I have just tried to use `uac_replace_from()` (which I wasn't familiar with - we haven't used the uac module in our setup up till now, and from its description I didn't think it would be appropriate as we are not initiating sessions from Kamailio).

We the following code in place, instead of the `remove_hf()` and `append_hf()`:

---8<--- uac_replace_from("sip:$var(from_user)@$var(from_domain)"); ---8<---

The headers are no longer broken, but do not look OK - I still have another copy of the $fU stuck somewhere in the middle. The `From` header no looks as thus:

---8<--- From: sip:972515340014@example.com972515340014;tag=as0c32fd2e ---8<---

Any suggestions?

On Sun, May 26, 2019 at 7:10 PM Karsten Horsmann khorsmann@gmail.com wrote:

Hi Oded,

I would better use uac_replace_from like described in the Kamailio wiki. (link below) This function takes also care of rewriting it back AFAIK.

http://www.kamailio.org/wiki/tutorials/faq/main#how_to_set_different_header_...

Cheers Karsten Horsmann

Oded Arbel odeda@cloudonix.io schrieb am So., 26. Mai 2019, 17:02:

...

I'm trying to rewrite the `From` header in an invite going out to an external system (*), by calling remove_hf("From"); followed by append_hf("From: <some value>\r\n");

but the result is an invalid SIP message, because apparently after `remove_hf()` finishes, it leaves the content of $fU in place of the `From` header - just the raw text, without a header name or even a "\r\n" - and that eats the next header line and breaks the INVITE.

Here is how the INVITE originally looks (without either `remove_hf()` or `append_hf()`):

---8<--- INVITE sip:972522353283@10.20.30.40:5060 SIP/2.0 Record-Route: sip:10.90.80.70;lr;did=649.f0b Via: SIP/2.0/UDP 10.90.80.70:5060;branch=z9hG4bK6188.ff5c4fd855bd2970fdf32d4807b114c2.0 Via: SIP/2.0/UDP 10.90.80.71:5066;received=10.90.80.71;branch=z9hG4bK581897a4;rport=5066 Max-Forwards: 69 From: sip:972515340014@example.com:5066;tag=as0bcee217 To: sip:972522353283@10.20.30.40:5060 Contact: sip:972515340014@10.90.80.71:5066 Call-ID: 722299dc15aad53d28172d433e8665f4@10.90.80.71:5066 CSeq: 102 INVITE User-Agent: Asterisk PBX 14.7.2 Date: Sun, 26 May 2019 14:42:52 GMT Allow: INVITE, ACK, CANCEL, OPTIONS, BYE, REFER, SUBSCRIBE, NOTIFY, INFO, PUBLISH, MESSAGE Supported: replaces X-CID: ad199638-4df3-4aef-81f1-1b50581032a7 Content-Type: application/sdp Content-Length: 280

v=0 o=root 817636675 817636675 IN IP4 10.50.60.70 s=Asterisk PBX 14.7.2 c=IN IP4 10.50.60.70 t=0 0 m=audio 39064 RTP/AVP 8 0 18 9 a=maxptime:60 a=rtpmap:8 PCMA/8000 a=rtpmap:0 PCMU/8000 a=rtpmap:18 G729/8000 a=rtpmap:9 G722/8000 a=sendrecv a=rtcp:39065 a=ptime:20 ---8<---

and here is another call with the same setup, with just `remove_hf("From");` being invoked (without calling `append_hf()` - this is just for reproduction):

---8<--- INVITE sip:972522353283@10.20.30.40:5060 SIP/2.0 Record-Route: sip:10.90.80.70;lr;did=f45.91b1 Via: SIP/2.0/UDP 10.90.80.70:5060;branch=z9hG4bK024d.2690ebf7a809edca2208e6b2561ccf9e.0 Via: SIP/2.0/UDP 10.90.80.71:5066;received=10.90.80.71;branch=z9hG4bK79ae3073;rport=5066 Max-Forwards: 69 972515340014To: sip:972522353283@10.20.30.40:5060 Contact: sip:972515340014@10.90.80.71:5066 Call-ID: 7b20ba5a549f6f886c22985146765736@10.90.80.71:5066 CSeq: 102 INVITE User-Agent: Asterisk PBX 14.7.2 Date: Sun, 26 May 2019 14:44:20 GMT Allow: INVITE, ACK, CANCEL, OPTIONS, BYE, REFER, SUBSCRIBE, NOTIFY, INFO, PUBLISH, MESSAGE Supported: replaces X-CID: ffbb4ac1-3a0f-4dd6-9ce9-7ad24361081a Content-Type: application/sdp Content-Length: 280

v=0 o=root 509740923 509740923 IN IP4 10.50.60.70 s=Asterisk PBX 14.7.2 c=IN IP4 10.50.60.70 t=0 0 m=audio 39102 RTP/AVP 8 0 18 9 a=maxptime:60 a=rtpmap:8 PCMA/8000 a=rtpmap:0 PCMU/8000 a=rtpmap:18 G729/8000 a=rtpmap:9 G722/8000 a=sendrecv a=rtcp:39103 a=ptime:20 ---8<---

As you can see, the `From` header is indeed gone, but it left the $fU just before the `To` header.

WTF?

I'm running Kamailio 5.2.2 built locally.

Notes: *) the problem I'm trying to solve is that I can't get my Asterisk installation to generate the `From` header I want, the reason is out of scope of this question, so I'm sending custom SIP headers over to Kamaiio and I want to rewrite the header in Kamailio to the "correct" value.

-- Oded Arbel

---

Kamailio (SER) - Users Mailing List sr-users@lists.kamailio.org https://lists.kamailio.org/cgi-bin/mailman/listinfo/sr-users

---

Kamailio (SER) - Users Mailing List sr-users@lists.kamailio.org https://lists.kamailio.org/cgi-bin/mailman/listinfo/sr-users

On Sun, May 26, 2019 at 8:59 PM Alex Balashov abalashov@evaristesys.com wrote:

Make sure you have the right reversion/cookie options set,

I'm not sure what that means...

and that you only call uac_replace_from() once in the message processing loop,

There is only one call in the code to `uac_replace_from()` which is run once (and only for INVITE messages).

and that you don’t modify From any other way besides that concurrently.

Aha! and that was apparently the problem. I have another place in the code (probably from an earlier work trying to do the same thing) that had:

$fU = $hdr(custom-caller-id-header)

Removing that solved the problem. I'm guessing its the same issue that I had with the `remove_hf()` method, that removed `From` but somehow somewhere got the value that was meant to replace $fU written into an arbitrary location in the headers buffer.

Thanks for the support!

...

On May 26, 2019, at 1:52 PM, Oded Arbel odeda@cloudonix.io wrote:

I have just tried to use `uac_replace_from()` (which I wasn't familiar with - we haven't used the uac module in our setup up till now, and from its description I didn't think it would be appropriate as we are not initiating sessions from Kamailio).

We the following code in place, instead of the `remove_hf()` and `append_hf()`:

---8<--- uac_replace_from("sip:$var(from_user)@$var(from_domain)"); ---8<---

The headers are no longer broken, but do not look OK - I still have another copy of the $fU stuck somewhere in the middle. The `From` header no looks as thus:

---8<--- From: sip:972515340014@example.com972515340014;tag=as0c32fd2e ---8<---

Hello Alex,

just a small comment:

If you use an appropriate db_mode for the dialog module storage then this mode is also restart-persistent. I did some improvements here recently, the fix was also back-ported.

Cheers,

Henning

Am 27.05.19 um 01:01 schrieb Alex Balashov:

On Sun, May 26, 2019 at 09:27:01PM +0300, Oded Arbel wrote:

...

On Sun, May 26, 2019 at 8:59 PM Alex Balashov abalashov@evaristesys.com wrote:

...

Make sure you have the right reversion/cookie options set,

I'm not sure what that means...

Sorry, that wasn't a very clear response on my part. I was referring to these parameters:

https://kamailio.org/docs/modules/5.2.x/modules/uac.html#uac.p.restore_mode https://kamailio.org/docs/modules/5.2.x/modules/uac.html#uac.p.restore_dlg

The way the uac_replace_{to,from}() mechanism works is that given an INVITE going from A to B, and passing through proxy P, P will modify the header in flight and store the original value in some fashion. Taking advantage of its role in series to the path of any replies or new in-dialog requests going from B back to A, P will statefully revert the header value to the stored original, so that A is none the wiser as to the fact that the header has been modified. Meanwhile, B will continue to see the modified value at all times. This removes the problem of the proxy P modifying these values en route in a way that gives rise to an rejection from A -- "Hey, I didn't send that!"

The settings in this area mostly relate to the way in which this original value is stored. There are two ways. One is to throw it into a Record-Route header parameter, taking advantage of the proxy's role (if record_route() is called/an RR header is added upon processing the initial INVITE) in the entire life cycle of the dialog. The proxy will intercept the original value in the RR header (encoded or encrypted somehow) and revert it in messages going back to "A".

The other way is to use the dialog module's runtime memory of dialogs, if enabled. This method is perhaps a bit more sanitary, but has the disadvantage of not being particularly stateless or restart-persistent.

-- Alex