To start safely you need a replication factor of 3 and at least 4 nodes (think size+1) to allow for smooth maintenance on your nodes.
On Fri, Jul 1, 2016 at 2:31 PM, Ashley Merrick <ashley@xxxxxxxxxxxxxx> wrote:
Hello,
Okie makes perfect sense.
So if run CEPH with a replication of 3, is it still required to run an odd number of OSD Nodes.
Or could I run 4 OSD Nodes to start with, with a replication of 3, with each replication on a separate server.
,Ashley Merrick
-----Original Message-----
From: ceph-users [mailto:ceph-users-bounces@xxxxxxxxxxxxxx] On Behalf Of Tomasz Kuzemko
Sent: 01 July 2016 13:28
To: ceph-users@xxxxxxxxxxxxxx
Subject: Re: CEPH Replication
Still in case of object corruption you will not be able to determine which copy is valid. Ceph does not provide data integrity with filestore (it's planned for bluestore).
On 01.07.2016 14:20, David wrote:
> It will work but be aware 2x replication is not a good idea if your
> data is important. The exception would be if the OSD's are DC class
> SSD's that you monitor closely.
>
> On Fri, Jul 1, 2016 at 1:09 PM, Ashley Merrick <ashley@xxxxxxxxxxxxxx
> <mailto:ashley@xxxxxxxxxxxxxx>> wrote:
>
> Hello,
>
> Perfect, I want to keep on separate node's, so wanted to make sure
> the expected behaviour was that it would do that.
>
> And no issues with running an odd number of nodes for a replication
> of 2? I know you have quorum, just wanted to make sure would not
> effect when running an even replication.
>
> Will be adding nodes in future as require, but will always keep an
> uneven number.
>
> ,Ashley
>
> -----Original Message-----
> From: ceph-users [mailto:ceph-users-bounces@xxxxxxxxxxxxxx
> <mailto:ceph-users-bounces@xxxxxxxxxxxxxx>] On Behalf Of
> ceph@xxxxxxxxxxxxxx <mailto:ceph@xxxxxxxxxxxxxx>
> Sent: 01 July 2016 13:07
> To: ceph-users@xxxxxxxxxxxxxx <mailto:ceph-users@xxxxxxxxxxxxxx>
> Subject: Re: CEPH Replication
>
> It will put each object on 2 OSD, on 2 separate node All nodes, and
> all OSDs will have the same used space (approx)
>
> If you want to allow both copies of an object to put stored on the
> same node, you should use osd_crush_chooseleaf_type = 0 (see
> http://docs.ceph.com/docs/master/rados/operations/crush-map/#crush-map-bucket-types
> and
>
> http://docs.ceph.com/docs/hammer/rados/configuration/pool-pg-config-re
> f/)
>
>
> On 01/07/2016 13:49, Ashley Merrick wrote:
> > Hello,
> >
> > Looking at setting up a new CEPH Cluster, starting with the following.
> >
> > 3 x CEPH OSD Servers
> >
> > Each Server:
> >
> > 20Gbps Network
> > 12 OSD's
> > SSD Journal
> >
> > Looking at running with replication of 2, will there be any issues
> using 3 nodes with a replication of two, this should "technically"
> give me ½ the available total capacity of the 3 node's?
> >
> > Will the CRUSH map automaticly setup each 12 OSD's as a separate
> group, so that the two replicated objects are put on separate OSD
> servers?
> >
> > Thanks,
> > Ashley
> >
> >
> >
> >
> > _______________________________________________
> > ceph-users mailing list
> > ceph-users@xxxxxxxxxxxxxx <mailto:ceph-users@xxxxxxxxxxxxxx>
> > http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
> >
>
> _______________________________________________
> ceph-users mailing list
> ceph-users@xxxxxxxxxxxxxx <mailto:ceph-users@xxxxxxxxxxxxxx>
> http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
> _______________________________________________
> ceph-users mailing list
> ceph-users@xxxxxxxxxxxxxx <mailto:ceph-users@xxxxxxxxxxxxxx>
> http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
>
>
>
>
> _______________________________________________
> ceph-users mailing list
> ceph-users@xxxxxxxxxxxxxx
> http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
>
--
Tomasz Kuzemko
tomasz.kuzemko@xxxxxxxxxxxx
_______________________________________________
ceph-users mailing list
ceph-users@xxxxxxxxxxxxxx
http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
_______________________________________________ ceph-users mailing list ceph-users@xxxxxxxxxxxxxx http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
