Hi all,
a customer of ours are having some trouble with the Content-Lenght header when a NOTIFY-message goes through our OpenSER 1.2.0.
Here is what seem to happend.
SIP NOTIFY from our PSTN-gateway to our OpenSER: --- CUT --- Content-Length: 21. . SIP/2.0 100 Trying. . --- CUT ---
SIP NOTIFY from our OpenSER to Customer: --- CUT --- Content-Length: 21. . SIP/2.0 100 Trying. --- CUT ---
Now, the customer has to manually do this in his OpenSER to make the UA on his side happy: if (method=="NOTIFY") { if (search_body("SIP/2.0 100 Trying")) { xlog("L_WARN", "R: Setting Content-Length to 20\n"); remove_hf("Content-Length"); append_hf("Content-Length: 20\r\n"); } }
Please advice on what we are doing wrong here?
Br, /Tobias
Hi,
Please advice on what we are doing wrong here?
What is the actual content length? If the GW sets it to 21 and the UA is only happy with 20, one of them has to be right. So it's either a bug in the GW's or UA's length calculation (and should be fixed there, not using OpenSER).
Andreas
Hello,
from what I can see (ngrep) it looks like the content-lenght is 21 (if you count the characters), but traversing my OpenSER it seems that one character is removed (a dot?) and the content-lenght then should be 20?
Br, /Tobias
Andreas Granig said the following on 2007-05-23 11:12:
Hi,
Please advice on what we are doing wrong here?
What is the actual content length? If the GW sets it to 21 and the UA is only happy with 20, one of them has to be right. So it's either a bug in the GW's or UA's length calculation (and should be fixed there, not using OpenSER).
Andreas
from what I can see (ngrep) it looks like the content-lenght is 21 (if you count the characters), but traversing my OpenSER it seems that one character is removed (a dot?) and the content-lenght then should be 20?
A dot in ngrep is I think a "non printable" character like newline/linefeed (\r, \n).
Use ethereal/wireshark to find out the real length of the content. It looks like 21 is correct. Maybe the receiver does not like the single CR at the end of the content.
regards klaus
Tobias Lindgren wrote:
Hello,
from what I can see (ngrep) it looks like the content-lenght is 21 (if you count the characters), but traversing my OpenSER it seems that one character is removed (a dot?) and the content-lenght then should be 20?
Br, /Tobias
Andreas Granig said the following on 2007-05-23 11:12:
Hi,
Please advice on what we are doing wrong here?
What is the actual content length? If the GW sets it to 21 and the UA is only happy with 20, one of them has to be right. So it's either a bug in the GW's or UA's length calculation (and should be fixed there, not using OpenSER).
Andreas
Users mailing list Users@openser.org http://openser.org/cgi-bin/mailman/listinfo/users
Hi all,
thanks for the hint. I used Wireshark now and it reports "21 bytes" on the message body on my package coming into my OpenSER and "20 bytes" on the package going out from OpenSER. However "content-lenght" header is still the same, "21".
Br, /Tobias
Klaus Darilion said the following on 2007-05-23 12:12:
Use ethereal/wireshark to find out the real length of the content. It looks like 21 is correct. Maybe the receiver does not like the single CR at the end of the content.
regards klaus
Tobias Lindgren wrote:
Hello,
from what I can see (ngrep) it looks like the content-lenght is 21 (if you count the characters), but traversing my OpenSER it seems that one character is removed (a dot?) and the content-lenght then should be 20?
Br, /Tobias
Andreas Granig said the following on 2007-05-23 11:12:
Hi,
Please advice on what we are doing wrong here?
What is the actual content length? If the GW sets it to 21 and the UA is only happy with 20, one of them has to be right. So it's either a bug in the GW's or UA's length calculation (and should be fixed there, not using OpenSER).
Andreas
Users mailing list Users@openser.org http://openser.org/cgi-bin/mailman/listinfo/users
I guess there is a CR at the end of the packet - correct?
I'm not sure about this, but as the CR is part of the content I think openser should forward it too - thus leave the content untouched.
Thus, maybe its a bug in openser.
regards klaus
Tobias Lindgren wrote:
Hi all,
thanks for the hint. I used Wireshark now and it reports "21 bytes" on the message body on my package coming into my OpenSER and "20 bytes" on the package going out from OpenSER. However "content-lenght" header is still the same, "21".
Br, /Tobias
Klaus Darilion said the following on 2007-05-23 12:12:
Use ethereal/wireshark to find out the real length of the content. It looks like 21 is correct. Maybe the receiver does not like the single CR at the end of the content.
regards klaus
Tobias Lindgren wrote:
Hello,
from what I can see (ngrep) it looks like the content-lenght is 21 (if you count the characters), but traversing my OpenSER it seems that one character is removed (a dot?) and the content-lenght then should be 20?
Br, /Tobias
Andreas Granig said the following on 2007-05-23 11:12:
Hi,
Please advice on what we are doing wrong here?
What is the actual content length? If the GW sets it to 21 and the UA is only happy with 20, one of them has to be right. So it's either a bug in the GW's or UA's length calculation (and should be fixed there, not using OpenSER).
Andreas
Users mailing list Users@openser.org http://openser.org/cgi-bin/mailman/listinfo/users
Hello,
yes, it seems there is a CR at the end of the incoming NOTIFY packet but not on the outgoing.
For other packets, like INVITE, the message body looks intact after going through openser.
Br, /Tobias
Klaus Darilion said the following on 2007-05-23 13:49:
I guess there is a CR at the end of the packet - correct?
I'm not sure about this, but as the CR is part of the content I think openser should forward it too - thus leave the content untouched.
Thus, maybe its a bug in openser.
regards klaus
Tobias Lindgren wrote:
Hi all,
thanks for the hint. I used Wireshark now and it reports "21 bytes" on the message body on my package coming into my OpenSER and "20 bytes" on the package going out from OpenSER. However "content-lenght" header is still the same, "21".
Br, /Tobias
Klaus Darilion said the following on 2007-05-23 12:12:
Use ethereal/wireshark to find out the real length of the content. It looks like 21 is correct. Maybe the receiver does not like the single CR at the end of the content.
regards klaus
Tobias Lindgren wrote:
Hello,
from what I can see (ngrep) it looks like the content-lenght is 21 (if you count the characters), but traversing my OpenSER it seems that one character is removed (a dot?) and the content-lenght then should be 20?
Br, /Tobias
Andreas Granig said the following on 2007-05-23 11:12:
Hi,
Please advice on what we are doing wrong here?
What is the actual content length? If the GW sets it to 21 and the UA is only happy with 20, one of them has to be right. So it's either a bug in the GW's or UA's length calculation (and should be fixed there, not using OpenSER).
Andreas
Users mailing list Users@openser.org http://openser.org/cgi-bin/mailman/listinfo/users
Maybe Bogdan or Daniel can comment on this, as I do not know how the content is treated during forwarding.
regards klaus
Tobias Lindgren wrote:
Hello,
yes, it seems there is a CR at the end of the incoming NOTIFY packet but not on the outgoing.
For other packets, like INVITE, the message body looks intact after going through openser.
Br, /Tobias
Klaus Darilion said the following on 2007-05-23 13:49:
I guess there is a CR at the end of the packet - correct?
I'm not sure about this, but as the CR is part of the content I think openser should forward it too - thus leave the content untouched.
Thus, maybe its a bug in openser.
regards klaus
Tobias Lindgren wrote:
Hi all,
thanks for the hint. I used Wireshark now and it reports "21 bytes" on the message body on my package coming into my OpenSER and "20 bytes" on the package going out from OpenSER. However "content-lenght" header is still the same, "21".
Br, /Tobias
Klaus Darilion said the following on 2007-05-23 12:12:
Use ethereal/wireshark to find out the real length of the content. It looks like 21 is correct. Maybe the receiver does not like the single CR at the end of the content.
regards klaus
Tobias Lindgren wrote:
Hello,
from what I can see (ngrep) it looks like the content-lenght is 21 (if you count the characters), but traversing my OpenSER it seems that one character is removed (a dot?) and the content-lenght then should be 20?
Br, /Tobias
Andreas Granig said the following on 2007-05-23 11:12:
Hi,
> Please advice on what we are doing wrong here? What is the actual content length? If the GW sets it to 21 and the UA is only happy with 20, one of them has to be right. So it's either a bug in the GW's or UA's length calculation (and should be fixed there, not using OpenSER).
Andreas
Users mailing list Users@openser.org http://openser.org/cgi-bin/mailman/listinfo/users
I submitted this to the buglist, let's wait and see what they say about it.
Thanks all for your help so far.
Br, /Tobias
Klaus Darilion said the following on 2007-05-23 14:11:
Maybe Bogdan or Daniel can comment on this, as I do not know how the content is treated during forwarding.
regards klaus
Tobias Lindgren wrote:
Hello,
yes, it seems there is a CR at the end of the incoming NOTIFY packet but not on the outgoing.
For other packets, like INVITE, the message body looks intact after going through openser.
Br, /Tobias
Klaus Darilion said the following on 2007-05-23 13:49:
I guess there is a CR at the end of the packet - correct?
I'm not sure about this, but as the CR is part of the content I think openser should forward it too - thus leave the content untouched.
Thus, maybe its a bug in openser.
regards klaus
Tobias Lindgren wrote:
Hi all,
thanks for the hint. I used Wireshark now and it reports "21 bytes" on the message body on my package coming into my OpenSER and "20 bytes" on the package going out from OpenSER. However "content-lenght" header is still the same, "21".
Br, /Tobias
Klaus Darilion said the following on 2007-05-23 12:12:
Use ethereal/wireshark to find out the real length of the content. It looks like 21 is correct. Maybe the receiver does not like the single CR at the end of the content.
regards klaus
Tobias Lindgren wrote:
Hello,
from what I can see (ngrep) it looks like the content-lenght is 21 (if you count the characters), but traversing my OpenSER it seems that one character is removed (a dot?) and the content-lenght then should be 20?
Br, /Tobias
Andreas Granig said the following on 2007-05-23 11:12: > Hi, > >> Please advice on what we are doing wrong here? > What is the actual content length? If the GW sets it to 21 and the UA is > only happy with 20, one of them has to be right. So it's either a bug in > the GW's or UA's length calculation (and should be fixed there, not > using OpenSER). > > Andreas _______________________________________________ Users mailing list Users@openser.org http://openser.org/cgi-bin/mailman/listinfo/users
Hi Tobias,
could you sent the inbound and outbound NOTIFY in text and pcap format? I will try to run some tests on it.
Thanks and regards, Bogdan
Tobias Lindgren wrote:
I submitted this to the buglist, let's wait and see what they say about it.
Thanks all for your help so far.
Br, /Tobias
Klaus Darilion said the following on 2007-05-23 14:11:
Maybe Bogdan or Daniel can comment on this, as I do not know how the content is treated during forwarding.
regards klaus
Tobias Lindgren wrote:
Hello,
yes, it seems there is a CR at the end of the incoming NOTIFY packet but not on the outgoing.
For other packets, like INVITE, the message body looks intact after going through openser.
Br, /Tobias
Klaus Darilion said the following on 2007-05-23 13:49:
I guess there is a CR at the end of the packet - correct?
I'm not sure about this, but as the CR is part of the content I think openser should forward it too - thus leave the content untouched.
Thus, maybe its a bug in openser.
regards klaus
Tobias Lindgren wrote:
Hi all,
thanks for the hint. I used Wireshark now and it reports "21 bytes" on the message body on my package coming into my OpenSER and "20 bytes" on the package going out from OpenSER. However "content-lenght" header is still the same, "21".
Br, /Tobias
Klaus Darilion said the following on 2007-05-23 12:12:
Use ethereal/wireshark to find out the real length of the content. It looks like 21 is correct. Maybe the receiver does not like the single CR at the end of the content.
regards klaus
Tobias Lindgren wrote:
> Hello, > > from what I can see (ngrep) it looks like the content-lenght is 21 (if > you count the characters), but traversing my OpenSER it seems that one > character is removed (a dot?) and the content-lenght then should be 20? > > Br, > /Tobias > > Andreas Granig said the following on 2007-05-23 11:12: > >> Hi, >> >> >>> Please advice on what we are doing wrong here? >>> >> What is the actual content length? If the GW sets it to 21 and the UA is >> only happy with 20, one of them has to be right. So it's either a bug in >> the GW's or UA's length calculation (and should be fixed there, not >> using OpenSER). >> >> Andreas >> > _______________________________________________ > Users mailing list > Users@openser.org > http://openser.org/cgi-bin/mailman/listinfo/users >
Users mailing list Users@openser.org http://openser.org/cgi-bin/mailman/listinfo/users
-
Andreas Granig -
Andreas Sikkema -
Bogdan-Andrei Iancu -
Klaus Darilion -
Tobias Lindgren