Bonding is just one particular case which HA handles. There are many other applications of HA.<br>Suppose you have 3 servers each exporting a storage brick and also an AFR volume of bricks from three servers, you can HA all the 3 AFR subvolumes and you will have availability of glusterfs till the last server is reachable.<br>
<br>--<br>Gowda<br><br><div class="gmail_quote">2009/2/23 William Hanwoody <span dir="ltr"><<a href="mailto:hanwoody@gmail.com">hanwoody@gmail.com</a>></span><br><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
<div><div></div><div class="Wj3C7c"><br>
> On Sun, Feb 22, 2009 at 11:06 PM, Jeffry Molanus<br>
><br>
> <<a href="mailto:jeffry.molanus@gmail.com">jeffry.molanus@gmail.com</a>>wrote:<br>
> > On Sun, 2009-02-22 at 21:34 +0530, Basavanagowda Kanur wrote:<br>
> > > Davide,<br>
> > > Replies inline.<br>
> > ><br>
> > > On Sun, Feb 22, 2009 at 2:24 PM, Davide D'Amico<br>
> > > <<a href="mailto:davide.damico@gmail.com">davide.damico@gmail.com</a>> wrote:<br>
> > > Hi,<br>
> > > I was looking for a scalable-ha nfs storage solution and<br>
> > > glusterfs seems to be<br>
> > > the only out there :-)<br>
> > ><br>
> > > I was reading the documentation and I cannot find the<br>
> > > difference<br>
> > > between "Replicate<br>
> > > Translator" and "HA Translator": the latest seems to be a<br>
> > > particular case of the<br>
> > > first one when I have only two bricks. Is it right?<br>
> > > Replicate translator provides replication feature, whereas HA (High<br>
> > > Availability) translator provides high availability, HA does not do<br>
> > > any replication.<br>
> ><br>
> > Care to explain how it does work then? How can a file be HA if the<br>
> > server that went down is not replicated?<br>
><br>
> HA is useful when you have multiple network interconnect to the server and<br>
> you want glusterfs to continue working seemlessly, then you would use HA.<br>
> HA fails over to secondary interconnect, if primary fails.<br>
<br>
</div></div>I think the name of "HA" is mismatch. For replication is also "HA" when one of<br>
storage node went down.<br>
<br>
HA of gluster is more like bonding of multiple network interface,and I think<br>
it is no need for glusterfs, because we can do bonding!!<br>
<div><div></div><div class="Wj3C7c"><br>
><br>
> > > I intend to create a scalable nfs cluster, so I can add<br>
> > > storage nodes<br>
> > > to increase<br>
> > > storage size: have I to use unify on top of AFR translator<br>
> > > (i.e. unify<br>
> > > on a replicated<br>
> > > volume?).<br>
> > > you can use 'distribute' (formerly DHT) translator over AFR.<br>
> > ><br>
> > ><br>
> > ><br>
> > > I've read that glusterfs has problems with self-healing and<br>
> > > opened<br>
> > > files: is it true?<br>
> > ><br>
> > ><br>
> > > Thanks in advance,<br>
> > ><br>
> > > --<br>
> > > d.<br>
> > ><br>
> > > _______________________________________________<br>
> > > Gluster-users mailing list<br>
> > > <a href="mailto:Gluster-users@gluster.org">Gluster-users@gluster.org</a><br>
> > > <a href="http://zresearch.com/cgi-bin/mailman/listinfo/gluster-users" target="_blank">http://zresearch.com/cgi-bin/mailman/listinfo/gluster-users</a><br>
> > ><br>
> > ><br>
> > ><br>
> > > --<br>
> > > Gowda<br>
> > > _______________________________________________<br>
> > > Gluster-users mailing list<br>
> > > <a href="mailto:Gluster-users@gluster.org">Gluster-users@gluster.org</a><br>
> > > <a href="http://zresearch.com/cgi-bin/mailman/listinfo/gluster-users" target="_blank">http://zresearch.com/cgi-bin/mailman/listinfo/gluster-users</a><br>
<br>
<br>
</div></div></blockquote></div><br><br clear="all"><br>