G1 warming on lucene wiki

classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view
|

G1 warming on lucene wiki

Jeff Courtade
Hello,

I have been looking into tuning the garbage collector for solr. I found
this entry on the lucene wiki that seems to be out of date.

The bug referenced is reported as resolved now. Could someone validate
whether it is safe to use G1 garbage collection with lucene?

"Do not, under any circumstances, run Lucene with the G1 garbage collector.
Lucene's test suite fails with the G1 garbage collector on a regular basis,
including bugs that cause index corruption. There is no person on this
planet that seems to understand such bugs (see
https://bugs.openjdk.java.net/browse/JDK-8038348, open for over a year), so
don't count on the situation changing soon. This information is not out of
date, and don't think that the next oracle java release will fix the
situation."

https://wiki.apache.org/lucene-java/JavaBugs
Reply | Threaded
Open this post in threaded view
|

Re: G1 warming on lucene wiki

Jerven Tjalling Bolleman
Hi Jeff,

Not sure about the specific bug. But as an data point for
www.uniprot.org we have run with G1 for about 5 years now
without this issue. Looking at the Bug report it is actually
a HotSpot C2 bug. If you are moving to JDK11 they are rarer
in this area as there is an consistent API for most GC work now.

Actually if you are on RedHat OpenJDK builds then ShenandoahGC
is very interesting as well. Especially regarding pausetimes.

Regards,
Jerven



On 09/27/2018 02:24 PM, Jeff Courtade wrote:

> Hello,
>
> I have been looking into tuning the garbage collector for solr. I found
> this entry on the lucene wiki that seems to be out of date.
>
> The bug referenced is reported as resolved now. Could someone validate
> whether it is safe to use G1 garbage collection with lucene?
>
> "Do not, under any circumstances, run Lucene with the G1 garbage collector.
> Lucene's test suite fails with the G1 garbage collector on a regular basis,
> including bugs that cause index corruption. There is no person on this
> planet that seems to understand such bugs (see
> https://bugs.openjdk.java.net/browse/JDK-8038348, open for over a year), so
> don't count on the situation changing soon. This information is not out of
> date, and don't think that the next oracle java release will fix the
> situation."
>
> https://wiki.apache.org/lucene-java/JavaBugs
>


---------------------------------------------------------------------
To unsubscribe, e-mail: [hidden email]
For additional commands, e-mail: [hidden email]