Re: [SR-Users] Crash Kamailio 4.1.3

Daniel-Constantin Mierla Thu, 12 Jun 2014 08:03:58 -0700

Hello,

if you get a record for 180 response, then you have also the early_mediaparameter set for acc module, isn't it?

In the morning I pushed a patch that should fix this issue. Use latestrelease 4.1.4 and see if works fine. Report the results to know that itis gone or not.


Cheers,
Daniel

On 12/06/14 16:55, Igor Potjevlesch wrote:

Hello,

We don't use $ai in xlog nor in other process. only in ACC.

parameters for ACC are :
modparam("acc", "db_flag", FLT_ACC)
modparam("acc", "db_missed_flag", 3)
modparam("acc", "db_url", DBURLWO)
modparam("acc", "db_extra",
"src_user=$fU;username=$Au;src_domain=$fd;src_ip=$si;src_pai=$ai;"
        "dst_ouser=$tU;dst_user=$rU;dst_domain=$rd")

For the 3781-4b1-572014182635-OGNAJ-1-A.B.C.D there is a code 180ringing in the INVITE in ACC table.


regards,

Igor

2014-06-11 23:01 GMT+02:00 Daniel-Constantin Mierla <mico...@gmail.com<mailto:mico...@gmail.com>>:


    Few more things...

    Are you recording 1xx events? Can you check to see if there is
    another record in acc table for the same call? You can search by
    call-id 3781-4b1-572014182635-OGNAJ-1-A.B.C.D

    Eventually, send also the parameters you set for acc module.

    Cheers,
    Daniel


    On 11/06/14 19:25, Daniel-Constantin Mierla wrote:

    Hello,

    so you don't print $ai in xlog() statements nor use it in any
    assignments or other functions besides acc parameter?

    Cheers,
    Daniel

    On 11/06/14 19:19, Igor Potjevlesch wrote:

    Hello,

    We do not access to the P-asserted-identity in our configuration
    but we added the field PAI in the db base ACC ( for INVITE, ACK
    and BYE) .
    I dont know if it's in request_route, failure_route or
    branch_route .

    This is the print :

    (gdb) p mem_block
    $3 = (struct qm_block *) 0x7f6a6bef1010
    (gdb) p shm_block
    $4 = (struct qm_block *) 0x7f6a5666a000

    Regards,

    Igor


    2014-06-11 18:02 GMT+02:00 Daniel-Constantin Mierla
    <mico...@gmail.com <mailto:mico...@gmail.com>>:

        Hello,

        cloning to shm for tm seems ok. Can you tell where you
        access P-Asserted-Identity header, via variables? Does it
        happen in request_route, failure_route or branch_route?

        Can you print from gdb, any frame:

        p mem_block
        p shm_block

        I want to see if parsed filed point to shm or pkg memory.

        Cheers,
        Daniel


        On 11/06/14 17:37, Daniel-Constantin Mierla wrote:

        Hello,

        at least I narrowed it down a bit. It is empty also in the
        clone stored in transaction, so it happens either during
        cloning or before. I will have to check these parts.

        Cheers,
        Daniel

        On 11/06/14 17:00, Igor Potjevlesch wrote:

        Hello,

        This is the result, always for frame 5 :

        (gdb) p *t->uas.request->pai
        $1 = {type = HDR_PAI_T, name = {
            s = 0x7f6a60cd34b8 "P-Asserted-Identity:
        \"0987654321\"<sip:0987654321@A.B.C.D>\r\nContact:
        <sip:0987654321@A.B.C.D:5060>\r\nAllow: INVITE, BYE,
        REGISTER, ACK, OPTIONS, CANCEL, SUBSCRIBE, NOTIFY, INFO,
        REFER, UPD"..., len = 19}, body = {
            s = 0x7f6a60cd34cd
        "\"0987654321\"<sip:0987654321@A.B.C.D>\r\nContact:
        <sip:0987654321@A.B.C.D:5060>\r\nAllow: INVITE, BYE,
        REGISTER, ACK, OPTIONS, CANCEL, SUBSCRIBE, NOTIFY, INFO,
        REFER, UPDATE\r\nSupported: path,"..., len = 43}, len =
        66, parsed = 0x7f6a6d81da88, next = 0x7f6a60cd3f10}

        (gdb) p *((p_id_body_t*)(t->uas.request->pai->parsed))
        $2 = {id = 0x0, num_ids = 0, next = 0x0}

        *Did *you find one thing in**common between the 2
        occurrences? Do you have any ideas about what is the cause
        of this pai reset?

        Regards,

        Igor




        2014-06-11 16:09 GMT+02:00 Daniel-Constantin Mierla
        <mico...@gmail.com <mailto:mico...@gmail.com>>:

            Hello,

            in the same frame 5, can yo get:

            p *t->uas.request->pai
            p *((p_id_body_t*)(t->uas.request->pai->parsed))

            Cheers,
            Daniel


            On 10/06/14 18:35, Igor Potjevlesch wrote:

            Hello,

            Here is the results :
            _
            _
            (gdb) frame 5
            #5 0x00007f6a687e9b43 in acc_onreply
            (t=0x7f6a60d16ff8, req=0x7f6a60cd2c10,
            reply=0x7f6a6c119aa8, code=200) at acc_logic.c:471
            471 acc_db_request(req);