Thanks Erick and Shawn for your reply. We have recently upgraded the server RAM from 64MB to 192MB, and I noticed that this caching occurs after we upgraded the RAM. Previously, the cache may not even be preserved in the same Solr session. So is it true that the upgrading of the server RAM creates enough spare memory for good caching?
Regards, Edwin On 28 January 2016 at 12:27, Shawn Heisey <apa...@elyograg.org> wrote: > On 1/27/2016 8:11 PM, Zheng Lin Edwin Yeo wrote: > > I would like to find out, is the cache in the Solr cleared when I shut > down > > the Solr instant and restart it? > > > > I am suspecting that the cache is not entirely cleared, because when I > try > > to do a search on the same query as I did before the search, it still > has a > > return QTime that is much faster than the initial search. However, when I > > do a search on a new query, the return QTime is the original speed. > > > > I am using Solr 5.4.0, and this is my setting for the queryResultCache. > > > > <queryResultCache class="solr.LRUCache" > > size="1024" > > initialSize="512" > > autowarmCount="0"/> > > The Solr caches are maintained in the Java heap, which is lost when Java > stops. > > Although the Solr caches are not preserved across a restart, the > operating system does cache actual index data in main memory, so when > Solr asks for the same index data off of the disk again, it is pulled > directly from RAM, which is a LOT faster than the disk. In order to > deliver good performance, Solr is extremely reliant on this built-in > feature of all modern operating systems, so there must be enough spare > memory for good caching. > > Here are a couple of pages with some more detail: > > https://wiki.apache.org/solr/SolrPerformanceProblems > https://en.wikipedia.org/wiki/Page_cache > > Thanks, > Shawn > >
