Re: Question about hugepages, shared memory, and /dev/shm....

From: Mark Bobak <Mark.Bobak_at_proquest.com>
Date: Wed, 16 Apr 2014 17:48:49 +0000
Message-ID: <CF743AA9.57F21%Mark.Bobak_at_ProQuest.com>



Thanks Amir. Yep, confirmed. LS Cheng just wrote the same thing a few minutes ago.

-Mark

From: <Hameed>, Amir Hameed <Amir.Hameed_at_xerox.com<mailto:Amir.Hameed_at_xerox.com>> Date: Wednesday, April 16, 2014 at 1:46 PM To: Mark Bobak <Mark.Bobak_at_ProQuest.com<mailto:Mark.Bobak_at_ProQuest.com>>, "Cunningham, Mike" <mcunningham_at_thedoctors.com<mailto:mcunningham_at_thedoctors.com>>, "Chris.Ruel_at_lfg.com<mailto:Chris.Ruel_at_lfg.com>" <Chris.Ruel_at_lfg.com<mailto:Chris.Ruel_at_lfg.com>>, "oracle-l_at_freelists.org<mailto:oracle-l_at_freelists.org>" <oracle-l_at_freelists.org<mailto:oracle-l_at_freelists.org>> Subject: RE: Question about hugepages, shared memory, and /dev/shm....

Mark,
I am running EBS and Oracle FMW SOA Suite 11g on two different Linux x86_64 servers. On both servers, I do see JOXSHM_EXT_.. files. These are not related to AMM and are most likely there because the DB was shutdown either via immediate or abort option (/dev/shm Filled Up With Files In Format JOXSHM_EXT_xxx_SID_xxx (Doc ID 752899.1). The ones associated AMM should have names like ora_<SID>_<SHMID>_nn.

From: oracle-l-bounce_at_freelists.org<mailto:oracle-l-bounce_at_freelists.org> [mailto:oracle-l-bounce_at_freelists.org] On Behalf Of Mark Bobak Sent: Wednesday, April 16, 2014 1:20 PM
To: Cunningham, Mike; Chris.Ruel_at_lfg.com<mailto:Chris.Ruel_at_lfg.com>; oracle-l_at_freelists.org<mailto:oracle-l_at_freelists.org> Subject: Re: Question about hugepages, shared memory, and /dev/shm....

Hi Mike,

AMM (Automatic Memory Management), which is setting memory_target and letting Oracle manage both SGA and PGA from that chunk of memory, is only available in 11g, and *not* compatible with hugepages. ASMM (Automatic Shared Memory Mangement), which is setting sga_target and pga_aggregate_target, is available in 10g, and is compatible with hugepages.

My understanding was that /dev/shm segments and hugepages were mutually exclusiveÖbut perhaps not?

Anyhow, Iím seeing ASMM + hugepages + segments in /dev/shm, which I thought wasnít possible.

-Mark

From: <Cunningham>, Mike <mcunningham_at_thedoctors.com<mailto:mcunningham_at_thedoctors.com>> Date: Wednesday, April 16, 2014 at 1:13 PM To: "Chris.Ruel_at_lfg.com<mailto:Chris.Ruel_at_lfg.com>" <Chris.Ruel_at_lfg.com<mailto:Chris.Ruel_at_lfg.com>>, Mark Bobak <Mark.Bobak_at_ProQuest.com<mailto:Mark.Bobak_at_ProQuest.com>>, "oracle-l_at_freelists.org<mailto:oracle-l_at_freelists.org>" <oracle-l_at_freelists.org<mailto:oracle-l_at_freelists.org>> Subject: RE: Question about hugepages, shared memory, and /dev/shm....

For what itís worth. While using AMM with huge pages the instance in my environment crashed on occasion during memory resizing. That was on version 10.2.0.3 with Linux 5.7 (x86_64).

I did not pay attention to the /dev/shm so I canít offer anything there. Also, learning from my past errors, I have never tried AMM in 11g with huge pages.

Michael Cunningham
Senior Database Administrator
The Doctors' Company
707.226.0221 - desk
707.337.0184 - cell

From:oracle-l-bounce_at_freelists.org<mailto:oracle-l-bounce_at_freelists.org> [mailto:oracle-l-bounce_at_freelists.org] On Behalf Of Ruel, Chris Sent: Wednesday, April 16, 2014 8:59 AM
To: Mark.Bobak_at_proquest.com<mailto:Mark.Bobak_at_proquest.com>; oracle-l_at_freelists.org<mailto:oracle-l_at_freelists.org> Subject: RE: Question about hugepages, shared memory, and /dev/shm....

I donít know if you have come across this MOS doc yet but itís pretty good at explaining some things relating to HPís:

Oracle Support Document 361323.1 (HugePages on Linux: What It Is... and What It Is Not...) can be found at: https://support.oracle.com/epmos/faces/DocumentDisplay?id=361323.1 I would also add that depending on the version of Linux, you need to disable Transparent Huge pages for OEL6...at least we did. Look at the referenced documents at the bottom and there is an article on this. If you donít disable THPís, it can cause problems in RAC environments.

For your questions below, your understanding correct for #1 and #2. I *think* you are correct on #3 and #4 but I have not dove in that far myself...have left it up to sysadmins to make sure it works...

Chris..

Chris Ruel * Oracle Database Administrator cruel_at_lfg.com<mailto:cruel_at_lfg.com> * Desk:317.759.2172 * Cell 317.523.8482

From:oracle-l-bounce_at_freelists.org<mailto:oracle-l-bounce_at_freelists.org> [mailto:oracle-l-bounce_at_freelists.org] On Behalf Of Mark Bobak Sent: Wednesday, April 16, 2014 11:47 AM To: oracle-l_at_freelists.org<mailto:oracle-l_at_freelists.org> Subject: Question about hugepages, shared memory, and /dev/shm....

Hi All,

So, I thought I really understood this stuff, but Iím a little baffled here, and I wonder if anyone can offer me a clue?

Hereís what I (think I) know:

1.)  AMM (setting memory_target) is *not* compatible with a hugepages configuration.  Any attempt to use hugepages will lock out the memory allocated to hugepages and AMM will only use non-hugepage memory allocations, the effect of which would be like removing the huge page allocated memory from the system.
2.)  ASMM (setting sga_target and pga_aggregate_target) and MMM (manually setting db_cache_size and pool sizes) *are* compatible with a hugepages configuration, and for any non-trivially sized SGA, hugepages is strongly recommended.
3.)  If hugepages are *not* configured, and AMM is used, memory segments will be mapped in /dev/shm.
4.)  If hugepages *are* used, no memory segments will be visible in /dev/shm.

So, thatís what I think is true about memory configuration and hugepages configuration.

That seems to be consistent throughout our environment, which mostly has ASMM or MMM and hugepages configuration.

However, and this is where my confusion comes in, we have several eBS environments, which seem to have a valid and active hugepages configuration, are using ASMM (not AMM), and *still* I can see memory segments allocated in /dev/shm?? Any idea how this is possible?

Hereís an example from our preprod environment: (Content was too long for Oracle-L, so hereís a paste bin URL)

http://pastebin.com/7w2V2jEa

So, Iím a little baffled here. I thought these were mutually exclusive features.

Note also that the timestamps on the /dev/shm segments is *after* instance startup time, so, I donít think these are ďorphanĒ memory segmentsÖ.

Anyone out there can clue me in?

Thanks,

-Mark

Notice of Confidentiality: **This E-mail and any of its attachments may contain Lincoln National Corporation proprietary information, which is privileged, confidential, or subject to copyright belonging to the Lincoln National Corporation family of companies. This E-mail is intended solely for the use of the individual or entity to which it is addressed. If you are not the intended recipient of this E-mail, you are hereby notified that any dissemination, distribution, copying, or action taken in relation to the contents of and attachments to this E-mail is strictly prohibited and may be unlawful. If you have received this E-mail in error, please notify the sender immediately and permanently delete the original and any copy of this E-mail and any printout. Thank You.**

Confidentiality Notice: This message and any attachments hereto may contain confidential and privileged communications or information and/or attorney client communications or work-product protected by law. The information contained herein is transmitted for the sole use of the intended recipient(s). If you are not the intended recipient or designated agent of the recipient of such information, you are hereby notified that any use, dissemination, copying or retention of this e-mail or the information contained herein is strictly prohibited and may subject you to penalties under federal and/or state law. If you received this e-mail in error, please notify the sender immediately and permanently delete this e-mail.

--

http://www.freelists.org/webpage/oracle-l Received on Wed Apr 16 2014 - 19:48:49 CEST

Original text of this message