On 11.08.2010 23:39, dotnetdub wrote:

The same endpoint never causes our Kamailio 1.4.1 servers to generate this message. I have watched the SIP transactions and the ACK always seems to arrive at the correct point in the dialog. The endpoint is Asterisk 1.4.22.

The reason why you never saw the message show up on Kamailio 1.4 was probably that the same race condition was fixed in Kamailio at some point (I believe the thread you referred to in your initial mail mentioned that, specifically by Ovidiu).

However, when the tm module changed in SIP-Router 3 somewhat, the fix was left out and only brought back in by Daniel at the beginning of July. As I pointed out in my last mail, this fix apparently didn't make it into any branch other than master yet.

Cheers,

--Timo

On Wednesday 11 August 2010, Timo Reimann wrote:

[..] It definitely doesn't cause any critical issues to the proper working of SIP-Router. The worst it can do is mess up a call's dialog state when the race condition kicks in. Once the race condition is ruled out, occurrences of the message boil down to broken UAs as Henning pointed out already.

Thanks for the clarification Timo,

i forgot about the state of the race condition in current 3.0 branch.

Henning

On 12 August 2010 10:21, Daniel-Constantin Mierla miconda@gmail.com wrote:

Hi Timo,

On 8/11/10 11:58 PM, Timo Reimann wrote:

...

Hey Brian,

On 11.08.2010 13:01, dotnetdub wrote:

...

I see this in every call setup..

CRITICAL: dialog [dlg_hash.c:591]: bogus event 6 in state 2 for dlg 0xaf933bd0 [3739:1530909163] with clid '5959362a340dfa7938cf52ae1ba4b9d7@my.proxy.com mailto:5959362a340dfa7938cf52ae1ba4b9d7@my.proxy.com' and tags 'as5b636a8a' ''

I am running SIP-Router 3.01

I note this thread: http://www.mail-archive.com/sr-dev@lists.sip-router.org/msg05668.html

but it's inconclusive.

The conclusion took place in a different thread: Daniel committed a patch to master on 2 July

http://lists.sip-router.org/pipermail/sr-dev/2010-July/008097.html

and asked people to test the patch on 5 July prior to backporting it into 3.0:

http://lists.sip-router.org/pipermail/sr-dev/2010-July/008109.html

Klaus Feichtinger took the time for validation, and things seem to be looking good. However, AFAICS the patch wasn't brought into 3.0 which means that you still run on a race-conditioned version of SIP-Router that causes the log message to appear quite frequently yet.

Daniel, considering Klaus' successful test runs would it be acceptable to get the patch (commit f1c35c1325) into 3.0 from your point of view?

it is backported, I did many cherry-picks resulting in a single email with bulk commits to kamailio_3.0 content: http://lists.sip-router.org/pipermail/sr-dev/2010-July/008213.html

Cheers, Daniel

It doesn't seem to cause any issue but I would like to get it sorted.

...

...

It definitely doesn't cause any critical issues to the proper working of SIP-Router. The worst it can do is mess up a call's dialog state when the race condition kicks in. Once the race condition is ruled out, occurrences of the message boil down to broken UAs as Henning pointed out already.

HTH,

--Timo

-- Daniel-Constantin Mierla http://www.asipto.com/

Thanks Gents.

I will wait until 3.1 drops and update these two boxes to get the fix.

Brian

On 8/12/10 6:02 PM, dotnetdub wrote:

On 12 August 2010 10:21, Daniel-Constantin Mierla <miconda@gmail.com mailto:miconda@gmail.com> wrote:

 Hi Timo,


On 8/11/10 11:58 PM, Timo Reimann wrote:

    Hey Brian,


    On 11.08.2010 13:01, dotnetdub wrote:

        I see this in every call setup..

        CRITICAL: dialog [dlg_hash.c:591]: bogus event 6 in state
        2 for
        dlg 0xaf933bd0 [3739:1530909163] with clid
        '5959362a340dfa7938cf52ae1ba4b9d7@my.proxy.com
        <mailto:5959362a340dfa7938cf52ae1ba4b9d7@my.proxy.com>
        <mailto:5959362a340dfa7938cf52ae1ba4b9d7@my.proxy.com
        <mailto:5959362a340dfa7938cf52ae1ba4b9d7@my.proxy.com>>'
        and tags
        'as5b636a8a' ''

        I am running SIP-Router 3.01

        I note this
        thread:
        http://www.mail-archive.com/sr-dev@lists.sip-router.org/msg05668.html

        but it's inconclusive.

    The conclusion took place in a different thread: Daniel
    committed a
    patch to master on 2 July

    http://lists.sip-router.org/pipermail/sr-dev/2010-July/008097.html

    and asked people to test the patch on 5 July prior to
    backporting it
    into 3.0:

    http://lists.sip-router.org/pipermail/sr-dev/2010-July/008109.html

    Klaus Feichtinger took the time for validation, and things
    seem to be
    looking good. However, AFAICS the patch wasn't brought into
    3.0 which
    means that you still run on a race-conditioned version of
    SIP-Router
    that causes the log message to appear quite frequently yet.

    Daniel, considering Klaus' successful test runs would it be
    acceptable
    to get the patch (commit f1c35c1325) into 3.0 from your point
    of view?


it is backported, I did many cherry-picks resulting in a single
email with bulk commits to kamailio_3.0 content:
http://lists.sip-router.org/pipermail/sr-dev/2010-July/008213.html

Cheers,
Daniel


        It doesn't seem to cause any issue but I would like to get
        it sorted.

    It definitely doesn't cause any critical issues to the proper
    working of
    SIP-Router. The worst it can do is mess up a call's dialog
    state when
    the race condition kicks in. Once the race condition is ruled out,
    occurrences of the message boil down to broken UAs as Henning
    pointed
    out already.


    HTH,

    --Timo


-- 
Daniel-Constantin Mierla
http://www.asipto.com/

Thanks Gents.

I will wait until 3.1 drops and update these two boxes to get the fix.

the patch is also in 3.0.x, you have to get latest git branch kamailio_3.0 -- no need to wait for 3.1.0.

Cheers, Daniel