[Gluster-devel] How do we identify peers?

Justin Clift jclift at redhat.com
Thu Jun 27 15:11:49 UTC 2013


On 27/06/2013, at 10:13 AM, Kaushal M wrote:
<snip>
> I'm right now leaning towards a hybrid approach of using a translation
> layer and using connection groups. We could use any of the addresses
> associated with a peer, with the commands. Internally, we will be
> maintaining a list of addresses associated with peers which will make
> the internal translation easier. We will use UUIDs exclusively
> internally, this includes the volfiles and other generated config
> files.
> I'll write a detailed description of this and send it here for comments soon.


Cool. :)

As a thought (maybe for a later implementation so the bar isn't too
high at the start) it would also be useful to be able to specify a
connection group to be used just for server-server replication traffic.
(at least for self heal traffic, unsure about geo-replication)

Something like this:

  Create two connection groups
  ****************************

  $ sudo gluster connection-group create somegroup1 gluster1:172.16.101.1 gluster2:172.16.101.2 gluster3:172.16.101.3
  connection-group created: somegroup1

   (creates a group using the 1st Infiniband interface IP on my 3 gluster dev boxes)

  $ sudo gluster connection-group create somegroup2 gluster1:172.16.102.1 gluster2:172.16.102.2 gluster3:172.16.102.3
  connection-group created: somegroup2

   (creates a group using the 2nd Infiniband interface IP on my 3 gluster dev boxes)

  Associate the new connection groups with volumes
  ************************************************

  $ sudo gluster volume groupreplication myfirstvolume somegroup1
  volume myfirstvolume replication group: somegroup1

   (self heal traffic for "myfirstvolume" will now use the first IB interface's IP for my boxes)

  $ sudo gluster volume groupreplication mysecondvolume somegroup2
  volume mysecondvolume replication group: somegroup2

   (self heal traffic for "mysecondvolume" will now use the second IB interface's IP for my boxes)

Seems like a sensible approach (thus far). :)

+ Justin

--
Open Source and Standards @ Red Hat

twitter.com/realjustinclift





More information about the Gluster-devel mailing list