Re: MMAN blocker

Hello Nenad,
have you checked the involved semaphore(s) for/with OS PID 26758? Was it running on the default time-out or was there also another PID involved (ipcs -si)?

> How does a session makes a decision whether to allocate a chunk from the free memory or to wait on 'SGA: allocation forcing component growth'?

Never researched in detail but according to the documentation ( https://docs.oracle.com/database/121/REFRN/GUID-B826385B-CA43-4018-836B-F40BB1BA3F5E.htm#REFRN00623 ): "Process waiting on an immediate mode memory transfer with auto-tune SGA after a 4031 for MMAN to get the memory and post it." - so the wait only happens after session ran into ORA-4031.

Best Regards
Stefan Koehler

Independent Oracle performance consultant and researcher Website: http://www.soocs.de
Twitter: _at_OracleSK

> Noveljic Nenad <nenad.noveljic_at_vontobel.com> hat am 25. Juli 2019 um 12:33 geschrieben:
>
> According to the system dump the affected sessions were waiting on MMAN to finish a memory resizing operation:
>
> ...
>       7: MMAN ospid 26758 sid 71 ser 33258, waiting for 'rdbms ipc message'
>
> ...
>
>       72: CJQ0 ospid 29161 sid 721 ser 22160, waiting for 'SGA: allocation forcing component growth'
>
>           Blocked by inst: 1, sid: 71, ser: 33258
>           Final Blocker inst: 1, sid: 71, ser: 33258
> ...
>
> But MMAN seemed idle ('rdbms ipc message').
>
>
> I concluded following:
>
>
> 1. Due to the shared pool shortage MMAN started growing it.
> 2. At some point, there was enough memory to fulfill the requests, but they were failing, because the sessions were waiting on MMAN to finish the resizing.
>
>
> My questions are:
>
>
> 1. Are my conclusions correct?
> 2. How does a session makes a decision whether to allocate a chunk from the free memory or to wait on 'SGA: allocation forcing component growth'?
>
>
> Best regards,
>
>
> Nenad
>
> https://nenadnoveljic.com/blog/

--
http://www.freelists.org/webpage/oracle-l