Hello all,
I am running DSpace 3.2 (JSPUI) on Debian Linux, using Tomcat7 (assigned
memory: 2GB), Java 1.8.0_66, a 29GB assetstore and a 1.2GB Solr index (the
PDF Text Extractor media filter is enabled).
I want to make a special item browse index that will only show items of
certain types. My approach was to do this on query-time, i.e. by adding a
filter query of the form *dc.type:(Book Tutorial "Lab Guide")* in
org.dspace.browse.SolrBrowseDAO.java for this browse index. However, the
browse page takes a lot of time to load: by adding appropriate statements
inside the code, I was able to identify that the main reason is the slow
Solr response (sometimes, as much as 40 sec), rather than the retrieval of
the necessary data for filling in the browse table in
org.dspace.browse.BrowseListTag (which also takes long: ~4 sec, but that
may be due to the inclusion of thumbnails and the fact that some of them
are in the order of ~100-200 KB). I tried disabling the Spellcheck and
MoreLikeThis components in solrconfig.xml and made dc.type a "stored" field
in Solr (by adding it to the *index.projection* configuration in
discovery.cfg), but I didn't observe any significant improvement.
Furthermore, in org.dspace.browse.BrowseEngine, I noticed that every query
is run at least twice, first to get the number of total results and then to
get the actual results. I expected that, under normal conditions, the
second query would execute much faster and I managed to verify this in
another DSpace installation by inspecting the Tomcat logs. However, in my
case, both queries take comparable time to execute. I also observe some
difference between QTime and the actual time that I measure, were they
supposed to be nearly the same? e.g. in Tomcat i see the following QTime
INFO: [search] webapp=/solr path=/select
params={fq=NOT(withdrawn:true)&fq=NOT(withdrawn:true)&fq=dc.type:(Book+Tutorial+"Lab+Guide")&fq=search.resourcetype:2&fq=read:(g0)&wt=javabin&sort=bi_sort_1_sort+asc&q=*:*&start=0&version=2&rows=40}
hits=252 status=0 QTime=1576
while my DSpace log shows that
org.dspace.browse.BrowseEngine.getTotalResults() took 19355 ms to run.
I would like to know if there is some sort of configuration that could be
made to improve Solr performance or any other ideas/ways to achieve the
same result (perhaps a more efficient Solr query?).
Thank you in advance, any hint or help is greatly appreciated.
--
You received this message because you are subscribed to the Google Groups
"DSpace Technical Support" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To post to this group, send email to [email protected].
Visit this group at https://groups.google.com/group/dspace-tech.
For more options, visit https://groups.google.com/d/optout.
