Received: from yw-out-1718.google.com (yw-out-1718.google.com [74.125.46.153]) by maillists.intel-research.net (8.13.8/8.13.8) with ESMTP id n3RG6Fbs009352 for ; Mon, 27 Apr 2009 09:06:16 -0700 Received: by yw-out-1718.google.com with SMTP id 9so1324218ywk.34 for ; Mon, 27 Apr 2009 08:49:49 -0700 (PDT) MIME-Version: 1.0 Sender: demmer@gmail.com Received: by 10.101.66.15 with SMTP id t15mr8839038ank.38.1240847389543; Mon, 27 Apr 2009 08:49:49 -0700 (PDT) In-Reply-To: References: From: Michael Demmer Date: Mon, 27 Apr 2009 08:49:29 -0700 X-Google-Sender-Auth: c31c365ea689fd42 Message-ID: <59e8babf0904270849h17d166d6wde1856a6fcaa7b55@mail.gmail.com> To: Teemu Content-Type: multipart/alternative; boundary=00163662e65cdfcc1d04688b4b93 Cc: dtn-users@maillists.intel-research.net Subject: Re: [dtn-users] TCPCL keepalive interval X-BeenThere: dtn-users@maillists.intel-research.net X-Mailman-Version: 2.1.9 Precedence: list List-Id: dtn-users mailing list List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Mon, 27 Apr 2009 16:06:17 -0000 --00163662e65cdfcc1d04688b4b93 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit I'm too far removed from the code to know why you're having the specific problem, but one reason why you might be seeing it intermittently is that there's a race between the two hosts to see which one establishes the TCP connection first, if you've configured a link on both sides. Other than that my guess is that this is a simple bug in the logic for setting up the various timers in the connection-oriented convergence layer and that the case in which one side wants keepalives but the other doesn't was simply not very well tested. -m On Mon, Apr 27, 2009 at 4:14 AM, Teemu wrote: > Hi, > > We're running into a randomly occurring problem where the TCPCL of DTN2 is > requesting a keepalive interval of 0 (i.e., to disable keepalives) in the > contact header. If we then do not send keepalives as requested, DTN2 will > timeout the connection due to inactivity after 2x the timeout we requested. > > According to draft-irtf-dtnrg-tcp-clayer-02: > > The keepalive_interval parameter is set to the minimum value > from both contact headers. If one or both contact headers > contains the value zero, then the keepalive feature (described > in Section 5.5) is disabled. > > However, it seems like DTN2 is expecting keepalives at the interval we > request in the contact header instead of disabling the feature as per the > above spec. Also, DTN2 will not send any keepalives in this case. > > This issue seems to occur and disappear unpredictably with both 2.6 release > and the current repository version. Restarting the dtnd instance or > changing > to a different version sometimes fixes this, but not always. > > Any insights on what might be going on? > > Cheers, > -Teemu > > > _______________________________________________ > dtn-users mailing list > dtn-users@maillists.intel-research.net > http://maillists.intel-research.net/mailman/listinfo/dtn-users > --00163662e65cdfcc1d04688b4b93 Content-Type: text/html; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable I'm too far removed from the code to know why you're having the spe= cific problem, but one reason why you might be seeing it intermittently is = that there's a race between the two hosts to see which one establishes = the TCP connection first, if you've configured a link on both sides.
Other than that my guess is that this is a simple bug in the logic for = setting up the various timers in the connection-oriented convergence layer = and that the case in which one side wants keepalives but the other doesn= 9;t was simply not very well tested.

-m

On Mon, Apr 27, 2009 at 4:14 AM, T= eemu <teemuk@n= etlab.tkk.fi> wrote:
Hi,

We're running into a randomly occurring problem where the TCPCL of DTN2= is
requesting a keepalive interval of 0 (i.e., to disable keepalives) in the contact header. If we then do not send keepalives as requested, DTN2 will timeout the connection due to inactivity after 2x the timeout we requested.=

According to draft-irtf-dtnrg-tcp-clayer-02:

=A0 =A0 =A0 =A0The keepalive_interval parameter is set to the minimum valu= e
=A0 =A0 =A0 =A0from both contact headers. =A0If one or both contact header= s
=A0 =A0 =A0 =A0contains the value zero, then the keepalive feature (descri= bed
=A0 =A0 =A0 =A0in Section 5.5) is disabled.

However, it seems like DTN2 is expecting keepalives at the interval we
request in the contact header instead of disabling the feature as per the above spec. Also, DTN2 will not send any keepalives in this case.

This issue seems to occur and disappear unpredictably with both 2.6 release=
and the current repository version. Restarting the dtnd instance or changin= g
to a different version sometimes fixes this, but not always.

Any insights on what might be going on?

Cheers,
-Teemu


_______________________________________________
dtn-users mailing list
dtn-users@maillis= ts.intel-research.net
http://maillists.intel-research.net/mailman/listinfo/dtn-= users

--00163662e65cdfcc1d04688b4b93-- Received: from smtp.netlab.hut.fi (keskus.netlab.hut.fi [130.233.154.176]) by maillists.intel-research.net (8.13.8/8.13.8) with ESMTP id n3RBXViH029289 for ; Mon, 27 Apr 2009 04:33:32 -0700 Received: from mailsrv.netlab.hut.fi (mailsrv.netlab.hut.fi [130.233.154.190]) by smtp.netlab.hut.fi (Postfix) with ESMTP id 53A45C5101 for ; Mon, 27 Apr 2009 14:17:10 +0300 (EET DST) Received: from [192.168.2.153] (pc94.netlab.hut.fi [130.233.154.94]) by mailsrv.netlab.hut.fi (Postfix) with ESMTP id 71044120050 for ; Mon, 27 Apr 2009 14:17:05 +0300 (EEST) User-Agent: Microsoft-Entourage/12.15.0.081119 Date: Mon, 27 Apr 2009 14:14:32 +0300 From: Teemu To: Message-ID: Thread-Topic: TCPCL keepalive interval Thread-Index: AcnHKVbDQANUDq7ykUmYzb+BKjI0jA== Mime-version: 1.0 Content-type: text/plain; charset="US-ASCII" Content-transfer-encoding: 7bit Subject: [dtn-users] TCPCL keepalive interval X-BeenThere: dtn-users@maillists.intel-research.net X-Mailman-Version: 2.1.9 Precedence: list List-Id: dtn-users mailing list List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Mon, 27 Apr 2009 11:33:33 -0000 Hi, We're running into a randomly occurring problem where the TCPCL of DTN2 is requesting a keepalive interval of 0 (i.e., to disable keepalives) in the contact header. If we then do not send keepalives as requested, DTN2 will timeout the connection due to inactivity after 2x the timeout we requested. According to draft-irtf-dtnrg-tcp-clayer-02: The keepalive_interval parameter is set to the minimum value from both contact headers. If one or both contact headers contains the value zero, then the keepalive feature (described in Section 5.5) is disabled. However, it seems like DTN2 is expecting keepalives at the interval we request in the contact header instead of disabling the feature as per the above spec. Also, DTN2 will not send any keepalives in this case. This issue seems to occur and disappear unpredictably with both 2.6 release and the current repository version. Restarting the dtnd instance or changing to a different version sometimes fixes this, but not always. Any insights on what might be going on? Cheers, -Teemu Received: from cs.tcd.ie (relay.cs.tcd.ie [134.226.32.56]) by maillists.intel-research.net (8.13.8/8.13.8) with ESMTP id n3MDBTJn000831; Wed, 22 Apr 2009 06:11:30 -0700 Received: from localhost (localhost [127.0.0.1]) by relay.cs.tcd.ie (Postfix) with ESMTP id A024954141; Wed, 22 Apr 2009 13:57:12 +0100 (IST) X-Virus-Scanned: amavisd-new at cs.tcd.ie Received: from cs.tcd.ie ([127.0.0.1]) by localhost (smtp.cs.tcd.ie [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id N365q6exnvCZ; Wed, 22 Apr 2009 13:57:11 +0100 (IST) Received: from [134.226.63.120] (cswireless63-120.cs.tcd.ie [134.226.63.120]) by smtp.cs.tcd.ie (Postfix) with ESMTP id E55EA3EE84; Wed, 22 Apr 2009 13:57:11 +0100 (IST) From: Alex McMahon To: DTN , dtn-users Content-Type: text/plain Organization: Trinity College Dublin Date: Wed, 22 Apr 2009 13:57:10 +0100 Message-Id: <1240405030.8846.46.camel@sphere> Mime-Version: 1.0 X-Mailer: Evolution 2.22.3.1 Content-Transfer-Encoding: 7bit Subject: [dtn-users] DTN mercurial repository on sourceforge X-BeenThere: dtn-users@maillists.intel-research.net X-Mailman-Version: 2.1.9 Precedence: list Reply-To: alex.mcmahon@cs.tcd.ie List-Id: dtn-users mailing list List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Wed, 22 Apr 2009 13:11:31 -0000 Hi, There are several folks with pending DTN2 contributions and there has been confusion on access to the repository and where the repository "lives". Basically, from here on I shall be responsible for that sort of thing and shall strive for continuity to the way Michael Demmer managed the code. The DTN2 repository is located at /home/groups/d/dt/dtn on web.sourceforge.net. It is probably a good idea to reiterate how to contribute to the main source tree. The conventions provided in http://www.dtnrg.org/docs/code/DTN2/CONTRIBUTING briefly describe how to get involved with DTN development. To contribute back to the main source tree, first do a commit to the clone of the repository that you are developing on and then export a changeset using: hg export REV Where REV is the revision number of your commit. Capture the output from the command into a file and email it as an attachment to dtn-users@mailman.dtnrg.org. Using the exported changeset (as opposed to a simple patch) is preferable because the changeset will include your name, email, and checkin comment which will be put into the mercurial change log. You can do this in bulk for multiple commits using: hg export -o mypatch-%n REV1 [REV2] [REV3] This will create N files, one for each revision, named "mypatch-1", "mypatch-2", etc. Run 'hg help export' for more information. Any bugs should be reported to dtn-users _at_ mailman _dot_ dtnrg _dot_ org. All the best, Alex McMahon