I'll guess that if you have wildcards with more letters in front of the * your queries are fine, right? The underlying problem is that Solr has to enumerate every term in your title field that starts with 'h' to see if they match your pattern.
You can require more letters at the beginning. You could also do something with reverse-wildcards if the pattern you're showing is typical. Possibly you could do something with NGrams, but that's questionable. Hope that helps Erick On Sat, Apr 16, 2011 at 4:55 PM, Charles Wardell <charles.ward...@bcsolution.com> wrote: > Hi All, > > I have an index with about 30M documents. For the most part queries are very > fast. However, when I add a wildcard to a search field. > +title:h*twitter it can take a few minutes. > > > 8GB > 1 quad core > CENTOS > > <useCompoundFile>false</useCompoundFile> > <mergeFactor>100</mergeFactor> > <ramBufferSizeMB>512</ramBufferSizeMB> > <maxBufferedDocs>100000</maxBufferedDocs> > > <maxFieldLength>10000</maxFieldLength> > <writeLockTimeout>1000</writeLockTimeout> > <commitLockTimeout>10000</commitLockTimeout> > > > <lockType>native</lockType> > <autoCommit> > <maxDocs>100000</maxDocs> > <maxTime>90000</maxTime> > </autoCommit> > > > > > >
