lucene-solr-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Erick Erickson <erickerick...@gmail.com>
Subject Re: shard lost - solr5.3
Date Mon, 04 Jan 2016 16:34:57 GMT
There's no reason to shut down your node. You should be able
to issue a REBALANCELEADERS command, see:
https://cwiki.apache.org/confluence/display/solr/Collections+API#CollectionsAPI-RebalanceLeaders

on a currently-running cluster and all your preferred leaders
(assuming the nodes are up) should become the leader of their respective shards.

I should emphasize, though, that this is rarely necessary unless you
have lots and lots
and lots of shards. The use-case that code was written for was,
literally, hundreds
of shards all had their leaders on a single node.

FWIW,
Erick

On Mon, Jan 4, 2016 at 3:26 AM, GOURAUD Emmanuel <egouraud@jouve.fr> wrote:
> hi there
>
> replying to myself
>
> i have set the replica property "preferredLeader" on this shard, shut down all replica
for this shard and started only the "preferred" one, this forced an election and save my "ops"
night and my new year party!!
>
> cheers,
>
> Emmanuel
>
>
>
> De: "GOURAUD Emmanuel" <egouraud@jouve.fr>
> À: solr-user@lucene.apache.org
> Envoyé: Jeudi 31 Décembre 2015 15:30:42
> Objet: shard lost - solr5.3
>
> Hi there,
>
> I have a collection that is composed of 8 shards with a replicationFactor of 2
>
> i found 2 cores of the same shard in recoveryfailed status so i decided to restart both,
>
> after having doing that , i do not have any leader on that shard... and both cores are
down
>
> is there a way to force a leader at startup or with the API? can fore election?
>
> thanks for your help
>
> Emmanuel
>

Mime
View raw message