@Udo, I think Kirk explained it well — This issue was introduced very recently (right before we cut the release branch) and it has serious consequences (requires restarting the server).
- Aaron > On Aug 14, 2019, at 9:06 AM, Kirk Lund <kl...@apache.org> wrote: > > +1 to include this fix in 1.10.0 > > FYI: The race condition for this code path to throw NPE (which is > catastrophic and requires restarting the server) was introduced by commit > 279fa0 on July 31 for GEODE-7001. > > On Tue, Aug 13, 2019 at 6:22 PM Anthony Baker <aba...@pivotal.io> wrote: > >> Given that we’re trying to stabilize the release branch and this fix seems >> to *help* that I’m in favor of merging it. >> >> Anthony >> >> >>> On Aug 13, 2019, at 5:32 PM, Udo Kohlmeyer <u...@apache.com> wrote: >>> >>> @Aaron, is this an existing issue (i.e this was not introduced in a >> current refactor)? >>> >>> If the answer is anything other that "This will make the system stop >> working", I would vote: -1 >>> >>> If this is an existing issue and has been around for a while, I think we >> hold off including this. >>> >>> I think the boat has sailed on the inclusion of issues into the 1.10 >> release. Sorry... >>> >>> --Udo >>> >>> On 8/13/19 4:58 PM, Aaron Lindsey wrote: >>>> I’d like to propose including >> https://github.com/apache/geode/pull/3913/commits/6f1814d1f719cc06b13769c40a9d6d01f99f927c >> < >> https://github.com/apache/geode/pull/3913/commits/6f1814d1f719cc06b13769c40a9d6d01f99f927c> >> in the Geode 1.10 release. >>>> >>>> This commit fixes an issue where a NullPointerException is thrown from >> PartitionedRegion.getLocalSize() when the statistics callback sampler is >> invoked before a PartitionedRegion is initialized. >>>> >>>> - Aaron >>>> >>>> >>>> >> >>
