Check the "capabilities" parameter for adding the ceph backend [1]. you need to set replication to 3 (min_replication should automatically be set to 2).

I think the "system" command for it should be: 
system storage-backend-add ceph -s cinder,glance,swift,nova replication=3 --confirmed 

The help of the command states: 
$ system help storage-backend-add
usage: system storage-backend-add [-s <services>] [-n <name>] [-t <tier_uuid>]
                                  [-c <ceph_conf>] [--confirmed]
                                  [--ceph-mon-gib <ceph-mon-gib>]
                                  <backend>
                                  [<parameter=value> [<parameter=value> ...]] 

Replication should be given as a <parameter=value>.

Storage nodes inside a replication group all share the same (replicated) data. So if one node fails for any reason, the data is still being served by the remaining nodes. As, I said, nodes are assigned automatically to a group based on replication, so you cannot manually assign each node to a different group.

It is recommended that your number of storage nodes be a multiple of your replication factor. So for replication 2, you should use 2,4,6.... storage nodes and for replication 3 you should use 3,6,9... storage nodes.


[1]: https://docs.starlingx.io/api-ref/stx-config/api-ref-sysinv-v1-config.html?expanded=add-a-ceph-storage-backend-detail#add-a-ceph-storage-backend

From: von Hoesslin, Volker [Volker.Hoesslin@swsn.de]
Sent: Tuesday, October 30, 2018 1:34 PM
To: 'starlingx-discuss@lists.starlingx.io'
Subject: Re: [Starlingx-discuss] move storage node in replication group

Ok, I understand. how can I define this when creating a storage node? neither cli nor gui can I find corresponding options.

how are the "replication-groups" to be understood? is it desirable to put all three storage-nodes into one group or should each storage-node have its own group?

 

greez & thx,

  volker...

 

Von: Dinescu, Stefan [mailto:Stefan.Dinescu@windriver.com]
Gesendet: Dienstag, 30. Oktober 2018 10:48
An: von Hoesslin, Volker; 'starlingx-discuss@lists.starlingx.io'
Betreff: RE: move storage node in replication group

 

Hello,

 

The default replication of ceph is 2. Nodes are assigned to each group automatically and cannot be manually changed. if you want replication 3, you need you set that before installing storage nodes.


From: von Hoesslin, Volker [Volker.Hoesslin@swsn.de]
Sent: Monday, October 29, 2018 5:43 PM
To: 'starlingx-discuss@lists.starlingx.io'
Subject: [Starlingx-discuss] move storage node in replication group

hello,

i have 3 storage nodes and after provisioning nodes 1+2 are in replication group 0 and node 3 are in replication group 1?! how can i move node 3 to group 0?

 

greetings, volker