Hi Alex,

 

Thanks for sharing more insights about that. Yes, indeed hard to say why its happens without digging into TM internals.

 

Does this also happens e.g. if you just create a reply with send_reply/t_reply, and not with the auth_challenge()?

If it’s a generic issue, it might be sensible to create an issue about it to at least document it in a structure way.

 

Cheers,

 

Henning

 

 

From: Alex Balashov <abalashov@evaristesys.com>
Sent: Thursday, December 15, 2022 9:03 PM
To: Kamailio (SER) - Users Mailing List <sr-users@lists.kamailio.org>
Subject: [SR-Users] Re: auth_challenge() and async

 

One more thing, and I apologise profusely for the spam:

This is not caused by immediate re-suspension of the transaction upon successful authorisation of new INVITE+credentials. Even if I attempt to complete the onward routing in the same route which does the authentication, without any further TM suspend/continue, I get the 407 retransmissions from the previous transaction, e.g.

route[RESUME] {

pv_auth_check() and the rest

 

$ru = "sip:asterisk-receiver:5060";

t_on_reply("MAIN_REPLY");

 

if(!t_relay())

t_reply("500", "Internal server error");

 

exit;

}

I can't say for sure what's going on, not knowing TM internals.


-- Alex


On Dec 15, 2022, at 2:40 PM, Alex Balashov <abalashov@evaristesys.com> wrote:

Setting tm.wt_timer to a very low value (e.g. 200 ms) does provide a hack around this behaviour, but it doesn't seem to me that this is the correct solution. 

                                                               │AC
             172.24.0.9:39777              172.24.0.7:5060     │K
          ───────────────────          ───────────────────   │si
  19:38:44.386913   │        INVITE (SDP)         │            │p:
        +0.001550   │ ──────────────────────────> │            │10
  19:38:44.388463   │  100 trying -- your call is │            │0@
        +0.003336   │ <────────────────────────── │            │si
  19:38:44.391799   │  407 Proxy Authentication R │            │p-
        +0.000234   │ <────────────────────────── │            │pr
  19:38:44.392033   │             ACK             │            │ox
        +0.201070   │ ──────────────────────────> │            │y-
  19:38:44.593103   │        INVITE (SDP)         │            │di
        +0.004226   │ ──────────────────────────> │            │ge
  19:38:44.597329   │  100 trying -- your call is │            │st
        +0.003063   │ <────────────────────────── │            │-a
  19:38:44.600392   │  407 Proxy Authentication R │            │ut
        +0.000489   │ <────────────────────────── │            │h:
  19:38:44.600881   │             ACK             │            │50
                    │ ──────────────────────────> │            │60
                    │                             │            │ S

The real question is why the negative ACK for the first transaction doesn't seem to be having the intended effect in this scenario.

-- Alex


On Dec 15, 2022, at 2:21 PM, Alex Balashov <abalashov@evaristesys.com> wrote:

Adding further to this, it seems to me the real problem is that I can't use t_release() in an async resume route, because it's internally structured to take place inside a failure_route context. If I could, I think that would rid me of the first transaction after I send the challenge and call 'exit'.


On Dec 15, 2022, at 12:42 PM, Alex Balashov <abalashov@evaristesys.com> wrote:

As a test, I tried to put the auth_challenge() in the request_route before any async suspension, and in that case works fine.

The issue is definitely with the way auth_challenge() issued from _within_ an async resume route (failure_route context) bears upon transaction state.

-- Alex


On Dec 15, 2022, at 12:23 PM, Alex Balashov <abalashov@evaristesys.com> wrote:

Well, the difference seems pretty clear. In a scenario with an auth challenge and no subsequent INVITE+credentials, the negative ACK is matched:

4(54) DEBUG: <core> [core/receive.c:389]: receive_msg(): --- received sip message - request - call-id: [01eed151-4234-4518-9a0e-9b9168f21a3f] - cseq: [288439 ACK]
4(54) DEBUG: <core> [core/receive.c:261]: ksr_evrt_pre_routing(): event route core:pre-routing not defined
4(54) DEBUG: <core> [core/receive.c:471]: receive_msg(): preparing to run routing scripts...
4(54) DEBUG: sl [sl_funcs.c:447]: sl_filter_ACK(): too late to be a local ACK!
4(54) DEBUG: <core> [core/parser/parse_hname2.c:301]: parse_sip_header_name(): parsed header name [Content-Length] type 12
4(54) DEBUG: <core> [core/parser/msg_parser.c:187]: get_hdr_field(): content_length=0
4(54) DEBUG: <core> [core/parser/msg_parser.c:91]: get_hdr_field(): found end of header
4(54) DEBUG: maxfwd [mf_funcs.c:55]: is_maxfwd_present(): max_forwards header not found!
4(54) DEBUG: siputils [checks.c:123]: has_totag(): totag found
4(54) DEBUG: rr [loose.c:108]: find_first_route(): No Route headers found
4(54) DEBUG: rr [loose.c:1006]: loose_route_mode(): There is no Route HF
4(54) DEBUG: tm [t_lookup.c:1053]: t_check_msg(): msg (0xffffa72f7088) id=14/54 global id=13/54 T start=0
4(54) DEBUG: tm [t_lookup.c:497]: t_lookup_request(): start searching: hash=42311, isACK=1
4(54) DEBUG: tm [t_lookup.c:439]: matching_3261(): RFC3261 transaction matched, tid=SG.ceb57d44-7388-4739-9a86-d44ea04d974d
4(54) DEBUG: tm [t_lookup.c:692]: t_lookup_request(): transaction found (T=0xffffa2f428a8)
4(54) DEBUG: tm [t_lookup.c:1122]: t_check_msg(): msg (0xffffa72f7088) id=14/54 global id=14/54 T end=0xffffa2f428a8
4(54) DEBUG: tm [t_reply.c:1763]: cleanup_uac_timers(): RETR/FR timers reset
4(54) DEBUG: tm [t_funcs.c:120]: put_on_wait(): put T [0xffffa2f428a8] on wait
4(54) DEBUG: <core> [core/timer.c:557]: timer_add_safe(): timer_add called on an active timer 0xffffa2f42930 (0xffffa2d05d08, 0xffffa2d05d08), flags 201
4(54) DEBUG: tm [t_funcs.c:143]: put_on_wait(): transaction 0xffffa2f428a8 already on wait

However, in a scenario with an auth challenge with subsequent INVITE+credentials, the same negative ACK is not matched to a known transaction.

2(52) DEBUG: <core> [core/receive.c:389]: receive_msg(): --- received sip message - request - call-id: [895a7051-3e0c-410a-88ea-4bad7a1c21b6] - cseq: [939189 ACK]
2(52) DEBUG: <core> [core/receive.c:261]: ksr_evrt_pre_routing(): event route core:pre-routing not defined
2(52) DEBUG: <core> [core/receive.c:471]: receive_msg(): preparing to run routing scripts...
2(52) DEBUG: sl [sl_funcs.c:447]: sl_filter_ACK(): too late to be a local ACK!
2(52) DEBUG: <core> [core/parser/parse_hname2.c:301]: parse_sip_header_name(): parsed header name [Content-Length] type 12
2(52) DEBUG: <core> [core/parser/msg_parser.c:187]: get_hdr_field(): content_length=0
2(52) DEBUG: <core> [core/parser/msg_parser.c:91]: get_hdr_field(): found end of header
2(52) DEBUG: maxfwd [mf_funcs.c:55]: is_maxfwd_present(): max_forwards header not found!
2(52) DEBUG: siputils [checks.c:123]: has_totag(): totag found
2(52) DEBUG: rr [loose.c:108]: find_first_route(): No Route headers found
2(52) DEBUG: rr [loose.c:1006]: loose_route_mode(): There is no Route HF
2(52) DEBUG: tm [t_lookup.c:1053]: t_check_msg(): msg (0xffffa72f7088) id=19/52 global id=18/52 T start=0
2(52) DEBUG: tm [t_lookup.c:497]: t_lookup_request(): start searching: hash=21251, isACK=1
2(52) DEBUG: tm [t_lookup.c:455]: matching_3261(): RFC3261 transaction matching failed - via branch [z9hG4bKSG.c52861b7-2535-4080-84f5-2819c4169843]
2(52) DEBUG: tm [t_lookup.c:675]: t_lookup_request(): no transaction found

This makes sense intuitively; the auth_challenge(), and resulting 407 challenge, should have ended the old transaction, so the negative ACK should just be absorbed.

But in that case, why does the 407 keep being retransmitted?

-- Alex


On Dec 15, 2022, at 12:00 PM, Alex Balashov <abalashov@evaristesys.com> wrote:

Hi Henning,


On Dec 15, 2022, at 11:51 AM, Henning Westerholt <hw@gilawa.com> wrote:

Hi Alex,
it might not help you much, but recently I was implementing a similar structure in one larger migration project, and it seems to work fine.
I am not using any special flags for the challenge etc..
It’s basically like this (pseudo-code)
route{
if no auth user -> auth_challenge()
else -> http_async_query(API, AUTH)
}
route[AUTH] {
get API result for password
if API failure -> auth_challenge()
else -> pv_auth_check(..)
route(next steps)
}


Yeah, that's more or less what I've got, except the first part.

I don't auth_challenge() every request because some requests are allowed by static IP, and I don't know whether to auth_challenge() them unless I am already in the async resume context.

I have eliminated the independent credentials query. At this point my process is more:

request_route {
  ...

  http_async_query(API, RESUME)
}

route[RESUME] {
  if(method == "INVITE") {
     if(has_auth_attrib()) {
        if(!pv_auth_check(...)) {
           auth_challenge("realm", "1");
           exit;
        }
     }

     # Get more routing info.
     http_async_query(API, RESUME2)
     return;
  }
}

route[RESUME2} {
  t_relay() etc
}

-- Alex

--
Alex Balashov | Principal | Evariste Systems LLC

Tel: +1-706-510-6800 / +1-800-250-5920 (toll-free)
Web: http://www.evaristesys.com/, http://www.csrpswitch.com/


--
Alex Balashov | Principal | Evariste Systems LLC

Tel: +1-706-510-6800 / +1-800-250-5920 (toll-free)
Web: http://www.evaristesys.com/, http://www.csrpswitch.com/


--
Alex Balashov | Principal | Evariste Systems LLC

Tel: +1-706-510-6800 / +1-800-250-5920 (toll-free)
Web: http://www.evaristesys.com/, http://www.csrpswitch.com/


--
Alex Balashov | Principal | Evariste Systems LLC

Tel: +1-706-510-6800 / +1-800-250-5920 (toll-free)
Web: http://www.evaristesys.com/, http://www.csrpswitch.com/


-- 
Alex Balashov | Principal | Evariste Systems LLC

Tel: +1-706-510-6800 / +1-800-250-5920 (toll-free)
Web: http://www.evaristesys.com/http://www.csrpswitch.com/

 

-- 
Alex Balashov | Principal | Evariste Systems LLC

Tel: +1-706-510-6800 / +1-800-250-5920 (toll-free)
Web: http://www.evaristesys.com/http://www.csrpswitch.com/