Heap areas larger than 200 GB (history)

[Former Member]

Hi,

From the HANA mini checks. I see CHID 422 .  Heap areas larger than 200 GB (history)    <hostname> Pool/RowEngine/MonitorView (1081 GB)    none    X  .

Can anyone suggest how to get Pool/RowEngine/MonitorView  reduced to <200. I checked 1999997 but could not find procedure to reduce.

From cmd hdbcons ---> mm bl

mm bl -t Pool/RowEngine/MonitorView

742934976b (1658337 blocks) in use at: Dumping frame 0x00007efcd58ca590:

1: 0x00007efcd58ca590 in ptime::CSColumnMonitor::get_object()+0x0 at CSColumnMonitor.cc:695 (libhdbrskernel.so)

35041984b (382111 blocks) in use at: Dumping frame 0x00007efcd52eb190:

1: 0x00007efcd52eb190 in ltt::string_base<char, ltt::char_traits<char> >::enlarge_(unsigned long)+0xe0 at char_traits.hpp:208 (libhdbrskernel.so)

21231232b (88491 blocks) in use at: Dumping frame 0x00007efcd233cc20:

1: 0x00007efcd233cc20 in Diagnose::StatisticsBuffer::create(Diagnose::StatisticsBuffer::Callback&, unsigned long, Diagnose::StatisticsWrapper&, Diagnose::StatisticsData&, ltt::vector<bool> const&, bool, unsigned long)+0x9c0 at StatisticsBuffer.cpp:276 (libhdbbasis.so)

1632960b (8505 blocks) in use at: Dumping frame 0x00007efcd54e17d0:

1: 0x00007efcd54e17d0 in ptime::Catalog::add_monitor_view(ptime::Transaction&, ptime::CreateMonitorViewInfo&)+0x130 at catalog_monitor.cc:130 (libhdbrskernel.so)

288000b (1000 blocks) in use at: Dumping frame 0x00007efcd5760280:

1: 0x00007efcd5760280 in ptime::StatisticsMonitorHandle::createStatisticsMonitorHandles()+0x30 at memory.hpp:1397 (libhdbrskernel.so)

88704b (462 blocks) in use at: Dumping frame 0x00007efcd54e1730:

1: 0x00007efcd54e1730 in ptime::Catalog::add_monitor_view(ptime::Transaction&, ptime::CreateMonitorViewInfo&)+0x90 at catalog_monitor.cc:105 (libhdbrskernel.so)

73464b (462 blocks) in use at: Dumping frame 0x00007efcd551c2c0:

1: 0x00007efcd551c2c0 in ptime::MonitorViewInfo::createFieldInfoArray()+0x10 at monitorviewinfo.cc:27 (libhdbrskernel.so)

52224b (5 blocks) in use at: Dumping frame 0x00007efcd52eb19d:

1: 0x00007efcd52eb19d in ltt::string_base<char, ltt::char_traits<char> >::enlarge_(unsigned long)+0xe9 at memory.hpp:845 (libhdbrskernel.so)

1536b (12 blocks) in use at: Dumping frame 0x00007efcd25e9f30:

1: 0x00007efcd25e9f30 in MemoryManager::PoolAllocator::createSubAllocator(char const*, ltt::hana_component const&)+0x50 at handle_ref.hpp:479 (libhdbbasis.so)

288b (1 block) in use at: Dumping frame 0x00007efcd7c5ae30:

1: 0x00007efcd7c5ae30 in init_StreamingServicesHandle::init_StreamingServicesHandle()+0x10 at memory.hpp:1397 (libhdbrskernel.so)

Regards,

DK

[former_member183326 · Accepted solution]

What revision are you running?

The only way to release this memory is a restart of the indexserver.

There is a known issue to SAP that is fixed as of Revision 112.05. No work around is available (except for restarting the indexserver).

If your revision is higher than 112.05 then open an incident and send me the incident number.

This may of interest too 2343177 - Memory leak in Pool/RowEngine/MonitorView in scale-out system

[Former Member · Accepted solution]

Hi Michael,

HDB 102.06 is the version.

We have planned upgrade in near future - SP12. I hope it should clear then.

Regards,

DK

[former_member183326 · Accepted solution]

Yep, this is the only fix.

[former_member183326]

Please close off

Thanks

[former_member182967]

Hi Dinesh,

Check if note 2112732 - Pool/RowEngine/MonitorView allocates large amount of memory helps or not.

Regards,

Ning