IIRC, a Get request will query both the Lucene index and the transaction
log (real-time), while any search will only access the Lucene index (near
real time).
correct. Whenever you are doing a search, you are always relying on lucene
search capabilities (even in case you are doing a simple id filter), which
means a refresh has to happen, before the data you indexed is available for
search. A search does not query the transaction log (because it is a simple
journal and not an inverted index suitable for search), where as a GET
(which simply uses the id) can check in the transaction log.
I am currently trying to sum up a little bit, what happens under the hood,
how the transaction log is used to provide real time get (how the
transaction log is used for recovery on startup), etc... hopefully I can
make a real blog post from that.
--Alex
On Wed, Apr 10, 2013 at 6:10 PM, Ivan Brusic ivan@brusic.com wrote:
IIRC, a Get request will query both the Lucene index and the transaction
log (real-time), while any search will only access the Lucene index (near
real time).
Apache, Apache Lucene, Apache Hadoop, Hadoop, HDFS and the yellow elephant
logo are trademarks of the
Apache Software Foundation
in the United States and/or other countries.