[ML] Autodetect fatal error in gallery 302

[sophiec20]

Found in "native_code_info": { "version": "7.0.0-alpha1-SNAPSHOT", "build_hash": "d219beab8c7d34" }

Linux version 3.10.0-693.11.1.el7.x86_64 ([email protected]) (gcc version 4.8.5 20150623 (Red Hat 4.8.5-16) (GCC) ) #1 SMP Mon Dec 4 23:52:40 UTC 2017

• Using gallery2018 dataset
• count at bucket_span=15m
• datafeed query {"bool":{"must":[{"term":{"status":{"value":"302"}}}]}}
• Datafeed started (from: 1970-01-02T10:00:00.000Z to: 2018-12-31T00:00:00.000Z)

Job fails with unexpected death of autodetect.
Occurs in repeated runs, at same latest timestamp 2018-10-09 15:29:11 (UTC).
This corresponds with the end of the timeseries for 302; however job end date was specified as 2018-12-31 00:00:00

[2018-05-14T10:14:04,309][INFO ][o.e.x.m.a.TransportPutDatafeedAction] [node1] Created datafeed [datafeed-ga3-count-302-15m]
[2018-05-14T10:14:04,499][INFO ][o.e.x.m.j.p.a.AutodetectProcessManager] [node1] Opening job [ga3-count-302-15m]
[2018-05-14T10:14:04,555][INFO ][o.e.x.m.j.p.a.AutodetectProcessManager] [node1] [ga3-count-302-15m] Loading model snapshot [N/A], job latest_record_timestamp [N/A]
[2018-05-14T10:14:04,705][INFO ][o.e.x.m.j.p.l.CppLogMessageHandler] [ga3-count-302-15m] [autodetect/6086] [CResourceMonitor.cc@67] Setting model memory limit to 20 MB
[2018-05-14T10:14:04,740][INFO ][o.e.x.m.j.p.a.AutodetectProcessManager] [node1] Successfully set job state to [opened] for job [ga3-count-302-15m]
[2018-05-14T10:14:05,227][INFO ][o.e.x.m.d.DatafeedJob    ] [ga3-count-302-15m] Datafeed started (from: 1970-01-02T10:00:00.000Z to: 2018-12-31T00:00:00.000Z) with frequency [450000ms]
[2018-05-14T10:14:06,887][INFO ][o.e.x.m.j.p.DataCountsReporter] [node1] [ga3-count-200-15m] 700000 records written to autodetect; missingFieldCount=0, invalidDateCount=0, outOfOrderCount=0
[2018-05-14T10:14:26,360][INFO ][o.e.x.m.a.TransportPutDatafeedAction] [node1] Created datafeed [datafeed-ga3-count-400-15m]
[2018-05-14T10:14:48,150][INFO ][o.e.x.m.a.TransportPutDatafeedAction] [node1] Created datafeed [datafeed-ga3-count-303-15m]
[2018-05-14T10:14:51,946][INFO ][o.e.x.m.j.p.DataCountsReporter] [node1] [ga3-count-200-15m] 800000 records written to autodetect; missingFieldCount=0, invalidDateCount=0, outOfOrderCount=0
[2018-05-14T10:14:52,463][ERROR][o.e.x.m.j.p.a.NativeAutodetectProcess] [ga3-count-302-15m] autodetect process stopped unexpectedly: Fatal error: 'si_signo 11, si_code: 1, si_errno: 0, address: 0x7f39a97c3eaa, library: /opt/elastic/7.0.0-20180511/elasticsearch-7.0.0-alpha1-SNAPSHOT/modules/x-pack/x-pack-ml/platform/linux-x86_64/bin/../lib/libMlMaths.so, base: 0x7f39a948e000, normalized address: 0x335eaa'

[2018-05-14T10:14:52,465][INFO ][o.e.x.m.j.p.a.NativeAutodetectProcess] [ga3-count-302-15m] State output finished
[2018-05-14T10:14:52,482][ERROR][o.e.x.m.j.p.l.CppLogMessageHandler] [controller/24993] [CDetachedProcessSpawner.cc@184] Child process with PID 6086 was terminated by signal 11
[2018-05-14T10:14:52,529][INFO ][o.e.x.m.j.p.a.AutodetectProcessManager] [node1] Successfully set job state to [failed] for job [ga3-count-302-15m]
[2018-05-14T10:14:52,763][ERROR][o.e.x.m.j.p.a.AutodetectCommunicator] [ga3-count-302-15m] Unexpected exception writing to process
org.elasticsearch.ElasticsearchException: [ga3-count-302-15m] Unexpected death of autodetect: Fatal error: 'si_signo 11, si_code: 1, si_errno: 0, address: 0x7f39a97c3eaa, library: /opt/elastic/7.0.0-20180511/elasticsearch-7.0.0-alpha1-SNAPSHOT/modules/x-pack/x-pack-ml/platform/linux-x86_64/bin/../lib/libMlMaths.so, base: 0x7f39a948e000, normalized address: 0x335eaa'

        at org.elasticsearch.xpack.ml.job.process.autodetect.AutodetectCommunicator.checkProcessIsAlive(AutodetectCommunicator.java:307) ~[?:?]
        at org.elasticsearch.xpack.ml.job.process.autodetect.AutodetectCommunicator.waitFlushToCompletion(AutodetectCommunicator.java:282) ~[?:?]
        at org.elasticsearch.xpack.ml.job.process.autodetect.AutodetectCommunicator.lambda$flushJob$4(AutodetectCommunicator.java:241) ~[?:?]
        at org.elasticsearch.xpack.ml.job.process.autodetect.AutodetectCommunicator$1.doRun(AutodetectCommunicator.java:363) ~[?:?]
        at org.elasticsearch.common.util.concurrent.ThreadContext$ContextPreservingAbstractRunnable.doRun(ThreadContext.java:724) ~[elasticsearch-7.0.0-alpha1-SNAPSHOT.jar:7.0.0-alpha1-SNAPSHOT]
        at org.elasticsearch.common.util.concurrent.AbstractRunnable.run(AbstractRunnable.java:37) ~[elasticsearch-7.0.0-alpha1-SNAPSHOT.jar:7.0.0-alpha1-SNAPSHOT]
        at org.elasticsearch.xpack.ml.job.process.autodetect.AutodetectProcessManager$AutodetectWorkerExecutorService.start(AutodetectProcessManager.java:678) ~[?:?]
        at java.util.concurrent.Executors$RunnableAdapter.call(Executors.java:511) [?:1.8.0_151]
        at java.util.concurrent.FutureTask.run(FutureTask.java:266) [?:1.8.0_151]
        at org.elasticsearch.common.util.concurrent.ThreadContext$ContextPreservingRunnable.run(ThreadContext.java:625) [elasticsearch-7.0.0-alpha1-SNAPSHOT.jar:7.0.0-alpha1-SNAPSHOT]
        at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149) [?:1.8.0_151]
        at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624) [?:1.8.0_151]
        at java.lang.Thread.run(Thread.java:748) [?:1.8.0_151]
[2018-05-14T10:14:52,767][ERROR][o.e.x.m.j.p.a.AutodetectProcessManager] [node1] [ga3-count-302-15m] exception while flushing job
[2018-05-14T10:14:52,772][INFO ][o.e.x.m.d.DatafeedManager] [no_realtime] attempt to stop datafeed [datafeed-ga3-count-302-15m] for job [ga3-count-302-15m]
[2018-05-14T10:14:52,780][INFO ][o.e.x.m.d.DatafeedManager] [no_realtime] try lock [20s] to stop datafeed [datafeed-ga3-count-302-15m] for job [ga3-count-302-15m]...
[2018-05-14T10:14:52,780][INFO ][o.e.x.m.d.DatafeedManager] [no_realtime] stopping datafeed [datafeed-ga3-count-302-15m] for job [ga3-count-302-15m], acquired [true]...
[2018-05-14T10:14:52,781][INFO ][o.e.x.m.d.DatafeedManager] [no_realtime] datafeed [datafeed-ga3-count-302-15m] for job [ga3-count-302-15m] has been stopped
[2018-05-14T10:14:54,614][WARN ][o.e.x.m.j.p.a.o.AutoDetectResultProcessor] [ga3-count-302-15m] some results not processed due to the termination of autodetect

[sophiec20]

(gdb) bt
#0  0x00007f9a42d4a59b in raise () from /lib64/libpthread.so.0
#1  0x00007f9a41ffed80 in ml::core::crashHandler(int, siginfo*, void*) () from /opt/elastic/7.0.0-20180511/elasticsearch-7.0.0-alpha1-SNAPSHOT/modules/x-pack/x-pack-ml/platform/linux-x86_64/bin/../lib/libMlCore.so
#2  <signal handler called>
#3  0x00007f9a41b97eaa in ml::maths::CTimeSeriesDecompositionDetail::CComponents::reweightOutliers(long, long, std::function<double (long)>, std::vector<ml::maths::CBasicStatistics::SSampleCentralMoments<ml::core::CFloatStorage, 1u>, std::allocator<ml::maths::CBasicStatistics::SSampleCentralMoments<ml::core::CFloatStorage, 1u> > >&) const () from /opt/elastic/7.0.0-20180511/elasticsearch-7.0.0-alpha1-SNAPSHOT/modules/x-pack/x-pack-ml/platform/linux-x86_64/bin/../lib/libMlMaths.so
#4  0x00007f9a41ba05b5 in ml::maths::CTimeSeriesDecompositionDetail::CComponents::addSeasonalComponents(ml::maths::CPeriodicityHypothesisTestsResult const&, ml::maths::CExpandingWindow const&, std::function<double (long)> const&, ml::maths::CTrendComponent&, std::vector<ml::maths::CSeasonalComponent, std::allocator<ml::maths::CSeasonalComponent> >&, std::vector<ml::maths::CTimeSeriesDecompositionDetail::CComponents::CComponentErrors, std::allocator<ml::maths::CTimeSeriesDecompositionDetail::CComponents::CComponentErrors> >&) const ()
   from /opt/elastic/7.0.0-20180511/elasticsearch-7.0.0-alpha1-SNAPSHOT/modules/x-pack/x-pack-ml/platform/linux-x86_64/bin/../lib/libMlMaths.so
#5  0x00007f9a41ba1534 in ml::maths::CTimeSeriesDecompositionDetail::CComponents::handle(ml::maths::CTimeSeriesDecompositionDetail::SDetectedSeasonal const&) ()
   from /opt/elastic/7.0.0-20180511/elasticsearch-7.0.0-alpha1-SNAPSHOT/modules/x-pack/x-pack-ml/platform/linux-x86_64/bin/../lib/libMlMaths.so
#6  0x00007f9a41b939b7 in ml::maths::CTimeSeriesDecompositionDetail::CPeriodicityTest::test(ml::maths::CTimeSeriesDecompositionDetail::SAddValue const&) ()
   from /opt/elastic/7.0.0-20180511/elasticsearch-7.0.0-alpha1-SNAPSHOT/modules/x-pack/x-pack-ml/platform/linux-x86_64/bin/../lib/libMlMaths.so
#7  0x00007f9a41b93bb5 in ml::maths::CTimeSeriesDecompositionDetail::CPeriodicityTest::handle(ml::maths::CTimeSeriesDecompositionDetail::SAddValue const&) ()
   from /opt/elastic/7.0.0-20180511/elasticsearch-7.0.0-alpha1-SNAPSHOT/modules/x-pack/x-pack-ml/platform/linux-x86_64/bin/../lib/libMlMaths.so
#8  0x00007f9a41b8666c in ml::maths::CTimeSeriesDecomposition::addPoint(long, double, boost::array<double, 4ul> const&) () from /opt/elastic/7.0.0-20180511/elasticsearch-7.0.0-alpha1-SNAPSHOT/modules/x-pack/x-pack-ml/platform/linux-x86_64/bin/../lib/libMlMaths.so
#9  0x00007f9a41bd0a13 in ml::maths::CUnivariateTimeSeriesModel::updateTrend(std::vector<ml::core::CTriple<long, ml::core::CSmallVector<double, 2ul>, unsigned long>, std::allocator<ml::core::CTriple<long, ml::core::CSmallVector<double, 2ul>, unsigned long> > > const&, std::vector<boost::array<ml::core::CSmallVector<double, 2ul>, 4ul>, std::allocator<boost::array<ml::core::CSmallVector<double, 2ul>, 4ul> > > const&) () from /opt/elastic/7.0.0-20180511/elasticsearch-7.0.0-alpha1-SNAPSHOT/modules/x-pack/x-pack-ml/platform/linux-x86_64/bin/../lib/libMlMaths.so
#10 0x00007f9a41bd0f5e in ml::maths::CUnivariateTimeSeriesModel::addSamples(ml::maths::CModelAddSamplesParams const&, std::vector<ml::core::CTriple<long, ml::core::CSmallVector<double, 2ul>, unsigned long>, std::allocator<ml::core::CTriple<long, ml::core::CSmallVector<double, 2ul>, unsigned long> > >) () from /opt/elastic/7.0.0-20180511/elasticsearch-7.0.0-alpha1-SNAPSHOT/modules/x-pack/x-pack-ml/platform/linux-x86_64/bin/../lib/libMlMaths.so
#11 0x00007f9a4138a295 in ml::model::CEventRateModel::sample(long, long, ml::model::CResourceMonitor&) () from /opt/elastic/7.0.0-20180511/elasticsearch-7.0.0-alpha1-SNAPSHOT/modules/x-pack/x-pack-ml/platform/linux-x86_64/bin/../lib/libMlModel.so
#12 0x00007f9a412d49c7 in ml::model::CAnomalyDetector::sample(long, long, ml::model::CResourceMonitor&) () from /opt/elastic/7.0.0-20180511/elasticsearch-7.0.0-alpha1-SNAPSHOT/modules/x-pack/x-pack-ml/platform/linux-x86_64/bin/../lib/libMlModel.so
#13 0x00007f9a412d831c in ml::model::CAnomalyDetector::buildResults(long, long, ml::model::CHierarchicalResults&) () from /opt/elastic/7.0.0-20180511/elasticsearch-7.0.0-alpha1-SNAPSHOT/modules/x-pack/x-pack-ml/platform/linux-x86_64/bin/../lib/libMlModel.so
#14 0x00007f9a40f2fce0 in ml::api::CAnomalyJob::outputResults(long) () from /opt/elastic/7.0.0-20180511/elasticsearch-7.0.0-alpha1-SNAPSHOT/modules/x-pack/x-pack-ml/platform/linux-x86_64/bin/../lib/libMlApi.so
#15 0x00007f9a40f30553 in ml::api::CAnomalyJob::outputBucketResultsUntil(long) () from /opt/elastic/7.0.0-20180511/elasticsearch-7.0.0-alpha1-SNAPSHOT/modules/x-pack/x-pack-ml/platform/linux-x86_64/bin/../lib/libMlApi.so
#16 0x00007f9a40f316a8 in ml::api::CAnomalyJob::handleRecord(boost::unordered::unordered_map<std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >, boost::hash<std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > >, std::equal_to<std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > >, std::allocator<std::pair<std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > > > > const&) () from /opt/elastic/7.0.0-20180511/elasticsearch-7.0.0-alpha1-SNAPSHOT/modules/x-pack/x-pack-ml/platform/linux-x86_64/bin/../lib/libMlApi.so
#17 0x00007f9a40fb6e30 in ml::api::CLengthEncodedInputParser::readStream(std::function<bool (boost::unordered::unordered_map<std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >, boost::hash<std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > >, std::equal_to<std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > >, std::allocator<std::pair<std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > > > > const&)> const&) () from /opt/elastic/7.0.0-20180511/elasticsearch-7.0.0-alpha1-SNAPSHOT/modules/x-pack/x-pack-ml/platform/linux-x86_64/bin/../lib/libMlApi.so
#18 0x00007f9a40f5d647 in ml::api::CCmdSkeleton::ioLoop() () from /opt/elastic/7.0.0-20180511/elasticsearch-7.0.0-alpha1-SNAPSHOT/modules/x-pack/x-pack-ml/platform/linux-x86_64/bin/../lib/libMlApi.so
#19 0x00005627315a37c9 in main ()

[hendrikmuhs]

Looking at the build hash, this seems to be a regression introduced with #92. I assume this test has been run before.

[tveasey]

I suspect that this is probably caused by numberOutliers in reweightOutliers being less than 0.5 for very sparse data. In which case, wrapping the code which does the reweighting in a if (numberOutliers > 1.0) would be the natural fix. (Note that this isn't released (nor targeted at 6.3) so also marking this as a non-issue.)

[hendrikmuhs]

@tveasey This is indeed the problem and can/should be fixed as you proposed.

But: I would like to further harden the code.