Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 04 Jan 2013 19:10:21 +0100
From:      Johan Hendriks <joh.hendriks@gmail.com>
To:        FreeBSD <freebsd-current@freebsd.org>
Subject:   Re: ZFS/RAIDZ and SAMBA: abyssimal performance
Message-ID:  <50E71B0D.602@gmail.com>
In-Reply-To: <23BF8538-FB5A-4432-A4E1-721B5F566CA2@my.gd>
References:  <50E6DE91.7010404@zedat.fu-berlin.de> <1ADC2ECB-70FF-4DDD-9D62-16E2EEECDD8B@my.gd> <CAGH67wTeF8dxHFCF53mmAmZzbGf47SQy4yoH58L1gibbUs1tPA@mail.gmail.com> <50E6F2FC.3060903@zedat.fu-berlin.de> <23BF8538-FB5A-4432-A4E1-721B5F566CA2@my.gd>

next in thread | previous in thread | raw e-mail | index | archive | help
Fleuriot Damien schreef:
> On Jan 4, 2013, at 4:19 PM, O. Hartmann <ohartman@zedat.fu-berlin.de> wrote:
>
>> Am 01/04/13 15:45, schrieb Garrett Cooper:
>>> On Fri, Jan 4, 2013 at 6:06 AM, Fleuriot Damien <ml@my.gd> wrote:
>>>
>>> ...
>>>
>>>> And this is under [global] in /usr/local/etc/smb.conf:
>>>>    min receivefile size = 16384
>>>>    aio read size = 16384
>>>>    aio write size = 16384
>>>>    aio write behind = yes
>>> These are still pretty low, depending on what your networking/disk
>>> setup is like; my important performance settings are:
>>>
>>>         socket options = SO_RCVBUF=64240 SO_SNDBUF=64240 TCP_NODELAY
>>> IPTOS_LOWDELAY IPTOS_THROUGHPUT
>>>         write cache size = 65536
>>>         aio read size = 65536
>>>         aio write size = 65536
>>>         directory name cache size = 0
>>>
>>> HTH,
>>> -Garrett
>> Well, now I have peak values ~ 120 MB/s when copying. I applied Fleuriot
>> Damien's values to /boot/loader.conf and yours to the smb.conf.
>> Somewhere in the handbook this should be documented! it is to much
>> efford to get SAMBA working properly with ZFS, if the tricks and
>> problems are so widespread over several architectural aspects of the system.
>>
>> It could save a lot of time for adminsitartors and those which try
>> FreeBSD as a serving system instead of Linux.
>>
>> Just for the record. I feel a bit confused about all the tricks and
>> tweak now "published" for ZFS, its magic L2ARC, the kernel_vmem wizzardy
>> thingis. The ZFS Wiki seems to be a bit outdated and confusing, it would
>> be a great deal if all these things could be lined up a s a primer with
>> a bit more explanations than "put this number there".
>>
>> And by the way, it is like changing from hell to heaven having now ~ 100
>> MB/s throughput compared to ~1/500!
>>
>> Thanks a lot,
>> Oliver
>>
>
> The problem, Oliver, is that these values are system dependant.
>
> Notice how Garret replied that these values are a bit low (and they might be indeed !).
>
> However, while you have 16gb RAM, my ZFS NAS only has 4gb.
>
>
>
> Basically and as Jeremy Chadwick pointed out at the time, there is no one set of correct values for 100% of the population.
>
> One has to adjust them step by step and decide what is best for them.
>
>
>
> @garret: I'll try with the values you posted, although I get 90-120mbytes/s most of the time so I pretty much saturate my 1gbs link.
>
> _______________________________________________
> freebsd-current@freebsd.org mailing list
> http://lists.freebsd.org/mailman/listinfo/freebsd-current
> To unsubscribe, send any mail to "freebsd-current-unsubscribe@freebsd.org"
The only zfs tunable i use in my /boot/loader.conf is vfs.zfs.arc_max=
I  leave 4 GB for the system itself in most cases.

In my smb.conf i use the following
socket options = TCP_NODELAY SO_RCVBUF=131072 SO_SNDBUF=131072
max protocol = SMB2
But if you set that then you can not use the aio settings as samba will 
crash then.

if you use aio also make sure you load the aio module by setting the 
following in the /boot/loader.conf
aio_load="YES"

@Oliver
I see you use one device for ZIL/log and L2ARC/cache.
Be aware that you can lose the L2ARC/cache without data corruption, but 
losing the ZIL/log device might cause corrupted data.
So always create a mirror for the ZIL/log device.

gr
Johan Hendriks










Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?50E71B0D.602>