[OpenSIPS-Users] TCP Errors

Bogdan-Andrei Iancu bogdan at opensips.org
Wed Aug 6 17:07:10 CEST 2014


Hi,

Make a pcap of a call to see if the what's broken in that ACK.

Regards,

Bogdan-Andrei Iancu
OpenSIPS Founder and Developer
http://www.opensips-solutions.com

On 02.07.2014 18:30, Gary Nyquist wrote:
> Thanks Liviu for your advice,
> Wondering how can this be fixed... it is just filling up the log files 
> with hundreds of repeating lines.
> Also, thinking how to stop the "out-of-dialog ACKs to 487 responses" 
> to be generated in the first place (if it is the culprit indeed).
> Any suggestion?
> Thanks again...
> BR
> - Gary
> *Sent:* Tuesday, July 01, 2014 at 8:05 AM
> *From:* "Liviu Chircu" <liviu at opensips.org>
> *To:* users at lists.opensips.org
> *Subject:* Re: [OpenSIPS-Users] TCP Errors
> Hello Gary,
>
> That "error" is most likely due to out-of-dialog ACKs to 487 responses 
> (CANCEL scenarios). Thank you for reporting - it should actually be a 
> debugging message.
>
> Best regards,
> Liviu Chircu
> OpenSIPS Developer
> http://www.opensips-solutions.com
> On 06/30/2014 07:00 PM, Gary Nyquist wrote:
>
>     Hi Vlad,
>     Thanks for taking up this issue.
>     1.  Yes. tcp_async=1
>     2.  Running on CentOS ( 2.6.32-358.6.2.el6.x86_64 )
>     3.  It's occurring on every INVITE
>     May be unrelated to this bug; but I am also seeing error messages
>     like:
>     ERROR:script_helper:run_helper_logic: failed to match dialog, ci
>     '4E14BD6E-53B17A8F00026071-6B1EA700'
>     Thanks again...
>     BR
>     -Gary
>     *Sent:* Monday, June 30, 2014 at 5:06 AM
>     *From:* "Vlad Paiu" <vladpaiu at opensips.org>
>     *To:* users at lists.opensips.org
>     *Cc:* gn62 at gmx.us
>     *Subject:* Re: [OpenSIPS-Users] TCP Errors
>     Hello,
>
>     First of all, are you using the async_tcp option in your OpenSIPS
>     script ?
>     Also, what architecture / OS are you running on ?
>     What steps are you taking in reproducing this ?
>
>     Best Regards,
>
>     Vlad Paiu
>     OpenSIPS Developer
>     http://www.opensips-solutions.com  
>
>     On 27.06.2014 19:59, Gary Nyquist wrote:
>
>         Thanks Bogdan for looking into it.
>         Here is the version:
>         opensips -V
>         version: opensips 1.11.1-tls (x86_64/linux)
>         flags: STATS: On, USE_TCP, USE_TLS, DISABLE_NAGLE, SHM_MEM,
>         SHM_MMAP, PKG_MALLOC, F_MALLOC, USE_SHM_MEM,
>         FAST_LOCK-ADAPTIVE_WAIT
>         ADAPTIVE_WAIT_LOOPS=1024, MAX_RECV_BUFFER_SIZE 262144,
>         MAX_LISTEN 16, MAX_URI_SIZE 1024, BUF_SIZE 65535
>         poll method support: poll, epoll_lt, epoll_et, sigio_rt, select.
>         git revision: dbc8653
>         main.c compiled on 16:17:43 Jun 26 2014 with gcc 4.4.7
>         Yes, this bug is reproducable.
>         When it appears, the log file instantly gets filled with
>         hundreds of repeating lines like this:
>         Jun 26 20:30:31 ip-10-0-0-30 /usr/sbin/opensips[15229]:
>         DBG:core:handle_tcpconn_ev: data available on 0x7f40bcec7398 6
>         Jun 26 20:30:31 ip-10-0-0-30 /usr/sbin/opensips[15229]:
>         DBG:core:io_watch_del: io_watch_del op on index -1 6
>         (0x7dff20, 6, -1, 0x0,0x1) fd_no=110 called
>         Jun 26 20:30:31 ip-10-0-0-30 /usr/sbin/opensips[15229]:
>         ERROR:core:io_watch_del: BUG - trying to del fd 6 with flags 2 1
>         Jun 26 20:30:31 ip-10-0-0-30 /usr/sbin/opensips[15229]:
>         DBG:core:handle_tcpconn_ev: data available on 0x7f40bcec7398 6
>         Jun 26 20:30:31 ip-10-0-0-30 /usr/sbin/opensips[15229]:
>         DBG:core:io_watch_del: io_watch_del op on index -1 6
>         (0x7dff20, 6, -1, 0x0,0x1) fd_no=110 called
>         Jun 26 20:30:31 ip-10-0-0-30 /usr/sbin/opensips[15229]:
>         ERROR:core:io_watch_del: BUG - trying to del fd 6 with flags 2 1
>         Thanks again for your help.
>         BR
>         -Gary
>         *Sent:* Thursday, June 26, 2014 at 10:56 AM
>         *From:* "Bogdan-Andrei Iancu" <bogdan at opensips.org>
>         *To:* "Gary Nyquist" <gn62 at gmx.us>
>         *Cc:* "OpenSIPS users mailling list" <users at lists.opensips.org>
>         *Subject:* Re: [OpenSIPS-Users] TCP Errors
>         Hi,
>
>         That bug log actually says OpenSIPS tries to remove a
>         connection marked as READ from a list for WRITEs :)..
>
>         I will look into that. What exact version do you use (opensips
>         -V) ? also, can you reproduce this  bug ?
>
>         Thanks and regards,
>
>         Bogdan-Andrei Iancu
>         OpenSIPS Founder and Developer
>         http://www.opensips-solutions.com
>
>         On 25.06.2014 20:27, Gary Nyquist wrote:
>
>             Thanks Bogdan for the detailed reply.
>             I implemented your advice.
>             Those error messages are now gone.
>             But seeing a new error now; not sure if it is related to
>             that...
>             ERROR:core:io_watch_del: BUG - trying to del fd 36 with
>             flags 2 1
>             Any advice?
>             BR
>             -Gary
>             *Sent:* Wednesday, June 25, 2014 at 5:54 AM
>             *From:* "Bogdan-Andrei Iancu" <bogdan at opensips.org>
>             *To:* "OpenSIPS users mailling list"
>             <users at lists.opensips.org>, gn62 at gmx.us
>             *Subject:* Re: [OpenSIPS-Users] TCP Errors
>             Hi,
>
>             Those messages say that OpenSIPS tried to open a TCP
>             connection to a party which does not respond -> there was
>             a timeout for connect in 10 seconds. Because of this
>             blocking in connects, there were not more opensips workers
>             available to handle other traffic.
>
>             So what you need to do is :
>                 - minimize the impact of the blocking connect - see my
>             previous email on reducing the connect timeout
>                 - you may configure OpenSIPS not to open new TCP
>             connect (but to reuse the existing ones, open by clients).
>             See tcp_no_new_conn_bflag
>             http://www.opensips.org/Documentation/Script-CoreParameters-1-11#toc96
>                 - try to understand the SIP patterns where such TCP
>             connect fails so you can avoid them at script level.
>
>             Regards,
>
>             Bogdan-Andrei Iancu
>             OpenSIPS Founder and Developer
>             http://www.opensips-solutions.com
>
>             On 24.06.2014 21:34, Gary Nyquist wrote:
>
>                 Hi,
>
>                 The following lines are repeating in the log.
>
>                 ERROR:core:tcp_blocking_connect: timeout 10 s elapsed
>                 from 10 s
>                 ERROR:core:tcpconn_connect: tcp_blocking_connect failed
>                 ERROR:core:tcp_send: connect failed
>                 ERROR:tm:msg_send: tcp_send failed
>
>                 INFO:core:send2child: no free tcp receiver, connection
>                 passed to the least busy one
>
>                 Any guess, what could be the reason?
>
>                 Best
>
>                 --Gary
>
>                 _______________________________________________
>                 Users mailing list
>                 Users at lists.opensips.org
>                 http://lists.opensips.org/cgi-bin/mailman/listinfo/users
>
>         _______________________________________________
>         Users mailing list
>         Users at lists.opensips.org
>         http://lists.opensips.org/cgi-bin/mailman/listinfo/users
>
>     _______________________________________________
>     Users mailing list
>     Users at lists.opensips.org
>     http://lists.opensips.org/cgi-bin/mailman/listinfo/users
>
>
> _______________________________________________ Users mailing list 
> Users at lists.opensips.org 
> http://lists.opensips.org/cgi-bin/mailman/listinfo/users
>
>
> _______________________________________________
> Users mailing list
> Users at lists.opensips.org
> http://lists.opensips.org/cgi-bin/mailman/listinfo/users

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.opensips.org/pipermail/users/attachments/20140806/292f43e3/attachment-0001.htm>


More information about the Users mailing list