lucene-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Mark Miller (JIRA)" <>
Subject [jira] [Commented] (SOLR-4434) Developer Curb Appeal: Better options than the manual copy step, and doc changes
Date Thu, 21 Feb 2013 01:56:11 GMT


Mark Miller commented on SOLR-4434:

bq. That really only makes sense if you're trying to run multiple nodes on a single laptop.

Which is what you are doing in the documentation that you refer to on the Apache wiki.

bq. I don't fully understand the distribution of labor between the wiki and Lucid's search
hub. Not sure who "keeps them in sync".

Lucid does - like I said, it has nothing to do with the Apache community. Our stuff is only
on the Apache Solr wiki.
> Developer Curb Appeal: Better options than the manual copy step, and doc changes
> --------------------------------------------------------------------------------
>                 Key: SOLR-4434
>                 URL:
>             Project: Solr
>          Issue Type: Improvement
>          Components: SolrCloud
>    Affects Versions: 4.1
>            Reporter: Mark Bennett
>             Fix For: 4.2
> We make developers manually copy the example directory to a named shard directory.
> Doc references:
> Sample commands:
> cp -r example shard1
> cp -r example shard2
> The doc is perhaps geared towards a developer laptop, so in that case you really would
need to make sure they have different names.
> But if you're running on a more realistic multi-node system, let's say 4 nodes handling
2 shards, the the actual shard allocation (shard1 vs. shard2) will be fixed by the order each
node is started in FOR THE FIRST TIME.
> At a minimum, we should do a better job of explaining the somewhat arbitrary nature of
the destination directories, and that the start order is what really matters.
> We should also document that the actual shard assignment will not change, regardless
of the name, and where this information is persisted?
> Could we have an intelligent guess as to what template directory to use, and do the copy
when the node is first started.
> It's apparently also possible to startup the first Solr node with no cores and just point
it at a template.  This would be good to document.  There's currently a bug in the Web UI
if you do this, but I'll be logging another JIRA for that.
> When combined with all the other little details of bringing up Solr Cloud nodes, this
is confusing to a newcomer and midly annoying.  Other engines don't require this.

This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see:

To unsubscribe, e-mail:
For additional commands, e-mail:

View raw message