Skip site navigation (1)Skip section navigation (2)
Date:      Wed, 16 Mar 2005 15:34:01 +0200
From:      Giorgos Keramidas <keramida@ceid.upatras.gr>
To:        "Colin J. Raven" <colin@kenmore.kozy-kabin.nl>
Cc:        freebsd-questions@freebsd.org
Subject:   Re: Illogical usage of swap
Message-ID:  <20050316133401.GA8571@orion.daedalusnetworks.priv>
In-Reply-To: <20050316141635.A2949@kenmore.kozy-kabin.nl>
References:  <20050316123901.U2949@kenmore.kozy-kabin.nl> <444qfbpydd.fsf@be-well.ilk.org> <20050316133153.M2949@kenmore.kozy-kabin.nl> <20050316125446.GA7986@orion.daedalusnetworks.priv> <20050316140047.S2949@kenmore.kozy-kabin.nl> <20050316130754.GC7986@orion.daedalusnetworks.priv> <20050316141635.A2949@kenmore.kozy-kabin.nl>

next in thread | previous in thread | raw e-mail | index | archive | help
On 2005-03-16 14:17, "Colin J. Raven" <colin@kenmore.kozy-kabin.nl> wrote:
>On Mar 16 at 15:07, Giorgos Keramidas said:
>>On 2005-03-16 14:02, "Colin J. Raven" <colin@kenmore.kozy-kabin.nl> wrote:
>>>On Mar 16 at 14:54, Giorgos Keramidas asked:
>>>>>Free memory (or the lack thereof) isn't the issue though.
>>>>>
>>>>>The issue is this:
>>>>>Swap: 8192M Total, 116K Used, 8192M Free
>>>>>and that's the piece of the puzzle that has us all utterly baffled.
>>>>>
>>>>>No way in creation this box should be swapping.
>>>>
>>>>Do you, by any chance, have vm.swap_idle_enabled set to 1?
>>>
>>>No, it isn't:
>>>gonzo# sysctl -a |grep vm.swap_idle_enabled
>>>vm.swap_idle_enabled: 0
>>
>>Strange.  AFAIK, processes are not swapped out to disk unless there is a
>>severe memory shortage.  The usual case is to have them "paged" out,
>>which is a bit different.
>
> That was my understanding also, and that's what's baffling us

Unless, of course, there _was_ a short period of memory shortage, some
processes were swapped out (a pair of idle getty instances, for example)
and they were never swapped back in because they are still idle.

You can probably track this down to the specific process or processes
that have the PS_INMEM bit turned off in their proc->p_sflag.  But 116K
seems too small to be a complete process.



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20050316133401.GA8571>