lucene-solr-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Markus Jelsma <>
Subject RE: Java 9
Date Tue, 07 Nov 2017 08:27:56 GMT

There won't be a Java 10, we'll get Java 18.3 instead. After 9 it is a guess when CMS and
friends are gone.


-----Original message-----
> From:Shawn Heisey <>
> Sent: Tuesday 7th November 2017 0:24
> To:
> Subject: Re: Java 9
> On 11/6/2017 3:07 PM, Petersen, Robert (Contr) wrote:
> > Anyone else been noticing this this msg when starting up solr with java 9? (This
is just an FYI and not a real question)
> >
> > Java HotSpot(TM) 64-Bit Server VM warning: Option UseConcMarkSweepGC was deprecated
in version 9.0 and will likely be removed in a future release.
> > Java HotSpot(TM) 64-Bit Server VM warning: Option UseParNewGC was deprecated in
version 9.0 and will likely be removed in a future release.
> I have not tried Java 9 yet.
> Looks like G1 is now the default garbage collector.  I did not know that
> they were deprecating CMS and ParNew ... that's a little surprising. 
> Solr's default garbage collection tuning uses those two collectors.  It
> is likely that those choices will be available in all versions of Java
> 9.  It would be very uncharacteristic for Oracle to take action on
> removing them until version 10, possibly later.
> If it were solely up to me, I would adjust Solr's startup script to use
> the G1 collector by default, eliminating the warnings you're seeing. 
> It's not just up to me though.  Lucene documentation says to NEVER use
> the G1 collector because they believe it to be unpredictable and have
> the potential to cause problems.  I personally have never had any issues
> with it.  There is *one* Lucene issue mentioning problems with G1GC, and
> that issue is *specific* to the 32-bit JVM, which is not recommended
> because of the limited amount of memory it can use.
> My experiments with GC tuning show the G1 collector (now default in Java
> 9) to have very good characteristics with Solr.  I have a personal page
> on the Solr wiki that covers those experiments.
> Thanks,
> Shawn

View raw message