[reSIProcate] Duplicate transaction ids on Windows

[Scott Godin]

In windows.... If you are using resip 0.9 or earlier - you must make
sure that you seed each thread independently.  This is fixed in SVN
head.  

 

You shouldn't be ending up with so many tid collisions if you follow the
above rule.

 

Scott

 

________________________________

From: resiprocate-devel-bounces at list.sipfoundry.org
[mailto:resiprocate-devel-bounces at list.sipfoundry.org] On Behalf Of
Robert Mansfield
Sent: Wednesday, December 21, 2005 2:05 PM
To: resiprocate-devel at list.sipfoundry.org
Subject: [reSIProcate] Duplicate transaction ids on Windows

 

All,

 

I've read through the previous correspondance on this issue and am
hoping someone can clarify the best practice for a multithreaded
application on widows. In particular the use of the 'Random' class for
generating unique transaction ids. 

 

I'm running two windows applications on seperate hosts, one acting as a
UAC, the other a UAS. There are 60 simumtaneous dialogs being created.

 

During load testing I submit a BYE request to the stack but the stack
doesn't transmit the message to the wire, claiming the transaction id
already exists in the transaction map.

 

The application has two threads, one processing SIP messages from the
stack (g_SipStack->receive()) and the other thread sending messages to
the stack (g_SipStack->send(sipMessage). The stack object is global to
both threads, but only one thread is responsible for creating and
submitting SIP requests. 

 

On the UAC side I have seen a duplicate transaction ids submitted to the
TransactionManager from the application, typically a BYE is discarded
because the tid has been used for an ACK from a different dialog.

 

In the creation of both the ACK  and the BYE message the 'via' header
branch parameter is instantiated with a call to "via.param(p_branch)".

 

In this particular stack trace instance the transaction id appears to
have been added from a received response, here is a fragment from the
stack trace:

 

########################################################################
############################################

DEBUG | 20051221-152514.156 | RT-SIP-00-19-07-5.bb.cc.dd.exe | RESIP |
1940 | SipStack.cxx:375 | RECV: SipResp: 200 tid=8664a730b52dc246
cseq=INVITE contact=RedwoodDevEngC1P18 at 10.0.0.126 / 1 from(wire)

 

DEBUG | 20051221-152521.562 | RT-SIP-00-19-07-5.bb.cc.dd.exe | RESIP |
1196 | SipStack.cxx:269 | SEND: SipReq:  BYE
RedwoodDevEngC1P21 at 10.0.0.126 tid=8664a730b52dc246 cseq=BYE / 2 from(tu)

 

STACK | 20051221-152522.281 | RT-SIP-00-19-07-5.bb.cc.dd.exe |
RESIP:TRANSACTION | 1940 | TransactionState.cxx:155 | Found matching
transaction for SipReq:  BYE RedwoodDevEngC1P21 at 10.0.0.126
tid=8664a730b52dc246 cseq=BYE / 2 from(tu) -> tid=8664a730b52dc246 [
ClientStale/Terminated unreliable target=[ V4 0.0.0.0:0
UNKNOWN_TRANSPORT connectionId=0 ]]
STACK | 20051221-152522.281 | RT-SIP-00-19-07-5.bb.cc.dd.exe |
RESIP:TRANSACTION | 1940 | TransactionState.cxx:1151 |
TransactionState::processClientStale: SipReq:  BYE
RedwoodDevEngC1P21 at 10.0.0.126 tid=8664a730b52dc246 cseq=BYE / 2 from(tu)
STACK | 20051221-152522.281 | RT-SIP-00-19-07-5.bb.cc.dd.exe |
RESIP:TRANSACTION | 1940 | TransactionState.cxx:1185 | Discarding extra
response: BYE sip:RedwoodDevEngC1P21 at 10.0.0.126 SIP/2.0

 

To: <sip:P335544401223 at 10.0.0.126>;tag=3c6fe50d

 

From: <sip:DNX_TESTC2P8 at 10.0.0.212>;tag=9b540c2f

 

Via: SIP/2.0/UDP
10.0.0.212:5060;branch=z9hG4bK-d87543-8664a730b52dc246-1--d87543-;rport

 

Call-ID: 4767b466384e6543 at UmVkVk9JUA..

 

CSeq: 2 BYE

 

Content-Length: 0

########################################################################
#############################################

 

So I guess I have two questions:

a) if two resiprocate applications are running back to back would a
received tid (generated remotely for dialog A) confuse the stack when
submitting a request with the same tid generated locally for dialog B
(branch parameter unique across space and time)?

b) just how random is the tid generated via the Random class. Should I
expect so many 'duplicates' after running just one iteration of the
test?

 

Regards,

 

Rob Mansfield.

 

============================

Robert Mansfield

Senior Software Engineer

Redwood Technologies Limited

Tel +[44] (0)1344 304 344

Fax +[44] (0)1344 304 345

E:mail mailto:rjm at redwoodtech.com <mailto:rjm at redwoodtech.com> 

Web http://www.redwoodtech.com <http://www.redwoodtech.com/> 

============================

Email Disclaimer

The information in this email is confidential and may be legally
privileged. It is intended solely for the addressee. Access to this
email by anyone else is unauthorised. If you are not the intended
recipient, any disclosure, copying, distribution or any action taken or
omitted to be taken in reliance on it is prohibited and may be unlawful.
When addressed to our clients any opinions or advice contained in this
email are subject to the limitations of Redwood Technologies Limited's
standard terms and conditions of contract.

 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://list.resiprocate.org/pipermail/resiprocate-devel/attachments/20051221/feab09bf/attachment.htm>