Yes, this looks like a bug in the Google AllocationRecorder <https://github.com/google/allocation-instrumenter/blob/master/src/main/java/com/google/monitoring/runtime/instrumentation/AllocationRecorder.java> :
The NullPointerException we are seeing occurs when a shutdown hook allocates a new object. The instrumented bytecode still invokes the recordAllocation method. The AllocationRecorder also has a shutdown hook <https://github.com/google/allocation-instrumenter/blob/master/src/main/java/com/google/monitoring/runtime/instrumentation/AllocationRecorder.java#L49> which sets its instrumentation field to null. These shutdown hook threads run at the same time, so if the instrumentation field is set to null after the null check <https://github.com/google/allocation-instrumenter/blob/master/src/main/java/com/google/monitoring/runtime/instrumentation/AllocationRecorder.java#L236> but before it is used in getObjectSize <https://github.com/google/allocation-instrumenter/blob/master/src/main/java/com/google/monitoring/runtime/instrumentation/AllocationRecorder.java#L192>, we get the exception. On Wed, Apr 27, 2016 at 12:29 AM, Ralph Goers <ralph.go...@dslextreme.com> wrote: > Any idea how to determine what caused this to fail? > > Ralph > > > On Apr 26, 2016, at 8:07 AM, Apache Jenkins Server < > jenk...@builds.apache.org> wrote: > > > > See <https://builds.apache.org/job/Log4j%202.x/1909/changes> > > > > > > --------------------------------------------------------------------- > > To unsubscribe, e-mail: log4j-dev-unsubscr...@logging.apache.org > > For additional commands, e-mail: log4j-dev-h...@logging.apache.org > > > > > > > > --------------------------------------------------------------------- > To unsubscribe, e-mail: log4j-dev-unsubscr...@logging.apache.org > For additional commands, e-mail: log4j-dev-h...@logging.apache.org > >
