Re: Auto-sharding and numShard parameter

2013-07-22 Thread Michael Della Bitta
That would be great. One step toward this goal is to stop treating the situation where there are no collections or cores as an error condition. It took me a while to get out of the mindset when bringing up a Solr install that I had to avoid that scenario at all costs, because red text == bad.

Re: Auto-sharding and numShard parameter

2013-07-22 Thread Mark Miller
There is a reason of course, or else it wouldn't be like that. We addressed it recently. https://issues.apache.org/jira/browse/SOLR-3633 https://issues.apache.org/jira/browse/SOLR-3677 https://issues.apache.org/jira/browse/SOLR-4943 - Mark On Jul 22, 2013, at 10:57 AM, Michael Della Bitta

Re: Auto-sharding and numShard parameter

2013-07-20 Thread Erick Erickson
Flavio: One of the great things about having people continually using Solr (and SolrCloud) for the first time is the opportunity to improve the docs. Anyone can update/add to the docs, all it takes is a signon. Unfortunately we has a bunch of spam bots a while ago, so it's now a two step process

Re: Auto-sharding and numShard parameter

2013-07-20 Thread Mark Miller
A lot has changed since those example were written - in general, we are moving away from that type of collection initialization and towards using the Collections API. Eventually, I'd personally like SolrCloud to ship with no predefined collections and have users simply start it and then start

Re: Auto-sharding and numShard parameter

2013-07-19 Thread Erick Erickson
First the numShards parameter is only relevant the very first time you create your collection. It's a little confusing because in the SolrCloud examples you're getting collection1 by default. Look further down the SolrCloud Wiki page, the section titled Managing Collections via the Collections API

Re: Auto-sharding and numShard parameter

2013-07-19 Thread Flavio Pompermaier
Thank you for the reply Erick, I was facing exactly with that problem..from the documentation it seems that those parameter are required to run SolrCloud, instead they are just used to initialize a sample collection.. I think that in the examples on the user doc it should be better to separate

Auto-sharding and numShard parameter

2013-07-18 Thread Flavio Pompermaier
Hi to all, Probably this question has a simple answer but I just want to be sure of the potential drawbacks..when I run SolrCloud I run the main solr instance with the -numShard option (e.g. 2). Then as data grows, shards could potentially become a huge number. If I hadstio to restart all nodes