[sr-dev] tm error: status rewrite by UAS

Juha Heinanen jh at tutpro.com
Mon Jun 22 17:03:18 CEST 2009


Andrei Pelinescu-Onciul writes:

 > > i set fr_timer_avp to 3000 (3 sec) before calling t_relay.  i have
 >                        ^^^^^^ the avp timers  are in seconds, so you should
 > set them it to 3. All the other timeouts are in ms (I'll add a note to
 > the docs).

i did so, but it didn't affect the error messages.  while doing it i
noticed that fr_timer is not very accurate.  i call next_gw() when
previous call times out and syslog reports 5 second intervals:

Jun 22 17:56:46 localhost /usr/sbin/sip-router[5881]: INFO: Routing initial INVITE to <sip:0407058050 at 192.98.101.21:5060> and <<null>>
Jun 22 17:56:51 localhost /usr/sbin/sip-router[5909]: INFO: Routing next INVITE to <sip:0407058050 at 192.98.101.24:5060>
Jun 22 17:56:56 localhost /usr/sbin/sip-router[5909]: INFO: Routing next INVITE to <sip:0407058050 at 192.98.101.23:5060>
Jun 22 17:57:01 localhost /usr/sbin/sip-router[5909]: INFO: Routing next INVITE to <sip:0407058050 at 192.98.101.22:5060>
Jun 22 17:57:06 localhost /usr/sbin/sip-router[5909]: INFO: Routing next INVITE to <sip:0407058050 at 192.98.100.10:5060>

 > An avp always has a value if it exists. If its value is the empty string
 > or 0 => the avp integer value will be 0 => in this case the timer will
 > be set to 0.

i added xlog statement before first next_gw() call and it reports that
fr_inv_timer_avp has value null:

Jun 22 17:56:46 localhost /usr/sbin/sip-router[5881]: INFO: fr_inv_timer_avp <null>

 > I'll add a check for 0 in sr too (so that it will be compatible with the
 > k way).

that would be nice so that script writer can leave the avp unset if
there is no need to change the default set by the module parameter
variable.

 > Note that in ser the modern way to use avps for timers is via t_set_fr()
 > & friends (you can set everything from fr timers to retransmission and
 > lifetime, you can later reset to default via t_reset_fr() and you can
 > use it only when needed).

thanks for the pointer. i'll take a look at those.

-- juha



More information about the sr-dev mailing list