MMS service down - SMPP failure

MMS service down - SMPP failure SearchSearch
Author Message
Antoine Martin
New member
Username: Anmartin

Post Number: 1
Registered: 04-2015
Posted on Thursday, April 09, 2015 - 05:05 pm:   

Hello,

our MMS service is currently down, after the following events :

1) Initially we had a complaint from a third-party operator which was not receiving MMS from our network. When checking on NowSMS/MMS Gateway, there was this alarm :

"MMSC Route Aicent Orange : MM4 - unable to connect to server xx.xx.xx.xx"


2) To solve the problem we tried first to restart MMSC service via a Stop then Start. => no change, alarm appeared again


3) We continued checking and decided to restart MMSC Service as well as SMS Gateway Service, via a Stop then Start on both services.
It did not solve the issue, but unfortunately the situation became worst as we lost the connectivity to SMPP gateway and the following alarm appeared :

"SMPP Failure -- Unable to connect to SMPP server at xx.xx.xx.xx"

On our SMPP gateway we noticed that the bind was disconnected, but then there is no more tentative of reconnection from the NowSMS/MMS Gateway...



Could you help us to check how we can investigate further on the problem ? and why the SMPP bind is not coming up again ?


thanks for your support!

Antoine
Des - NowSMS Support
Board Administrator
Username: Desosms

Post Number: 5253
Registered: 08-2008
Posted on Thursday, April 09, 2015 - 05:16 pm:   

Hi Antoine,

The first thing I would suggest, since it sounds like a basic connectivity issue, I'd suggest rebooting the server OS.

What is the MMSC version?

--
Des
NowSMS Support
Antoine Martin
New member
Username: Anmartin

Post Number: 2
Registered: 04-2015
Posted on Thursday, April 09, 2015 - 07:05 pm:   

Hi Des,

thanks for the quick reply, the MMSC version is v2014.06.30

we are checking if any connectivity issue, but ping are OK to both addresses in alarms: SMPP GW and third-party MM4 server


Is there not any other solution to re-launch the service properly wihtout rebooting ?


thanks and regards,
Antoine
Des - NowSMS Support
Board Administrator
Username: Desosms

Post Number: 5254
Registered: 08-2008
Posted on Thursday, April 09, 2015 - 07:24 pm:   

Hi Antoine,

Ping may be ok...but something seems very wrong with TCP/IP connectivity, as the connection attempts are failing.

You could try telnet to verify that you can connect from the NowSMS Server to the SMPP server, but I am expecting that will also fail. My best guess is that OS resources are constrained for some reason, which is why I suggest a full reboot.

I've gone back through some notes, and this sounds like s time where a virus scanner running on an MMSC had gotten overwhelmed and consumed system resources.

--
Des
Antoine Martin
New member
Username: Anmartin

Post Number: 3
Registered: 04-2015
Posted on Thursday, April 09, 2015 - 08:22 pm:   

Hi Des,

yes the reboot solved the issue. Actually, it seems that the server was not able to open/establish new IP:port connections.

For example, when I tried to open a ftp connection from the MMSC server to an external FTP, I got this error :
ftp : bind : no buffer space is supported

So your guess is right, there are some OS resources shortage, do you have any suggestion to avoid this kind of issues ? (apart from a regular reboot :) )


thanks again for your support,
Antoine
Des - NowSMS Support
Board Administrator
Username: Desosms

Post Number: 5255
Registered: 08-2008
Posted on Thursday, April 09, 2015 - 09:30 pm:   

Hi Antoine,

I'm glad to hear the problem was resolved quickly.

There is no reason to schedule a periodic server reboot, unless there is a recurring issue that cannot be resolved.

My advice would be to start monitoring resource utilization periodically in Windows task manager. In particular look for processes with excessive memory usage and/or handle/object counts.

What we have seen in the past is processes leaking handles...I'm talking thousands/tens of thousands...large numbers...not just a few. The biggest culprit is usually virus scanners, especially those with real time scanning. They are not designed to run on servers and get overwhelmed by large numbers of connections. Sometimes the handle leaks appear in the smsgws.exe or mmsc.exe process, but go away if real time scanning is disabled.

--
Des
NowSMS Support