricks by any IP in the configuration.
For example, Brick A has '192.168.1.2' & '192.168.2.2'. There will be no constant connection over both IPs, but as far as I know the client will try the second IP.
Best Regards,
Strahil Nikolov
On Aug 23, 2019 20:35, Matthew Evans <runmatt@xxxxxxxx> wrote:
Does Gluster support any sort of MPIO-like functionality for a single client? Maybe when using a Distributed Stripe? Or is a single client always limited to the throughput of a single node for each session?
_______________________________________________ Gluster-users mailing list Gluster-users@xxxxxxxxxxx https://lists.gluster.org/mailman/listinfo/gluster-users