[SR-Users] Catching websocket / TCP errors

Marrold kamailio at marrold.co.uk
Tue Oct 5 17:47:27 CEST 2021


We're currently implementing WebRTC (SIP over a websocket) on our Kamailio
proxies and we're experiencing a minor issue after a restart.

We write all USRLOC entries to a shared database to implement HA. These are
then loaded on a restart, which leaves us with entries with TCP connections
that no longer exist. When a call attempt is made the proxy can't find the
connection ID and immediately rejects the call with a "500 No error (2/SL)"

Oct  5 15:38:24 proxy-01 /sbin/kamailio[32087]: WARNING: <core>
[core/msg_translator.c:2861]: via_builder(): TCP/TLS connection (id: 0) for
WebSocket could not be found
Oct  5 15:38:24 proxy-01 /sbin/kamailio[32087]: ERROR: <core>
[core/msg_translator.c:2038]: build_req_buf_from_sip_req(): could not
create Via header
Oct  5 15:38:24 proxy-01 /sbin/kamailio[32087]: ERROR: tm [t_fwd.c:479]:
prepare_new_uac(): could not build request
Oct  5 15:38:24 proxy-01 /sbin/kamailio[32087]: ERROR: tm [t_fwd.c:1759]:
t_forward_nonack(): failure to add branches
Oct  5 15:38:24 proxy-01 /sbin/kamailio[32087]: ERROR: sl [sl_funcs.c:414]:
sl_reply_error(): stateless error reply used: No error (2/SL)

How can we catch this error to avoid terminating the call prematurely? Or
should we flush these TCP connections on start up?

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.kamailio.org/pipermail/sr-users/attachments/20211005/2e1cb05a/attachment.htm>

More information about the sr-users mailing list