Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 30 Oct 2009 16:16:12 -0400 (EDT)
From:      Rick Macklem <rmacklem@uoguelph.ca>
To:        Olaf Seibert <O.Seibert@cs.ru.nl>
Cc:        dfr@freebsd.org, freebsd-current@freebsd.org, freebsd-stable@freebsd.org
Subject:   Re: 8.0-RC1 NFS client timeout issue
Message-ID:  <Pine.GSO.4.63.0910301607570.27313@muncher.cs.uoguelph.ca>
In-Reply-To: <20091029135239.GX841@twoquid.cs.ru.nl>
References:  <20091027164159.GU841@twoquid.cs.ru.nl> <Pine.GSO.4.63.0910281624440.18390@muncher.cs.uoguelph.ca> <20091029135239.GX841@twoquid.cs.ru.nl>

next in thread | previous in thread | raw e-mail | index | archive | help


On Thu, 29 Oct 2009, Olaf Seibert wrote:

>
> After writing, I realised that it is indeed perfectly allowed for the
> client to send data. But since the server already sent its FIN, it can't
> send anything more, not even an error message. So with that in mind, the
> client shouldn't send anything any more either.
>
Here is what I am seeing without and with the patch. The client is a
pretty recent FreeBSD-CURRENT (nfsv4-test) and the server Solaris10
(nfsv4-solaris). I don't get the 5 minute reconnect delay without
the patch. I think the reason is that the resets (Rst's) "inspire"
the FreeBSD-CURRENT TCP to do the new connection. (I don't remember
seeing any RSTs in your tcpdump?) I deleted a few irrelevant lines
(packets not between nfsv4-test and nfsv4-solaris), which is why
the packet #s aren't contiguous. (and appologies for the long lines)

Hopefully someone with TCP expertise will know if the RSTs are
done correctly and whether or not your server should be generating
them.

After the patch things look fine to me. Hopefully your (and others)
testing will go ok.

Snoop trace without the patch (what FreeBSD-CURRENT will do now):
   2 209.18617 nfsv4-solaris -> nfsv4-test.cis.uoguelph.ca TCP D=699 S=2049 Fin Ack=1066292217 Seq=1580479914 Len=0 Win=49232 Options=<nop,nop,tstamp 25243063 5673171>
   3 209.18645 nfsv4-test.cis.uoguelph.ca -> nfsv4-solaris TCP D=2049 S=699 Ack=1580479915 Seq=1066292217 Len=0 Win=16588 Options=<nop,nop,tstamp 6031630 25243063>
   4 209.18656 nfsv4-test.cis.uoguelph.ca -> nfsv4-solaris TCP D=2049 S=699 Rst Ack=1580479915 Seq=1066292217 Len=0 Win=0
   5 209.18662 nfsv4-test.cis.uoguelph.ca -> nfsv4-solaris TCP D=2049 S=699 Rst Ack=1580479915 Seq=1066292217 Len=0 Win=0
   7 528.97250 nfsv4-test.cis.uoguelph.ca -> nfsv4-solaris NFS C FSSTAT3 FH=9D01
   8 528.97261 nfsv4-solaris -> nfsv4-test.cis.uoguelph.ca TCP D=699 S=2049 Rst Seq=1580479915 Len=0 Win=0
   9 528.97311 nfsv4-test.cis.uoguelph.ca -> nfsv4-solaris TCP D=2049 S=818 Syn Seq=3293207355 Len=0 Win=65535 Options=<mss 1460,nop,wscale 3,sackOK,tstamp 6350137 0>
  10 528.97329 nfsv4-solaris -> nfsv4-test.cis.uoguelph.ca TCP D=818 S=2049 Syn Ack=3293207356 Seq=1757551152 Len=0 Win=49232 Options=<nop,nop,tstamp 25275042 6350137,mss 1460,nop,wscale 0,nop,nop,sackOK>
  11 528.97354 nfsv4-test.cis.uoguelph.ca -> nfsv4-solaris TCP D=2049 S=818 Ack=1757551153 Seq=3293207356 Len=0 Win=8326 Options=<nop,nop,tstamp 6350138 25275042>
  12 528.97375 nfsv4-test.cis.uoguelph.ca -> nfsv4-solaris NFS C FSSTAT3 FH=9D01
  13 528.97382 nfsv4-test.cis.uoguelph.ca -> nfsv4-solaris TCP D=2049 S=818 Rst Ack=1757551153 Seq=3293207356 Len=0 Win=0
  14 528.97439 nfsv4-solaris -> nfsv4-test.cis.uoguelph.ca TCP D=818 S=2049 Ack=3293207488 Seq=1757551153 Len=0 Win=49100 Options=<nop,nop,tstamp 25275042 6350138>
  15 528.97524 nfsv4-solaris -> nfsv4-test.cis.uoguelph.ca NFS R FSSTAT3 OK
  16 529.07565 nfsv4-test.cis.uoguelph.ca -> nfsv4-solaris TCP D=2049 S=818 Ack=1757551325 Seq=3293207488 Len=0 Win=16588 Options=<nop,nop,tstamp 6350240 25275042>

and what happens after the patch is applied:
   6   1.35481 nfsv4-test.cis.uoguelph.ca -> nfsv4-solaris NFS C FSSTAT3 FH=9D01
   7   1.35516 nfsv4-solaris -> nfsv4-test.cis.uoguelph.ca NFS R FSSTAT3 OK
   8   1.45564 nfsv4-test.cis.uoguelph.ca -> nfsv4-solaris TCP D=2049 S=651 Ack=2612339569 Seq=837212133 Len=0 Win=16588 Options=<nop,nop,tstamp 144535 25632953>
   9 361.34400 nfsv4-solaris -> nfsv4-test.cis.uoguelph.ca TCP D=651 S=2049 Fin Ack=837212133 Seq=2612339569 Len=0 Win=49232 Options=<nop,nop,tstamp 25668953 144535>
  10 361.34434 nfsv4-test.cis.uoguelph.ca -> nfsv4-solaris TCP D=2049 S=651 Ack=2612339570 Seq=837212133 Len=0 Win=16588 Options=<nop,nop,tstamp 502984 25668953>
  11 361.34441 nfsv4-test.cis.uoguelph.ca -> nfsv4-solaris TCP D=2049 S=651 Rst Ack=2612339570 Seq=837212133 Len=0 Win=0
  12 361.34447 nfsv4-test.cis.uoguelph.ca -> nfsv4-solaris TCP D=2049 S=651 Rst Ack=2612339570 Seq=837212133 Len=0 Win=0
  14 501.80966 nfsv4-test.cis.uoguelph.ca -> nfsv4-solaris TCP D=2049 S=881 Syn Seq=1926848102 Len=0 Win=65535 Options=<mss 1460,nop,wscale 3,sackOK,tstamp 642887 0>
  15 501.80979 nfsv4-solaris -> nfsv4-test.cis.uoguelph.ca TCP D=881 S=2049 Syn Ack=1926848103 Seq=2754679721 Len=0 Win=49232 Options=<nop,nop,tstamp 25682999 642887,mss 1460,nop,wscale 0,nop,nop,sackOK>
  16 501.81006 nfsv4-test.cis.uoguelph.ca -> nfsv4-solaris TCP D=2049 S=881 Ack=2754679722 Seq=1926848103 Len=0 Win=8326 Options=<nop,nop,tstamp 642888 25682999>
  17 501.81024 nfsv4-test.cis.uoguelph.ca -> nfsv4-solaris NFS C FSSTAT3 FH=9D01
  18 501.81089 nfsv4-solaris -> nfsv4-test.cis.uoguelph.ca TCP D=881 S=2049 Ack=1926848235 Seq=2754679722 Len=0 Win=49100 Options=<nop,nop,tstamp 25682999 642888>
  19 501.81169 nfsv4-solaris -> nfsv4-test.cis.uoguelph.ca NFS R FSSTAT3 OK
  20 501.91218 nfsv4-test.cis.uoguelph.ca -> nfsv4-solaris TCP D=2049 S=881 Ack=2754679894 Seq=1926848235 Len=0 Win=16588 Options=<nop,nop,tstamp 642990 25682999>

Anyone with TCP expertise have opinions on these? rick



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.GSO.4.63.0910301607570.27313>