[SR-Users] Strange behavior for unregister() function. Wrong docs? Bug?

Tue Aug 13 15:41:42 CEST 2019

It is a bit counterintuitive, though. Like Joel, I would have thought such restrictions ultimately should ultimately be based on the type of the toplevel action.

Then again, I make use of this a lot from failure_route, calling the same route blocks over and over for route-advancing, some of them containing function calls which are not strictly possible in failure_route. The only thing one really has to be careful of in that scenario is transaction state; can’t send stateless replies out of failure_route, so accordingly cannot send them from a descendant route either. That’s an excellent use-case for send_reply(), which does the right thing (in effect, sl_send_reply() vs t_reply()) depending on whether a transaction context exists.

—
Sent from mobile, with due apologies for brevity and errors.

> On Aug 13, 2019, at 3:38 AM, Joel Serrano <joel at textplus.com> wrote:
> 
> Great!!! This is good to know!!
> 
> Thank you Daniel!
> Joel.
> 
>> On Tue, Aug 13, 2019 at 12:36 AM Daniel-Constantin Mierla <miconda at gmail.com> wrote:
>> Hello,
>> 
>> that is a "known hack" to go around the route block type restrictions for functions.
>> 
>> The purpose of the restrictions is to limit the usage in the blocks where the developer intended to allow the function to be used, like she/he tested for those route blocks or thought it is enough there and didn't analyze the effects when using in other route block types.
>> 
>> However, many functions are just safe to use in other route block types than those imposed by the developer.
>> 
>> Probably the unregister() is safe in onreply block, because it doesn't need to generate a reply or use information from request URI.
>> 
>> Cheers,
>> Daniel
>> 
>>> On 13.08.19 07:28, Joel Serrano wrote:
>>> Forgot to add:
>>> 
>>> root at csbc01:~# kamailio -V
>>> version: kamailio 5.3.0-dev7 (x86_64/linux)
>>> flags: STATS: Off, USE_TCP, USE_TLS, USE_SCTP, TLS_HOOKS, USE_RAW_SOCKS, DISABLE_NAGLE, USE_MCAST, DNS_IP_HACK, SHM_MMAP, PKG_MALLOC, Q_MALLOC, F_MALLOC, TLSF_MALLOC, DBG_SR_MEMORY, USE_FUTEX, FAST_LOCK-ADAPTIVE_WAIT, USE_DNS_CACHE, USE_DNS_FAILOVER, USE_NAPTR, USE_DST_BLACKLIST, HAVE_RESOLV_RES
>>> ADAPTIVE_WAIT_LOOPS 1024, MAX_RECV_BUFFER_SIZE 262144, MAX_URI_SIZE 1024, BUF_SIZE 65535, DEFAULT PKG_SIZE 8MB
>>> poll method support: poll, epoll_lt, epoll_et, sigio_rt, select.
>>> id: unknown
>>> compiled with gcc 8.3.0
>>> root at csbc01:~#
>>> 
>>> 
>>> 
>>>> On Mon, Aug 12, 2019 at 2:13 PM Joel Serrano <joel at textplus.com> wrote:
>>>> Hello, 
>>>> 
>>>> I'm playing around with the unregister() function from the registrar module:
>>>> 
>>>> https://www.kamailio.org/docs/modules/devel/modules/registrar.html#registrar.f.unregister
>>>> 
>>>> Docs say that that it can be used in REQUEST_ROUTE or FAILURE_ROUTE.
>>>> 
>>>> I needed it in REPLY_ROUTE, and before reading the docs I tried it and got this error:
>>>> 
>>>> Aug 12 15:52:44 csbc01 kamailio: CRITICAL: <core> [core/cfg.y:3526]: yyerror_at(): parse error in config file /etc/kamailio/csbc/reply_routes.cfg, line 27, column 71: Command cannot be used in the block#012
>>>> Aug 12 15:52:44 csbc01 kamailio[12737]: ERROR: bad config file (1 errors)
>>>> 
>>>> But, I have this working on another setup, I compared, and my difference is that I'm running unregister() inside a route, and calling that route from reply_route, and that works (and by works, I mean it also does what it is supposed to do, which is unregister the AoR).
>>>> 
>>>> So this doesn't work:
>>>> 
>>>> ...
>>>> onreply_route[MANAGE_REG_REPLY] {
>>>>      if(status=~"2[0-9][0-9]") {
>>>>          if(($sel(contact.expires)==0) || ($hdr(Expires)==0)){
>>>>             if (unregister("location", "$sht(sipserver=>$fU::contact)")) {
>>>>                 xlog("L_NOTICE", "[usrloc] removed user location\n");
>>>>             }
>>>>             route(DEL_SIPSERVER);
>>>> }
>>>> ...
>>>> 
>>>> But this works:
>>>> 
>>>> ...
>>>> onreply_route[MANAGE_REG_REPLY] {
>>>>      if(status=~"2[0-9][0-9]") {
>>>>          if(($sel(contact.expires)==0) || ($hdr(Expires)==0)){
>>>>             route(UNREG_CUSTOMER);
>>>>             route(DEL_SIPSERVER);
>>>> }
>>>> ...
>>>> 
>>>> ...
>>>> route[UNREG_CUSTOMER] {
>>>>             if (unregister("location", "$sht(sipserver=>$fU::contact)")) {
>>>>                 xlog("L_NOTICE", "[usrloc] removed user location\n");
>>>>             }
>>>> }
>>>> ...
>>>> 
>>>> 
>>>> 
>>>> So I'm in no-mans-land right now, docs say it's not allowed in REPLY_ROUTE, and if I try it it will fail, but if I try it inside a route, it woks (hack?).
>>>> 
>>>> What is the correct approach here? Should unregister() be valid in reply_route (like other similar functions, for example: save()) and therefor docs should be updated and the error in the logs should also be updated, or if it's really not supported, shouldn't Kamailio not let me use it anywhere that a ends up being called from reply_route?
>>>> 
>>>> I hope I have explained myself correctly. As always, please let me know if this should be a GH issue instead and I'll happily create it.
>>>> 
>>>> Thanks, 
>>>> Joel.
>>>> 
>>>> 
>>>> 
>>> 
>>> 
>>> _______________________________________________
>>> Kamailio (SER) - Users Mailing List
>>> sr-users at lists.kamailio.org
>>> https://lists.kamailio.org/cgi-bin/mailman/listinfo/sr-users
>> -- 
>> Daniel-Constantin Mierla -- www.asipto.com
>> www.twitter.com/miconda -- www.linkedin.com/in/miconda
> _______________________________________________
> Kamailio (SER) - Users Mailing List
> sr-users at lists.kamailio.org
> https://lists.kamailio.org/cgi-bin/mailman/listinfo/sr-users
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.kamailio.org/pipermail/sr-users/attachments/20190813/0b3cda90/attachment.html>