See my comments inline, marked AndersW>

regards,

Anders Widell


On 02/08/2018 10:36 AM, Ravi Sekhar Reddy Konda wrote:
Hi Gary,

Have query regarding quorum selection when raft servers are external to the 
OpenSAF Cluster

In the document we are saying  "The consensus service uses quorum to prevent state 
changes in network partitions that don't include more than half of the nodes in the 
cluster"

=> This is possible if the raft server is installed on the OpenSAF Cluster 
Nodes, as Raft decides which partition has more no of nodes.
but in the case where raft servers run on external nodes outside of the OpenSAF 
Cluster, how the quorum is decided

AndersW> If the consensus service is running on external servers then you need to have an appropriate number of them (probably three or five). Quorum is determined as the majority of these external servers, and is not in any way related to majority of the OpenSAF nodes. The consensus service will prevent split-brain within the OpenSAF cluster, but in case of a network partition it will not guarantee that the active system controller will be located in the largest partition. This situation is actually similar to the situation when you use TIPC for internal OpenSAF communication. You can have a split-brain in the TIPC network (for example due to misconfiguration or a bug in TIPC), but at the same time have full connectivity on the IP network which is used by RAFT. I think there were some review comments about this for ticket [#64] and I will write a follow-up ticket where we can address the possibility of moving the active system controller to a node in the largest network partition.



=> If the Raft Servers are external to OpenSAF Cluster, do we need to make any 
configuration so that etcd client on the OpenSAF nodes
communicates with Raft Leader
Also it will be good if we give some details about how to install and configure 
raft(raft servers within and external to the opensaf cluster)

AndersW> This is slightly out of scope since there are many RAFT implementations, but I agree it could be a good idea to provide a sample configuration for etcd along with the sample etcd plugin.


Thanks,
Ravi

-----Original Message-----
From: Gary Lee [mailto:gary....@dektech.com.au]
Sent: Friday, January 26, 2018 11:28 AM
To: Hans Nordeb├Ąck <hans.nordeb...@ericsson.com>; Anders Widell 
<anders.wid...@ericsson.com>; Ravi Sekhar Reddy Konda <ravisekhar.ko...@oracle.com>
Cc: opensaf-devel@lists.sourceforge.net
Subject: Review Request for doc: update overview PR for split brain prevention 
with consensus service [#64]

Hi

I have updated the OpenSAF Overview PR document for ticket #64.

Please have a look.

https://urldefense.proofpoint.com/v2/url?u=https-3A__sourceforge.net_p_opensaf_tickets_-5Fdiscuss_thread_0d47d4b9_5489_attachment_OpenSAF-5FOverview-5FPR.odt&d=DwICaQ&c=RoP1YumCXCgaWHvlZYR8PZh8Bv7qIrMUB65eapI_JnE&r=xBh_3WtlS1YjXd3Bui_nVjh5qwhU2UamdAhSfqynLU4&m=xCEIb5x0gLGfoZW5uOWz23MZa6HzmOa6Vhywz3WeIQs&s=RF6RsX3xhby4k4PnwA8WEXCWKg0JbFyGNgaiery9iDk&e=

Thanks
Gary




------------------------------------------------------------------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
_______________________________________________
Opensaf-devel mailing list
Opensaf-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/opensaf-devel

Reply via email to