[gentoo-amd64] What should the swap size be for 4G ram?

[gentoo-amd64] What should the swap size be for 4G ram?

[P.V.Anthony]

Hi,

Installing gentoo amd64 on a Supermicro PDSMi+.

It is using 4G ram and Intel Xeon x3220 with 4 cores.

Not sure how much of swap that should be allocated.

If following the old rule, 8G should be allocated for swap. I feel that 
is too much. Does 2.6 kernel really need so much of swap with 4G of ram?

Was thinking of just using a 1G swap file for safety. Please share some 
thoughts on the this swap size issue.

P.V.Anthony
-- 
gentoo-amd64@gentoo.org mailing list

Re: [gentoo-amd64] What should the swap size be for 4G ram?

[P.V.Anthony]

On this day, 22-September-2007 11:16 AM,  P.V.Anthony wrote:
> Hi,
> 
> Installing gentoo amd64 on a Supermicro PDSMi+.
> 
> It is using 4G ram and Intel Xeon x3220 with 4 cores.
> 
> Not sure how much of swap that should be allocated.
> 
> If following the old rule, 8G should be allocated for swap. I feel that 
> is too much. Does 2.6 kernel really need so much of swap with 4G of ram?
> 
> Was thinking of just using a 1G swap file for safety. Please share some 
> thoughts on the this swap size issue.
> 
> P.V.Anthony

Please ignore this email. It looks like I have asked something similar
to this before. I will read the old thread.

Sorry for the noise.

P.V.Anthony

-- 
gentoo-amd64@gentoo.org mailing list

Re: [gentoo-amd64] What should the swap size be for 4G ram?

[Richard Freeman]

[-- Attachment #1: Type: text/plain, Size: 1457 bytes --]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

P.V.Anthony wrote:
>> If following the old rule, 8G should be allocated for swap. I feel
>> that is too much. Does 2.6 kernel really need so much of swap with 4G
>> of ram?
>>
>> Was thinking of just using a 1G swap file for safety. Please share
>> some thoughts on the this swap size issue.
>>
> Please ignore this email. It looks like I have asked something similar
> to this before. I will read the old thread.
> 


That's Ok, I got a chuckle out of it.  You have Duncan who doesn't use
swap at all (I think), and you have me:

             total       used       free     shared    buffers     cached
Mem:       2058448    2041388      17060          0      82084     420860
- -/+ buffers/cache:    1538444     520004
Swap:     17514480    1152972   16361508

I guess there is a happy medium.  But what else am I going to do with
that odd space that doesn't fit easily into a RAID-5?  I figure that if
the kernel can find a use for it I might as well let it...  :)  I
probably have 50G more of unpartitioned space lying around since I've
installed my 2 RAID-5s on non-identical drives.  I guess I'll just have
to wait until ZFS takes off on linux...  :)
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.7 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFG9TGIG4/rWKZmVWkRAjLuAJ94LT0OzzgLBW6yDOgte0YJQowLTgCeK91I
Gi77TEE3GJllSyad0lr5ECQ=
=XF5V
-----END PGP SIGNATURE-----

[-- Attachment #2: S/MIME Cryptographic Signature --]
[-- Type: application/x-pkcs7-signature, Size: 4101 bytes --]

Re: [gentoo-amd64] What should the swap size be for 4G ram?

[Volker Armin Hemmann]

On Samstag, 22. September 2007, Richard Freeman wrote:
> P.V.Anthony wrote:
> >> If following the old rule, 8G should be allocated for swap. I feel
> >> that is too much. Does 2.6 kernel really need so much of swap with 4G
> >> of ram?
> >>
> >> Was thinking of just using a 1G swap file for safety. Please share
> >> some thoughts on the this swap size issue.
> >
> > Please ignore this email. It looks like I have asked something similar
> > to this before. I will read the old thread.
>
> That's Ok, I got a chuckle out of it.  You have Duncan who doesn't use
> swap at all (I think), and you have me:
>
>              total       used       free     shared    buffers     cached
> Mem:       2058448    2041388      17060          0      82084     420860
> -/+ buffers/cache:    1538444     520004
> Swap:     17514480    1152972   16361508
>
> I guess there is a happy medium.  But what else am I going to do with
> that odd space that doesn't fit easily into a RAID-5?  I figure that if
> the kernel can find a use for it I might as well let it...  :)  I
> probably have 50G more of unpartitioned space lying around since I've
> installed my 2 RAID-5s on non-identical drives.  I guess I'll just have
> to wait until ZFS takes off on linux...  :)

why? zfs is slow and is mixing things that should be in different layers. One 
argument against reiser4 always was 'it violates the layering' - well this is 
even more true for zfs.

And from this numbers:
http://tastic.brillig.org/~jwb/zfs-xfs-ext4.html

it doesn't look so great.
-- 
gentoo-amd64@gentoo.org mailing list

Re: [gentoo-amd64] What should the swap size be for 4G ram?

[B Vance]

On Sat, 2007-09-22 at 17:22 +0200, Volker Armin Hemmann wrote:
> On Samstag, 22. September 2007, Richard Freeman wrote:
> > P.V.Anthony wrote:
> > >> If following the old rule, 8G should be allocated for swap. I feel
> > >> that is too much. Does 2.6 kernel really need so much of swap with 4G
> > >> of ram?
> > >>
> > >> Was thinking of just using a 1G swap file for safety. Please share
> > >> some thoughts on the this swap size issue.
> > >
> > > Please ignore this email. It looks like I have asked something similar
> > > to this before. I will read the old thread.
> >
> > That's Ok, I got a chuckle out of it.  You have Duncan who doesn't use
> > swap at all (I think), and you have me:
> >
> >              total       used       free     shared    buffers     cached
> > Mem:       2058448    2041388      17060          0      82084     420860
> > -/+ buffers/cache:    1538444     520004
> > Swap:     17514480    1152972   16361508
> >
> > I guess there is a happy medium.  But what else am I going to do with
> > that odd space that doesn't fit easily into a RAID-5?  I figure that if
> > the kernel can find a use for it I might as well let it...  :)  I
> > probably have 50G more of unpartitioned space lying around since I've
> > installed my 2 RAID-5s on non-identical drives.  I guess I'll just have
> > to wait until ZFS takes off on linux...  :)
> 
> why? zfs is slow and is mixing things that should be in different layers. One 
> argument against reiser4 always was 'it violates the layering' - well this is 
> even more true for zfs.
> 
> And from this numbers:
> http://tastic.brillig.org/~jwb/zfs-xfs-ext4.html
> 
> it doesn't look so great.
Maybe it's just me, but I tend to take comparisons of apples and oranges
with a salt block.  To make that a useful comparison, the kernel
parameters would have to be shown as well as the level of support for
the various hardware.  The second is his write-up does not match his
numbers.  He writes ZFS "...has very bad sequential transfer with
hardware RAID and appalling sequential transfer with software RAID" but
fails to mention how equally bad EXT4 was at the same task.  XFS does
appear to blow them both out of the water in this area.  The rest of the
tests have ZFS being roughly equal to (or in some cases better then)
then the other two options.  

Maybe the next time, the test will be run using the userland ZFS on the
same system as the EXT4 and XFS system.  At least that way you can
adjust the numbers for the overhead of userland talking to the kernel.
Apples to Apples with a handicap is far more accurate then apples to
oranges.

I don't know enough about the layers you mention so I can't say whether
that's a real problem or not.



-- 
gentoo-amd64@gentoo.org mailing list

Re: [gentoo-amd64] What should the swap size be for 4G ram?

[Richard Freeman]

[-- Attachment #1: Type: text/plain, Size: 1659 bytes --]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Volker Armin Hemmann wrote:
> why? zfs is slow and is mixing things that should be in different layers. One 
> argument against reiser4 always was 'it violates the layering' - well this is 
> even more true for zfs.
> 

If you can suggest a filesystem that has full redundancy and uses all my
disk space optimally I'll be happy to use it.  :)  Ideally I'd prefer
not to have to carefully map and partition my drives and be able to add
space on a whim, but maybe that is asking too much.  But ZFS allows all
of this. RAID5 does not.

In theory ZFS also has much better write performance than RAID5 -
because it doesn't need to read stripe before writing them most of the time.

Sure, it is a layering violation, but there really isn't any way to
implement this sort of scheme using layers.  If you separate the raid
layer from the FS layer then the raid has no idea whether a given block
of disk is safe to overwrite or not - so it has to read stripes before
writing them.  So, while I do like the flexibility of running any
filesystem on top of any lvm/md scheme I'll live with the inflexibility
if it gives me more capability.

As far as performance goes - ZFS is pretty immature - I'm sure it will
only improve.  Especially considering it only has the most minimal
support for linux right now.  Obviously I'd wait until something more
robust is available...
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.7 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFG9U+YG4/rWKZmVWkRAhM/AKC7oL99LD6rVBYWZT8FKX+3TfLGfQCgkToP
JO0yxcMT1ptvRbhN35GxNvk=
=Srvy
-----END PGP SIGNATURE-----

[-- Attachment #2: S/MIME Cryptographic Signature --]
[-- Type: application/x-pkcs7-signature, Size: 4101 bytes --]

[gentoo-amd64] Re: What should the swap size be for 4G ram?

[Duncan]

Richard Freeman <rich@thefreemanclan.net> posted
46F53191.8040204@thefreemanclan.net, excerpted below, on  Sat, 22 Sep 2007
11:15:29 -0400:

> That's Ok, I got a chuckle out of it.  You have Duncan who doesn't use
> swap at all (I think),

I ran entirely without for awhile back when I had a gig of memory, but I 
turned it back on to allow suspend to disk (aka hibernate), then got the 
4-way RAID before I upped memory, and set it up for 4x4-gig of striped 
swap.  The kernel will stripe swap automatically when you set swap 
partitions to the same priority, as I've done here.  I didn't setup the 
swap on top of RAID-0.

I do use swap occasionally; the last time was when I recompiled qt-3 I 
think -- with unlimited jobs and $PORTAGE_TMPDIR on tmpfs!  IIRC it ran 
up 200+ load average (1 minute), used up all 8 gigs memory, and ran 7 
gigs into swap, before the system got slow enough it wasn't updating 
ksysguard any more and I couldn't tell how much worse it got. =8^)

I'm glad I was running striped swap or it probably would have been there 
quite some time, but it ran a nearly frozen GUI for several minutes, then 
started coming back to life, as I could see the load and memory usage 
climbing back down.

FWIW I'm running the kernel 2.6.23-rcs, which of course have the new CFS 
CPU scheduler.  I wasn't too happy with it originally, as it'd zero out X 
responsiveness much faster than the old scheduler did, but it has gotten 
somewhat better as the rcs have progressed.  I don't think it's quite 
where the old one was yet, but if one truly wants "fair", one should 
expect to play with priorities/niceness a bit more to keep smooth X 
operation when running 200+ load average and heavy swapping.  It's 
certainly reasonable now; something I couldn't have said back around rc-3 
or 4, when I was pretty unhappy with it.  I guess we see how good they 
did when it releases and we see if there's any outcry on people having 
trouble with X or whatever.

Back to swap, striped swap isn't so bad.  Running more than 4 gig of 
single-spindle swap can be a killer tho, as the effects of a multi-gig 
swap-storm on a slow single-spindle disk setup aren't pretty! =8^(

That said, the main reason I run swap now is for suspend-to-disk aka 
hibernate.  It's nice to be able to have the system shut off when I'm at 
work or sleeping, and turn it back on to have it restore the session as I 
left it.  Unfortunately, it can only use a single swap device for the 
suspend image, so since I have the 16 gigs spread across 4 disks equally, 
the max suspend image I can create is 4 gigs, only half of my physical 
memory size!  I'm thinking about layering swap on top of RAID-0 
(assembled based on the kernel command line, so before the restore 
starts) to see if I can avoid that and get a full 8 gig suspend image to 
match my memory size, plus faster read-in as well, but haven't tried it 
yet and am not positive it's supported.  IOW, the automatic combination 
of swap partitions into striped swap works for swap, but not for the 
suspend image, unfortunately, so (if it works) there's still reason to 
layer swap on RAID-0 after all.

-- 
Duncan - List replies preferred.   No HTML msgs.
"Every nonfree program has a lord, a master --
and if you use the program, he is your master."  Richard Stallman

-- 
gentoo-amd64@gentoo.org mailing list

Re: [gentoo-amd64] Re: What should the swap size be for 4G ram?

[Volker Armin Hemmann]

On Samstag, 22. September 2007, Duncan wrote:

> FWIW I'm running the kernel 2.6.23-rcs, which of course have the new CFS
> CPU scheduler.  I wasn't too happy with it originally, as it'd zero out X
> responsiveness much faster than the old scheduler did, but it has gotten
> somewhat better as the rcs have progressed.  I don't think it's quite
> where the old one was yet, but if one truly wants "fair", one should
> expect to play with priorities/niceness a bit more to keep smooth X
> operation when running 200+ load average and heavy swapping.  It's
> certainly reasonable now; something I couldn't have said back around rc-3
> or 4, when I was pretty unhappy with it.  I guess we see how good they
> did when it releases and we see if there's any outcry on people having
> trouble with X or whatever.

two things:

I am using cfs for a long time now - since shortly after the first patches. 
And everytime a new patch arrived I compared it with an unpatched kernel (at 
the moment 2.6.22.5) and everytime I had the same result:

X is much, much, much and a lot better with cfs. I never had such a snappy X. 
X rules with cfs. It is just great. Under load, without load, it does not 
matter. cfs is better. I am really curios what I did right that you do wrong.

second:
a little bit less text would be great. This was almost TL;DR. You turned a 
simple question about swapsize into a large sermon about your cool box with 
it bazillion of gb ram, its superduper harddisk setup and its amazing 
computing power that makes it survive a 200 load.

Back at school you would have failed the test because of missing the point.

Remember: this is an international list. Less text is almost always much more 
readable. Not everybody enjoys picking the few cherries of relevant 
information out of a sea of words.

Glück Auf
--
gentoo-amd64@gentoo.org mailing list