riak-admin down NODE meaning

Vladyslav Zakhozhai v.zakhozhai at smartweb.com.ua
Thu Jun 16 16:40:19 EDT 2016

I want to add one thing to my letter.

Yesturday (maybe the day before yesturday) I've joinde 3 node to cluster
plan. The I've reviewed it (to understand in which way my cluster will be
rebalanced). Then I've cleared this plan without commitment.

riak-admin member-status does not show the new nodes (that's ok). But when
I put the following in the riak console:
RCap = fun() ->
               {ok, Ring} = riak_core_ring_manager:get_my_ring(),
               [{Node, Props} || {Node, {invalid, Props, _Capabilities}} <-
element(8, Ring)]

I see

{'riak at boa.pleiad.uaprom',[{'riak at anole.pleiad.uaprom',{1,
                            {'riak at boa.pleiad.uaprom',{3,63632777924}}]},
 {'riak at boiga.pleiad.uaprom',[{'riak at anole.pleiad.uaprom',{1,
                              {'riak at boiga.pleiad.uaprom
 {'riak at cascavel.pleiad.uaprom',[{'riak at anole.pleiad.uaprom',{1,
                                 {'riak at cascavel.pleiad.uaprom
 {'riak at chameleon.pleiad.uaprom',[{'riak at ameiva.pleiad.uaprom',{4,

                                  {'riak at anole.pleiad.uaprom
                                  {'riak at chameleon.pleiad.uaprom
                                  {'riak at chuckwalla.pleiad.uaprom
                                  {'riak at cobra.gigatrans.uaprom
                                  {'riak at sonoran.pleiad.uaprom
 {'riak at chuckwalla.pleiad.uaprom',[{'riak at chuckwalla.pleiad.uaprom',{934,

                                   {'riak at cobra.gigatrans.uaprom
                                   {'riak at mussurana.pleiad.uaprom
                                   {'riak at yarara.pleiad.uaprom

boa, boiga and cascavel - this new nodes that was not joined to cluster.
chameleon and chuckwalla (and others) are old members of riak cluster.

Is it correct behaviour of riak or this is an issue?

On Thu, Jun 16, 2016 at 11:27 PM Vladyslav Zakhozhai <
v.zakhozhai at smartweb.com.ua> wrote:

> Hello,
> I thought that I know the meaning of command riak-admin down NODE pretty
> sure. Till this evening.
> So the question is what is explicit meaning of this command?
> My opinion was the following. During ownership handoff (maybe hinted
> handoffs also, I am not sure) when one of the members of riak cluster fails
> handoff stops till riak-admin down FAILED_NODE_NAME provided.
> For example there is 7 nodes: A, B, C, D, E, F, G. F and G are the nodes
> just joined to riak cluster and are awaiting to receive partitions from A,
> B and C. During ownership handoff node E fails. In this case all ownership
> handoffs may be stopped till riak-admin down E command is provided.
> Am I right? Or when I provide such command all ring will be replanned?
> Unfortunately description of this command is not clear in official docs.
> Today when my cluster plan was almost completed one of the riak nodes
> failed (this node was not a participant in ownership handoffs). And then in
> riak-admin ring-status I see 5 partitions is awaiting ownership transfer
> instead of only one.
> It confused me very match. I was looking forward to complete this cluster
> plan to join new nodes to cluster. In my case I need to wait for completion
> of ownership transfer for this 5 partition that can took 5-7 days (existing
> nodes experiencing lack of resources). Unfortunately I do not have so much
> time.
> Thank you.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.basho.com/pipermail/riak-users_lists.basho.com/attachments/20160616/f7a5dcc7/attachment-0002.html>

More information about the riak-users mailing list