Solr Cloud and Replication request handler

2014-04-28 Thread Amanjit Gill 
Hi everybody,

Considering a solr cloud configuration (4.6+)

a) I am wondering if the solr replication handler always has to be
configured completely, aka by choosing one master, then setting the config
accordingly (enable, masterUrl) etc ...  Do we really need a replication
master?

solrconfig.xml excerpt

requestHandler name=/replication class=solr.ReplicationHandler
 lst name=master
 str name=enabletrue/str  !-- true on master instance --
[..]
   /lst
  lst name=slave
 str name=enablefalse/str !-- true on slave instance --
 str name=masterUrlhttp://mysolrinstance::port
/default/replication/str
[..]
   /lst
/requestHandler

b) what happens to the cloud if the master instance goes down?

Thanks for your info ...

All the best,
Amanjit

Re: Solr Cloud and Replication request handler

2014-04-28 Thread Shawn Heisey 
On 4/28/2014 3:33 AM, Amanjit Gill wrote:
 Hi everybody,
 
 Considering a solr cloud configuration (4.6+)
 
 a) I am wondering if the solr replication handler always has to be
 configured completely, aka by choosing one master, then setting the config
 accordingly (enable, masterUrl) etc ...  Do we really need a replication
 master?


You simply need the replication handler to be present with a name of
/replication for SolrCloud to work properly.  You do not need to
configure it for master or slave.  SolrCloud will take care of
configuring which instance needs to be a slave whenever it needs to
recover an index.  You literally just need one line in your solrconfig.xml:

  requestHandler name=/replication class=solr.ReplicationHandler /

Thanks,
Shawn

Re: Solr Cloud and Replication request handler

2014-04-28 Thread Amanjit Gill 
Hello Shawn,

Thanks for your reply, that's good news!

All the best.


2014-04-28 15:28 GMT+02:00 Shawn Heisey s...@elyograg.org:

 On 4/28/2014 3:33 AM, Amanjit Gill wrote:
  Hi everybody,
 
  Considering a solr cloud configuration (4.6+)
 
  a) I am wondering if the solr replication handler always has to be
  configured completely, aka by choosing one master, then setting the
 config
  accordingly (enable, masterUrl) etc ...  Do we really need a replication
  master?


 You simply need the replication handler to be present with a name of
 /replication for SolrCloud to work properly.  You do not need to
 configure it for master or slave.  SolrCloud will take care of
 configuring which instance needs to be a slave whenever it needs to
 recover an index.  You literally just need one line in your solrconfig.xml:

   requestHandler name=/replication class=solr.ReplicationHandler /

 Thanks,
 Shawn