Memory issue in Hana Box

[former_member356313]

Dear Team,

We are facing memory issue in our hana box.

On this Box, we have 2 devlopment and 2 quality instances. Along with those we have 2 PRD DR instances. The whole box is of 757GB RAM and 24GB SWAP. Out of which we have limited the ram usage on this different instances using global_allocation_limit paramter as below

           Allocated        Peak used memory  

DEV1 - 125G             75G

QAS1 - 275G             200G  

DEV2 - 125G             75G  

QAS2 - 175G             110G  

LIL PRD DR - 349.95G

LDK PRD DR - 230G

When the dev and quality instances are running, there is no problem of memory. But when once the DR of both the prd instances are setted (configured) the swap is also being utilized with free RAM going to 1 GB.

After some time when less load on dev or quality is happening, the RAM might get release but the swap remains the same 24GB used.

DR is configured with Delta shipping and ASYN mode

HANA version is HANA 1.0 SPS110

Can we reduced the already allocated memory if needed in that development and quality instances.

Kinldy suggest solution.

[former_member182967]

Hi Vikram,

Please set parameter global.ini/[system_replication]-> preload_column_tables=false on the secondary to switch off preload of tables.

Additionally, set global.ini/[memorymanager]-> global_allocation_limit to a minimum value of 64 GB or “row store size+ 20 GB on the secondary.

For more information, please refer to -> 4.3 Use secondary site for DEV/QA System

Regards,

Ning

[former_member356313]

Hi Ning,

Thanks for your replay..

As of now secondary one DR is configured, and its not possible to set any parameters through studio.

Then how to set them.

Also my actually primary prd instances size in 350 GB and 230 GB, is it OK to set the global allocation limit parameter to 64GB for both ....if so where to set from OS level.

Further where to find the row store size...?

Also let me know can we reduced the already allocated memory if needed in that development and quality instances.

Regards,

Vikram

[former_member182967]

Hi Vikram,

On secondary, directly edit the .ini file and adjust parameter.

For parameter locations, please refer to note 2186744 – FAQ: SAP HANA Parameters -> 3. Where can the configured SAP HANA parameters be found?.

In order to make sure that changes to online parameters are taken into account, you can force a reconfiguration using the following command:

hdbnsutil -reconfig

The row store size can be determined with this SQL statement which is already mentioned in the above document:

select host,round(sum(page_size*USED_BLOCK_COUNT)/1024/1024/1024,2) as "RowStore Size GB" from m_data_volume_page_statistics where page_sizeclass = '16k-RowStore' group by host;

Regards,

Ning

[former_member356313]

HI Ning,

Actually primary prd instances size in 350 GB and 230 GB, is it OK to set the global allocation limit parameter to 64GB for both secoundry PRD instances. (lesser value)

Also let me know can we reduced the already allocated memory if needed in that development and quality instances as the peak usage is less...

         Allocated        Peak used memory   

DEV1 - 125G             75G

QAS1 - 275G             200G   

DEV2 - 125G             75G   

QAS2 - 175G             110G   

Regards,

Vikram

[former_member182967]

Hi Vikram,

For reducing allocation limit for DEV/QA, please set global.ini/[memorymanager]-> global_allocation_limit for each database and then restart.

Regards,

Ning

[former_member356313]

Hi Ning,

Thanks for your information.

Let me know, if we takeover from primary to secoundry is it necessary to revert the preload_column_tables parameter from false to true.

Further we have builded another DR that is 2nd one (FAR). Is it same configuration as between primary and secoundry or any new settings needed.

Regards,

Vikram

[former_member182967]

Hi Vikarm,

In my understanding, you do not need to change.

(Notes: DEV/QA systems need to be shut down in case of a takeover.)

Sorry, about your other DR, since I do not know your configuration, I'm not sure.

Regards,

Ning

[former_member747409]

This message was moderated.