log file sync in Database KPI in EWA Report of ECC system

[Former Member]

Hi Folks,

In our Weekly EWA Report of ECC System (EHP6 SP08) , showing recommendation as log file sync observed value is 40 where as

reference value is 15.Could any one please suggest on how to further deep dive  .

Regards,

Sharma TVK

[stefan_koehler]

Hi Sharma,

are your end users really suffered by log file sync times (> 15) or is just because of that KPI in SAP Early Watch? If we talk about the latter, go over it and keep focusing on some important stuff. You may fix that possible unimportant stuff in addition, when focusing on the important ones.

I really wish that SAP would stop with that and develop some helpful statistic features (e.g. like percentile on STAD/ST03n aggregations) or business process related unique identifiers in statistic data sets instead.

Regards

Stefan

[ACE-SAP]

Hi

One little question, are you out of time to the point you cannot perform a little search on your own ?

This is not a very complicated point to investigate, even if you are a beginner...

At first check the instance parameter log_buffer and the redo log size.

You should also check the IO performance on the FS where you redo log are.

DBAs-Oracle.com: See How Easily you can Solve "Log File Sync" Wait Event ?

1627481 - Preemptive redolog switches in Oracle 11.2

• Set parameter "log_buffer" size & redolog file size in consideration of the number of DB CPU cores.
  Check parameter "cpu_count" for reference => "show parameter cpu_count")
  • For systems with <= 16 cores set the logbuffer to 4MB or don't set it at all.
  • For systems with more than 16 cores don't set the logbuffer at all.
  • If you have set the logbuffer to 4MB, then create redo logfiles of at least 200MB.This minimum configuration should only be used for DEV, Test or QA systems but not for production systems.
  • If you did not set the logbuffer, create redo logfiles of at least 1GB.
  • For systems with more than 64 cores or an SGA larger than 64 GB use at least 4GB redo logfiles.
  • If you use dataguard in Maximum availability or Maximum protection use always a logbuffer of 128 MB and redo logfiles of at least 4GB.

Regards

[Former Member]

Hi,

Thanks for the reply. We will check as per the above.

Regards,

Sharma TVK

[former_member207186]

Hi,

In most cases the increased "log file sync" times are related to I/O problems for the LGWR process. Please refer to point 15 of SAP note #619188 - "FAQ:Oracle wait events" and SAP note #793113 - "FAQ:Oracle I/O configuration" to check and optimize this.

Regards,

Bíborka