I'm not exactly a long-time user of Solr yet so forgive me if I'm
missing something really obvious here... but, if I'm not mistaken, you
just add <commit/> to the end of your update XML. This is how I've
done all of my commits, I wasn't even aware there was a CLI method to
this.
I would worry about the performance implications of doing things this
way if you have many frequent updates, though.
--
Steve
On Aug 7, 2008, at 5:26 PM, Jacob Singh wrote:
Hi,
I'm using the XML based update interface, and feeding requests to
update
the index via jetty. It all works great, however now I'm trying to
get
replication running, and here's what I understand:
1. An index update comes in.
2. Solr runs the commit script
3. a post-commit event is specified in solrconfig.xml:
<listener event="postCommit" class="solr.RunExecutableListener">
<str name="exe">snapshooter</str>
....
4. This creates a snapshot of the index and puts it in data/
SNAPSHOT...
5. The slaves run on cron, ssh in to the master, and basically just ls
the data dir, find if there is a new snapshot and rsync it
6. the slaves "warm" the index and swap it.
7. snapcleaner runs on both machines to clear out old snapshots.
Okay... so this all makes sense, and I can sorta make it work,
however,
the commit is never run by itself when the index is updated. I have
to
manually call the commit script from the command line to get it to
happen.
Do I have to call something additional via the webservice?
Am I missing a config option in solrconfig.xml? (I'm using the
example,
so I wouldn't think so).
Thanks a lot!
Jacob
