MM4 FWD-RES not received during peak time

MM4 FWD-RES not received during peak time SearchSearch
Author Message
David
New member
Username: Legolas23

Post Number: 12
Registered: 03-2011
Posted on Wednesday, September 28, 2011 - 03:28 pm:   

Hello support,

I come to you about an issue observed on the MM4-FWD-RES sent by NOWMMS Plaftorm during the peak time.

Description :

During the peak time (4 pm to midnight), the NOWMMS returns the FWD-RES with a long time.
But at off peak time, normally, it replies FWD-RES back to the broker no longer than 5 minutes.

The MMSCACK directory is empty. There is no files

Trace example :

The following FWD-REQ was delivered to NOWMMSC at Mon, 26 Sep 2011 14:38:39 -0700 (PDT). But it sent the FWD-RES at Mon, 26 Sep 2011 20:31:14 -0700 (PDT).
In this case, it took almost 6 hours to send FWD-RES back.
You will find a graph displaying in attachment this behavior.

Question :

We didn't have this delay before. Do you have an explanation?

image/bmp
traffic MMS.bmp (1399.7 k)
Des - NowSMS Support
Board Administrator
Username: Desosms

Post Number: 3493
Registered: 08-2008
Posted on Wednesday, September 28, 2011 - 03:39 pm:   

Hi David,

What version?

If the MMSCACK directory was empty during this time, then this indicates that an ACK route was not explicitly defined for the inbound route on which this message was received.

Check the "MMSC VASP" definitions to make sure an MM4 Ack Route is defined.

--
Des
NowSMS Support
David
New member
Username: Legolas23

Post Number: 13
Registered: 03-2011
Posted on Wednesday, September 28, 2011 - 10:30 pm:   

Hello,

The NOWMMS version is v2010.11.26.

About the MMSCACK directory,sorry for the misunderstanding but there are some files.

For example, in watching now at 5.23 pm, i have 552 RFC file in the directory. I have some files dating from 1 pm still in the MMSACK directory.

I confirm you that the MMSCVASP has been created in order to drain off the FWD-RES to the broker.
However, i can see sometines some connections loss between NOWMMS and the broker. These disconnections are visible on the software but the telnet test from NOWMMS to the SMTP server is ok.

Do you know why we have some delays during the drain off.

Thanks for your feedback
Regards
David
New member
Username: Legolas23

Post Number: 14
Registered: 03-2011
Posted on Thursday, September 29, 2011 - 02:49 pm:   

Hello,

An additional request, how long does NowMMS resend the message stored in the queue? Like retry every 30 mins, 1 hour or 2 hours,etc...

Thanks for your help
Regards
Des - NowSMS Support
Board Administrator
Username: Desosms

Post Number: 3495
Registered: 08-2008
Posted on Thursday, September 29, 2011 - 05:14 pm:   

Hi David,

Sorry for the delay in response, I had to go back and review one of the change notes for a 2011 release.

This is what it said:


quote:

* MMSC: Reduce DNS query timeouts that can backlog the processing of MM4 acknowledgments, especially when the "MM4 Ack Route" has not been set for a connection.




I needed to better understand this, and I think I can explain it now.

Take a look at a sampling of these RFC files in the MMSCACK directory. They are basically text files with queued MM4/SMTP acknowledgments.

If there is no "X-MMSC-MM4-Route-Info:" header in the message file, this will cause problems.

It means that the message requested an acknowledgment, but the "MM4 Ack Route" could not be determined.

If you want to e-mail me a ZIP file with the contents of the directory, I can look at it in more detail.

If you install the 2011 release, it won't generate acks if it does not know how to route them back. If there are some previously generated ones stuck in the queue, it won't spend as much time on them triggering this type of backlog.

So the update would probably help, but it is important to determine what these acks that are missing the header are being generated in response to, so that they can be routed properly.

--
Des
NowSMS Support
Des - NowSMS Support
Board Administrator
Username: Desosms

Post Number: 3496
Registered: 08-2008
Posted on Thursday, September 29, 2011 - 05:17 pm:   

Regarding the resend question, there is a timed delay between resends, waiting 1 minute after each failure. But the problem is that the DNS query timeouts are excessive ... 4 to 6 minutes per message ... and the processing of the ACKs is single threaded, because of the expectation that the routing is explicitly defined.
Des - NowSMS Support
Board Administrator
Username: Desosms

Post Number: 3538
Registered: 08-2008
Posted on Tuesday, October 18, 2011 - 08:38 pm:   

A belated follow-up for others who might search and find this thread in the future.

Everything was configured correctly.

In this case, the problem was not caused by the ACK Route not being configured, it was caused by a corrupt MMSC user database.

The following article was written to provide more information about recovering from this type of problem:

http://www.nowsms.com/repairing-a-corrupt-mmsc-user-database


--
Des
NowSMS Support