cancel
Showing results for 
Search instead for 
Did you mean: 

Out of Memory Issue After SP10 Upgrade

muthuram_shanmugavel2
Contributor
0 Kudos

Hi Experts,

We recently upgraded our Production system from SP07 to SP10.

Our production system is 4 node system (512 GB each) - 1 Master, 2 Slave and 1 Standby.

Master node stores 35 GB of data and two Slave nodes stores 391 GB and 383 GB of data respectively.

In SP07, system was running fine without any issues.

But after SP10 upgrade, We are getting continuous out of memory issues.

We analysed and find out heap memory leakage happened on the following categories, which is increased drastically

  1. “Pool/ColumnStoreTables/Main/Dictionary/RoDict”
  2. “Pool/ColumnStoreTables/Main/Dictionary/ValueDict”
  3. “Pool/ColumnStoreTables/Main/Compressed/Indirect”
  4. “Pool/ColumnStoreTables/Main/Uncompressed”
  5. “Pool/ColumnStoreTables/Main/Index/Single”
  6. “Pool/ColumnStoreTables/Main/Rowid”

Select HOST, (SUM("EXCLUSIVE_SIZE_IN_USE") / (1024*1024*1024)) As "HEAP_MEMORY_LEAGAGE_GB" from "_SYS_STATISTICS"."HOST_HEAP_ALLOCATORS_BASE"

WHERE category in ('Pool/ColumnStoreTables/Main/Dictionary/RoDict','Pool/ColumnStoreTables/Main/Dictionary/ValueDict',

'Pool/ColumnStoreTables/Main/Compressed/Indirect','Pool/ColumnStoreTables/Main/Uncompressed',

'Pool/ColumnStoreTables/Main/Index/Single','Pool/ColumnStoreTables/Main/Rowid')

AND SERVER_TIMESTAMP = '2016-08-16 01:02:47.793'

GROUP BY HOST

Order By Host

We noticed master node occupies 30 GB and

slave nodes occupies 332 GB and 298 GB of data respectively.

But this was not the case in SP07. It was running fine.

Attached snapshot before upgrade.

There is no heap memory leakage in “Pool/ColumnStoreTables/Main/*" categories.

As it is production system, our complete business operations are getting affected.

We need your advise How can we resolve this issue and get back the system into running stage.

Thanks & Regards,

Muthuram

Accepted Solutions (0)

Answers (1)

Answers (1)

Former Member
0 Kudos

Hi Muthuram,

Please let us know your HANA revision version for SP10, seems to be your hitting a Buggy release.Its recommend to update your revision to 105 ie SP10 Patch 5 which is stable version of HANA with respect to SP10 and covers fix for Fixed an out of memory issue when a prepared "like" filter was used.

2275609 - SAP HANA SPS 10 Database Maintenance Revision 102.05


As known only work around aviable for this OOM is you need to restart HANA instance.


So hope you update your revison to 105


Note: Mark this as helpful or correct answer if this fix your issue


Thanks,

Avinash