Problem with inbound MMS
Jim Udall
There appears to be a problem with inbound MMS – which arose fairly recently.
I’ve diagnosed it back to a change in the MMSC from Fido that delivers MMS messages to us.
Specifically when an MMS is delivered to our MMS gateway, it provides information about what number (i.e. longcode) that message is addressed to. Whereas previously this was a 10 digit North American number, then have now added the country code of ‘1’ to the front of that number.
In the platform, routing of messages is based in part on the longcode on which it was received. This routing is now receiving an 11 digit number and tried to find a route based on a 10 digit number.
As a consequent, the MMS is received, but it is placed in a black hole because now application route can be found.
I see the SAT folks are attempting furiously to send a ton of MMS messages – which are making it to the gateway – but unfortunately not to the application.
I will be trying to correct this in the next hour or so and advise when I have a solution to the problem
Jim
Jim Udall
CTO
QuickMobile
jim....@quickmobile.com
m: 604.716,1523
o: 604.875.0403
f: 604.875.0603
www.quickmobile.com
Jim Udall
I believe I have a fix for this problem.
However, it seems that only MMS ‘s originating from the Rogers network have the country code prefixed to the number. I don’t have any rogers phones to test with.
Can someone with Rogers service please give it a shot?
Jim
No virus
found in this incoming message.
Checked by AVG - www.avg.com
Version: 8.0.233 / Virus Database: 270.10.19/1939 - Release Date: 02/05/09
11:34:00
Glenn Iles
I have a rogers service - what number would you like me to MMS something to?
Best regards,
Glenn Iles
604.966.7474
Jim Udall
778-320-6873….. It really doesn’t matter for this test, but if you want, you could put ‘jim’ in the subject line
Jim
Glenn Iles
778-320-6873….. It really doesn't matter for this test, but if you want, you could put 'jim' in the subject line
Jim
From: muse...@googlegroups.com [mailto:muse...@googlegroups.com] On Behalf Of Glenn Iles
Sent: Saturday, February 07, 2009 3:38 PM
To: muse...@googlegroups.com
Subject: Re: Problem with inbound MMS
Hello Jim,
Glenn Iles
Jim Udall
Thanks – sort of J…. I received your MMS. However, it didn’t come in with the country code prefix of ‘1’ – so it got routed OK.
On the other hand…. I just HAPPENED to have in the last couple of hours received some spam SMS from Orange. And they included the country code – which my patch successfully matched to the 10 digit longcode – and routed it correctly. In other words: my fix works.
However, it’s peculiar that in previous logs, I say a number of MMS arrive with the country code include – and I did verify that the MMS originated from a Rogers subscriber.
In any event, I believe the patch is fine – and MMS inbound should be working properly again.
Thanks again Glenn
Maria
north american SMS come in with a prefix of '1' and all my NA MMS come
in with no prefix to the 10 digit number. Would be good to be
consistent for NA numbers just to ease array searches and the like.
I've got European numbers too so any consistencies across services
would be great.
On Feb 7, 5:21 pm, "Jim Udall" <jim.ud...@quickmobile.com> wrote:
> Thanks - sort of J.. I received your MMS. However, it didn't come in with
> the country code prefix of '1' - so it got routed OK.
>
> On the other hand.. I just HAPPENED to have in the last couple of hours
> received some spam SMS from Orange. And they included the country code -
> which my patch successfully matched to the 10 digit longcode - and routed it
>
> However, it's peculiar that in previous logs, I say a number of MMS arrive
> from a Rogers subscriber.
>
> In any event, I believe the patch is fine - and MMS inbound should be
>
> Thanks again Glenn
>
> Jim
>
> From: muse...@googlegroups.com [mailto:muse...@googlegroups.com] On
> Behalf Of Glenn Iles
> Sent: Saturday, February 07, 2009 4:41 PM
> To: muse...@googlegroups.com
> Subject: Re: Problem with inbound MMS
>
> Just sent it.
>
> Best regards,
>
> Glenn Iles
> 604.966.7474
>
>
> 778-320-6873... It really doesn't matter for this test, but if you want,
>
> Jim
>
> From: muse...@googlegroups.com [mailto:muse...@googlegroups.com] On
> Behalf Of Glenn Iles
> Sent: Saturday, February 07, 2009 3:38 PM
> To: muse...@googlegroups.com
> Subject: Re: Problem with inbound MMS
>
> Hello Jim,
>
> I have a rogers service - what number would you like me to MMS something to?
>
> Best regards,
>
> Glenn Iles
> 604.966.7474
>
>
> I believe I have a fix for this problem.
>
> However, it seems that only MMS 's originating from the Rogers network have
> the country code prefixed to the number. I don't have any rogers phones to
> test with.
>
> Can someone with Rogers service please give it a shot?
>
> Jim
>
> From: muse3-p...@googlegroups.com
> [mailto:muse3-p...@googlegroups.com] On Behalf Of Jim Udall
> Sent: Saturday, February 07, 2009 11:59 AM
> To: muse3-p...@googlegroups.com; muse...@googlegroups.com
> Subject: Problem with inbound MMS
>
>
> I've diagnosed it back to a change in the MMSC from Fido that delivers MMS
> messages to us.
>
> Specifically when an MMS is delivered to our MMS gateway, it provides
> information about what number (i.e. longcode) that message is addressed to.
> Whereas previously this was a 10 digit North American number, then have now
> added the country code of '1' to the front of that number.
>
> In the platform, routing of messages is based in part on the longcode on
> which it was received. This routing is now receiving an 11 digit number and
> tried to find a route based on a 10 digit number.
>
> As a consequent, the MMS is received, but it is placed in a black hole
> because now application route can be found.
>
> which are making it to the gateway - but unfortunately not to the
>
> I will be trying to correct this in the next hour or so and advise when I
> have a solution to the problem
>
> Jim
>
> Jim Udall
> CTO
> QuickMobile
Jim Udall
field) SOMETIMES have 10 digits and SOMETIMES have 11. I noticed this when
someone complained about not receiving MMS messages. This was a serious
issue for the platform since I use that TO number to find the appropriate
interface and callback handler. I have a bit of a patch to overcome this
deep within the MMS driver. However, I don't do anything with the FROM
field.
As an application, you can resolve these differences by invoking the method
'CleanPhoneList' on the number and a service point. This will give you a
consistent representation.
Jim
15:36:00
Jim Udall
I have had the support of the MMS stack vendor on the outbound MMS issue,
but I'm afraid they were of little help. It appears that the stack is
working correctly. For those interested in details, read the following
explanation:
When an MMS is sent, it's actually pretty simple. An http connection is
setup to Fido's MMSC. Once established, a single HTTP POST is sent to the
MMSC. The body of the post contains the actual data (say an image).
The problem (and it can be clearly seen) is that the HTTP POST goes out, but
the http server (in this case Fido's MMSC), closes the persistent TCP
connection used to carry the post. The stack then retries with another
persistent connection. And again the MMSC closes the persistent TCP
connection. After a few such tries, the MMS stack just gives up.
The question is: "Why is the Fido MMSC closing the persistent connection?"
The POST itself is fairly simple. Other than the data, there is not much
other stuff sent. Specifically there are only 2 HTTP headers. One contains
the WAP profile for the sending device and the other contains the User Agent
type (e.g. Firefox browser, IE browser). I thought perhaps he didn't like
the WAP profile I was sending, so changed it to a bonafide Fido device WAP
profile. This made not difference. So I suppose that theory doesn't wash.
I then have an issue with the modem used for MMS in that they don't support
hardware flow control. It's possible I could be overrunning the modem since
the UART rate is faster than the GPRS data rate. So I throttled back the
UART rate to be less than the GPRS data rate. However, the result was the
same.
The only thing I have left is the change the User Agent HTTP header -
however, I'm unable to do that not having the source code.
END OF LONG EXPLANATION----------------------------
So at this point I'm stumped. I believe Fido has made some changes to the
MMSC implementation over the past year or so. This would explain two
things:
1. Why most Nokia devices (even those bona fide Fido Nokia devices) don't
seem to be able to reliably send MMS anymore
2. Why for some reason, my MMS gateway can no longer sent MMS outbound when
over a year ago, this was not a problem.
In short, I'm stumped.
Jim
Sent: Saturday, February 21, 2009 12:40 AM
To: muse3 dev
Maria Lantin
Jim
Behalf Of Maria
Sent: Saturday, February 21, 2009 12:40 AM
To: muse3 dev
Jim Udall
Thanks but alas I’m well aware of those parameters. Actually 205.151.11.11 is supposed to be for WAP based MMS and 205.141.11.13 is supposed to be for HTTP base MMS. But it doesn’t matter. I’ve tried both those address with the same result.
Head scratching continues L
Jim
Maria Lantin
Any new clues?
Thanks but alas I’m well aware of those parameters. Actually 205.151.11.11 is supposed to be for WAP based MMS and 205.141.11.13 is supposed to be for HTTP base MMS. But it doesn’t matter. I’ve tried both those address with the same result.
Head scratching continues L
Jim
Roland Tanglao
Fido (which of course is now really Rogers) unlocked European Nokia
E71 to a SIFT Channel this morning. The number I sent to was:
778 320 6673
Maria Lantin
sending to Muse works well. It's the sending from Muse that is stuck. The subject of this thread has somehow migrated from it's original "inbound" problem.
Unfortunately, the alternative of sending an SMS with a link is not quite functional enough for what I need. Just because some phones don't even see it as a link so you have to remember the URL as you try to bring up a web browser...and that's assuming that you have a data plan and don't mind browsing on the phone. For some, this is just to big a barrier.
M!
Jim Udall
I would say the probability is zero. I spent some time over the weekend looking at it – trying a number of hypotheses and having no luck.
I’m swamped with Quick Mobile work so I really don’t have any cycles to devote to it.
Sorry!
Version: 8.0.237 / Virus Database: 270.11.3/1971 - Release Date: 02/26/09 14:51:00
Clay Howey
Hi Maria,
Regarding sending a link via SMS, something worth trying is sending *only* the link. In the past, we found that if the link is accompanied by text, the link isn't handled correctly by some microbrowsers.
Cheers, Clay
Maria Lantin <maria....@gmail.com>
Sent by: muse...@googlegroups.com Feb 26, 2009 01:20 PM
|
|
|
|
Maria Lantin
Jim Udall
That would be my (i.e. the platform’s fault). I’m inserting that. It would be a simple change to remove it if the text string is null.
Although to be truthful, I’m skeptical that there are any phones out there today that CAN’T parse out a URL in any SMS.
Jim
Maria Lantin
That would be my (i.e. the platform’s fault). I’m inserting that. It would be a simple change to remove it if the text string is null.
Although to be truthful, I’m skeptical that there are any phones out there today that CAN’T parse out a URL in any SMS.
Jim
From: muse...@googlegroups.com [mailto:muse...@googlegroups.com] On Behalf Of Maria Lantin
Sent: Thursday, February 26, 2009 4:08 PM
Subject: Re: Problem with inbound MMS
Jim Udall
Ok – I’ve changed the code. If you give a text string of null or “”, then I won’t put the “:” character in the string
Maria Lantin
Ok – I’ve changed the code. If you give a text string of null or “”, then I won’t put the “:” character in the string
Jim
Jim Udall
I’d be stunned if what you say is true about the LG. What model is it? Sometimes these things aren’t obvious
Version: 8.0.237 / Virus Database: 270.11.4/1976 - Release Date: 02/27/09 13:27:00
Maria Lantin
I’d be stunned if what you say is true about the LG. What model is it? Sometimes these things aren’t obvious
Jim
Maria Lantin
We're periodically getting no response from the callback when we send an MMS to the 7174 number. Most of the time it works but when it doesn't there is nothing in our log file to suggest that the callback even got called. The picture shows up on the fido picture messaging site but the callback does not get called. Is there any way you can check whether there is an issue with the callback address or something else? An image was sent about 3 hours ago and no response. We're just about to resend.
Thanks for any clues you have access to!
M!
Jim Udall
If you give me the phone number(s) from where you’re trying to send and the approximate times they were sent – I can correlate them to received messages. I see many messages coming in on that interface – and no obvious dropping of the message anywhere
Version: 8.0.237 / Virus Database: 270.11.5/1979 - Release Date: 03/03/09 07:25:00