Solr 4.x and master-slave schema

[Jorge Luis Betancourt Gonzalez <jl...@uci.cu>]

Hi all:

We have a small installation of Solr 3.6 in our hands, right now we have 3 physical servers (1 master and 2 slaves) the ingestion process it’s done in the master which replicates by solr internal mechanism into the slaves, which handles all the queries. We are trying to update to Solr 4.x, eventually we would like to migrate into SolrCloud, my question essentially is if we migrate our Solr 3.6 nodes into Solr 4.9 and keep the same master-slave schema, how hard it would be to migrate afterwards to SorlCloud.

Greetings,VII Escuela Internacional de Verano en la UCI del 30 de junio al 11 de julio de 2014. Ver www.uci.cu

Re: Solr 4.x and master-slave schema

[Erick Erickson <er...@gmail.com>]

Shouldn't be hard at all. The only trick will be putting the
configuration files into ZooKeeper
from the collection that exists already.

Here's what I'd try (WARNING: this worked for me on a simple test, I
didn't test it robustly) if you can't
re-index:

1> copy your index somewhere for backup. This is just the
<SOLR_HOME>/collection/data/index directory.
2> Set up your collection in SolrCloud. This'll require that you use
your original conf directory.
    At this point the collection will be completely empty.
3> Shut Solr down
4> Copy your saved index into your core/data/index directory, after
removing everything that's there.
5> Start Solr.
6> use the collections API to add as many replicas as you want,
they'll synch with
    the leader and start serving queries  automatically when they're ready.

One warning: You _cannot_ do the process above if you want to go from
a single shard
to multiple shards. In that case you would be _far_ better off
re-indexing in cloud mode.

********************
If you _can_ just re-index everything, then just set up your SolrCloud
instance and index to it. Poof,
you're done. If you're using SolrJ, all you have to do is use the
CloudSolrServer class to
communicate with Solr (actually, you don't even _have_ to do that,
it's just preferred). If you're
indexing with post.jar, you don't have to change anything about your
indexing process.

*******************

I have to ask why you want to use the old master/slave replication in
the first place in 4.9?
Another way to approach this would be to set up your single node in
4.x right now and just
set it all up and go. But you know your constraints better than I do.
One of the beauties of
SolrCloud is that there's really very little you have to do to start
using it based on an older
infrastructure. There are a few minor tweaks to _improve_ the process,
but very little
other than setup that's required.

Best
Erick

On Thu, Jul 10, 2014 at 10:24 AM, Jorge Luis Betancourt Gonzalez
<jl...@uci.cu> wrote:
> Hi all:
>
> We have a small installation of Solr 3.6 in our hands, right now we have 3 physical servers (1 master and 2 slaves) the ingestion process it’s done in the master which replicates by solr internal mechanism into the slaves, which handles all the queries. We are trying to update to Solr 4.x, eventually we would like to migrate into SolrCloud, my question essentially is if we migrate our Solr 3.6 nodes into Solr 4.9 and keep the same master-slave schema, how hard it would be to migrate afterwards to SorlCloud.
>
> Greetings,VII Escuela Internacional de Verano en la UCI del 30 de junio al 11 de julio de 2014. Ver www.uci.cu