Probably because you're indexing a lot of documents very quickly. It's entirely reasonable to have much shorter autoCommit times, all that does is 1> truncate the transaction log 2> close the current segment 3> start a new segment.
That should cut down your tlog files drastically. Try setting your autocommit time to, say, 15000 (15 seconds). Long blog here: http://searchhub.org/2013/08/23/understanding-transaction-logs-softcommit-and-commit-in-sorlcloud/ Best, Erick On Tue, Sep 17, 2013 at 5:16 AM, YouPeng Yang <yypvsxf19870...@gmail.com>wrote: > Hi > According to > http://wiki.apache.org/solr/SolrPerformanceProblems#Slow_startup。 > It explains that the tlog file will swith to a new when hard commit > happened. > > > However,my tlog shows different. > tlog.0000000000000000003 5.16GB > tlog.0000000000000000004 1.56GB > tlog.0000000000000000002 610.MB > > there are only a fewer tlogs which suppose to be ten files, and each one > is vary huge.Even there are lots of hard commit happened. > > So why the number of the tlog files does not increase ? > > > here are settings of the DirectUpdateHandler2: > <updateHandler class="solr.DirectUpdateHandler2"> > > <updateLog> > <str name="dir">${solr.ulog.dir:}</str> > </updateLog> > > <autoCommit> > <maxTime>1200000</maxTime> > <maxDocs>1000000</maxDocs> > <openSearcher>false</openSearcher> > </autoCommit> > > > <autoSoftCommit> > <maxTime>600000</maxTime> > <maxDocs>500000</maxDocs> > </autoSoftCommit> > > </updateHandler> >
