From owner-freebsd-hackers@freebsd.org Sun Nov 15 01:13:25 2015 Return-Path: Delivered-To: freebsd-hackers@mailman.ysv.freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:1900:2254:206a::19:1]) by mailman.ysv.freebsd.org (Postfix) with ESMTP id BC903A2ECE1 for ; Sun, 15 Nov 2015 01:13:25 +0000 (UTC) (envelope-from jmaloney@pcbsd.org) Received: from barracuda.ixsystems.com (mail.ixsystems.com [69.198.165.135]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (Client CN "*.ixsystems.com", Issuer "Go Daddy Secure Certificate Authority - G2" (verified OK)) by mx1.freebsd.org (Postfix) with ESMTPS id 93AE91387 for ; Sun, 15 Nov 2015 01:13:25 +0000 (UTC) (envelope-from jmaloney@pcbsd.org) X-ASG-Debug-ID: 1447550001-08ca040e840d810002-P5m3U7 Received: from [10.0.1.8] (ip72-209-175-134.ks.ks.cox.net [72.209.175.134]) by barracuda.ixsystems.com with ESMTP id rfFBIVD8xCTBCOHz (version=TLSv1 cipher=ECDHE-RSA-AES256-SHA bits=256 verify=NO); Sat, 14 Nov 2015 17:13:22 -0800 (PST) X-Barracuda-Envelope-From: jmaloney@pcbsd.org X-Barracuda-AUTH-User: jmaloney@pcbsd.org X-Barracuda-Apparent-Source-IP: 72.209.175.134 Mime-Version: 1.0 (Mac OS X Mail 9.1 \(3096.5\)) Subject: Re: FreeBSD smbfs horribly slow From: Joe Maloney X-ASG-Orig-Subj: Re: FreeBSD smbfs horribly slow In-Reply-To: <310B6D45-63E1-4012-BA35-8CB40CE9EBCF@pcbsd.org> Date: Sat, 14 Nov 2015 19:13:21 -0600 Cc: Mario Lobo , Mark Saad , Allan Jude , freebsd-hackers@freebsd.org Message-Id: <9044E3FC-BD0A-4F28-8AD3-98D743A99452@pcbsd.org> References: <20151113162548.61529137@Papi> <56463ACE.5020605@freebsd.org> <20151114115022.781c0bc1@Papi> <1312151904.87133792.1447539054621.JavaMail.zimbra@uoguelph.ca> <310B6D45-63E1-4012-BA35-8CB40CE9EBCF@pcbsd.org> To: Rick Macklem X-Mailer: Apple Mail (2.3096.5) X-Barracuda-Connect: ip72-209-175-134.ks.ks.cox.net[72.209.175.134] X-Barracuda-Start-Time: 1447550002 X-Barracuda-Encrypted: ECDHE-RSA-AES256-SHA X-Barracuda-URL: https://10.2.0.41:443/cgi-mod/mark.cgi X-Barracuda-BRTS-Status: 1 X-Virus-Scanned: by bsmtpd at ixsystems.com X-Barracuda-Spam-Score: 0.00 X-Barracuda-Spam-Status: No, SCORE=0.00 using global scores of TAG_LEVEL=1000.0 QUARANTINE_LEVEL=1000.0 KILL_LEVEL=8.0 tests=HTML_MESSAGE X-Barracuda-Spam-Report: Code version 3.2, rules version 3.2.3.24400 Rule breakdown below pts rule name description ---- ---------------------- -------------------------------------------------- 0.00 HTML_MESSAGE BODY: HTML included in message Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: quoted-printable X-Content-Filtered-By: Mailman/MimeDel 2.1.20 X-BeenThere: freebsd-hackers@freebsd.org X-Mailman-Version: 2.1.20 Precedence: list List-Id: Technical Discussions relating to FreeBSD List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sun, 15 Nov 2015 01:13:25 -0000 Just to clarify what I was intending to ask I am curious about the = state, and health of the current smbfs code in FreeBSD. That is why I = am trying to determine where it originated from, etc. If there might be = a newer version of it, or if it=E2=80=99s still the best way of mounting = a CIFS share on FreeBSD. Joe Maloney > On Nov 14, 2015, at 7:02 PM, Joe Maloney wrote: >=20 > I=E2=80=99ve noticed that with the freebsd version of mount_smbfs I am = not able to mount an airport disk. With FreeBSD I can use gvfs = available in ports to get around that issue. >=20 > https://github.com/freebsd/freebsd/tree/master/contrib/smbfs = > >=20 > However with the Mac OS X version of mount_smbfs I can mount an = airport disk. I realize this structure, and ASPL clobbering is pretty = gross to look at. Apologies in advance. I am curious if there is = anything useful to see here. >=20 > http://opensource.apple.com/source/smb/smb-759.40.1/ = = > >=20 > =46rom what I can tell it looks like the mount_smbfs tool originated = in FreeBSD, and was ported to other BSD=E2=80=99s? Or is FreeBSD using = this which has been abandoned by Linux? =20 >=20 > https://www.samba.org/samba/smbfs/ = > >=20 > Joe Maloney >=20 >> On Nov 14, 2015, at 4:10 PM, Rick Macklem > wrote: >>=20 >> Mario Lobo wrote: >>> On Fri, 13 Nov 2015 19:04:05 -0500 >>> Mark Saad wrote: >>>=20 >>>> Mario >>>> Can you share more about your setup . >>>> What filesystem is the samba share exported from ? >>>=20 >>> The shares tested were both from a FBSD (10.2-STABLE) samba4 and = Linux >>> (Centos) samba 3.6. >>>=20 >>>> What mount options >>>> on the filesystem level do you use ? >>>>=20 >>>=20 >>> smbfs rw,noatime,-N,-Iserverip 0 0 >>>=20 >>>> What version of samba , was it from ports or a package ? >>>>=20 >>>=20 >>> See above. >>>=20 >>>> On the samba level can you tell us about your config ? Have you = tried >>>> any of the tuning from https://calomel.org/samba_optimize.html >>>>=20 >>>=20 >>> Like I said, the problem is not with the server. >>>=20 >>>> Did you change any sysctls ? What did you set ? >>>>=20 >>>> Lastly what's the hardware like ; CPU, nic type , ram , etc >>>>=20 >>>=20 >>> I tried the same FBSD client on different hardware. Made no = difference. >>>=20 >> Did that different hardware have a different type of net interface = that >> uses a different net device driver? >>=20 >> I have no idea if smbfs can do the same thing, but both NFS and iSCSI >> can generate TCP TSO output segments of near 64K in data length and >> that can cause problems for some net device drivers. >> --> If the net interface has TSO enabled, try disabling it. >>=20 >> I never use smbfs, so I can't help more, rick >>=20 >>>=20 >>>=20 >>>> --- >>>> Mark Saad | nonesuch@longcount.org >>>>=20 >>>>> On Nov 13, 2015, at 6:13 PM, Mario Lobo wrote: >>>>>=20 >>>>> 2015-11-13 16:32 GMT-03:00 Allan Jude : >>>>>=20 >>>>>>> On 2015-11-13 14:25, Mario Lobo wrote: >>>>>>> Hi; >>>>>>>=20 >>>>>>> It seems no one in @questions had any info/pointers/interest on >>>>>>> this so I'm trying @hackers for some light. >>>>>>>=20 >>>>>>> Thanks, >>>>>>>=20 >>>>>>>=20 >>>>>>> Begin forwarded message: >>>>>>>=20 >>>>>>> Date: Wed, 11 Nov 2015 17:53:11 -0300 >>>>>>> From: Mario Lobo >>>>>>> To: freebsd-questions >>>>>>> Subject: FreeBSD smbfs horribly slow >>>>>>>=20 >>>>>>> Googling on this subject, I found: >>>>>> = http://lists.freebsd.org/pipermail/freebsd-questions/2005-September/098717= .html >>>>>>> = https://lists.freebsd.org/pipermail/freebsd-net/2013-January/034239.html >>>>>> = https://lists.freebsd.org/pipermail/freebsd-questions/2014-October/261804.= html >>>>>>>=20 >>>>>>> I am on 10.2-STABLE and using FreeBSD as a client to any amb = share >>>>>>> continues to be very slow. >>>>>>>=20 >>>>>>> The share is mounted through mount_smbfs. I tried smbnetfs = (fuse) >>>>>>> and it is just a tiny bit better but doesn't compare to other >>>>>>> clients (linux or win) when writing/reading files >>>>>>>=20 >>>>>>> It gets even worse if an application is doing operations with >>>>>>> variable size records inside a data file on the share. >>>>>>>=20 >>>>>>> Does anyone have any advice to improve this? >>>>>>>=20 >>>>>>> Thanks, >>>>>>>=20 >>>>>>>=20 >>>>>>>=20 >>>>>>> _______________________________________________ >>>>>>> freebsd-hackers@freebsd.org mailing list >>>>>>> https://lists.freebsd.org/mailman/listinfo/freebsd-hackers >>>>>>> To unsubscribe, send any mail to " >>>>>> freebsd-hackers-unsubscribe@freebsd.org" >>>>>>=20 >>>>>> What kind of operations are you doing? >>>>>>=20 >>>>>> I just mounted a share from my windows desktop on my FreeBSD >>>>>> -CURRENT machine, and was able to write new files at 64 >>>>>> megabytes/s (roughly 1/2 the available gigabit/sec) >>>>>>=20 >>>>>> Reading it back only got 50 megabytes/s, not sure why. >>>>>>=20 >>>>>> -- >>>>>> Allan Jude >>>>> Which one is the server? Windows or FBSD? >>>>>=20 >>>>> I have no problems with either one being the server. The problem = is >>>>> when FBSD is the client. >>>>>=20 >>>>> I wrote a daemon that executes operations on old DBF/NTX (clipper) >>>>> files (Yeah, I know ... but that's what they have for 20+ years = ..). >>>>>=20 >>>>> Anyway, a site interacts with this daemon via tcp, with commands = to >>>>> add/delete/update records/indexes, as well as finding keys on the >>>>> indexxes. >>>>>=20 >>>>> I prepared a test that has several of these routines together on a >>>>> 10.2-STABLE machine. >>>>>=20 >>>>> Enough to say that when executing the tests with the files stored >>>>> locally, the whole test takes 3-4 seconds to complete. >>>>>=20 >>>>> When doing the same test with the files on a share on the same = wire >>>>> (1G connection, no matter which OS runs the share), the test takes >>>>> around 3:50 minutes to complete! >>>>>=20 >>>>> I am preparing a Centos VM and compiling the deamon on it to check >>>>> the results. >>>>>=20 >>>>> -- >>>>> Mario Lobo >>>>> http://www.mallavoodoo.com.br >>>>> FreeBSD since version 2.2.8 [not Pro-Audio.... YET!!] (99,7% >>>>> winfoes FREE) _______________________________________________ >>>>> freebsd-hackers@freebsd.org mailing list >>>>> https://lists.freebsd.org/mailman/listinfo/freebsd-hackers >>>>> To unsubscribe, send any mail to >>>>> "freebsd-hackers-unsubscribe@freebsd.org" >>>=20 >>>=20 >>>=20 >>> -- >>> Mario Lobo >>> http://www.mallavoodoo.com.br >>> FreeBSD since 2.2.8 [not Pro-Audio.... YET!!] >>>=20 >>> "UNIX was not designed to stop you from doing stupid things, >>> because that would also stop you from doing clever things." >>> _______________________________________________ >>> freebsd-hackers@freebsd.org mailing list >>> https://lists.freebsd.org/mailman/listinfo/freebsd-hackers >>> To unsubscribe, send any mail to = "freebsd-hackers-unsubscribe@freebsd.org" >>>=20 >> _______________________________________________ >> freebsd-hackers@freebsd.org = > mailing list >> https://lists.freebsd.org/mailman/listinfo/freebsd-hackers = = > >> To unsubscribe, send any mail to = "freebsd-hackers-unsubscribe@freebsd.org = = >" >=20 > _______________________________________________ > freebsd-hackers@freebsd.org = mailing list > https://lists.freebsd.org/mailman/listinfo/freebsd-hackers = > To unsubscribe, send any mail to = "freebsd-hackers-unsubscribe@freebsd.org = "