Showing results for 
Search instead for 
Did you mean: 

Welcome to the Cisco Small Business Community

Have a question? Click on a topic board below to get started in the community.


SPA122 - Memory leak after receiving CANCEL

I have hundreds of units in the field with version 1.3.3_015.


I was troubleshooting an issue where the ATA was showing all LEDs green, Phone1 and Phone2 being steady but when picking up a phone, no dialtone was present. A reboot fixed it but since I had syslog turned on, I noticed the MemFree to be quite low (772 KB) just before the unit stopped sending stuff to us.

So I have plotted the MemFree as much as I could. It was already low (in the 1700 KB) but that unit had been running for quite some time. At some point, the MemFree dropped to 1000 KB right after a call came in.


The pcap of the call showed an INVITE sent to the unit, it responded with the 100 and 180, then 3.19 seconds later we send a CANCEL, another CANCEL at 4.19 seconds and got a 487 at 4.80 seconds.


Before that call, MemFree was 1764 Kb. After this event, MemFree dropped to 1012 Kb so a drop of 752 bytes. This is about the size of the two CANCEL requests ...


Anyone noticed this memory leak issue?



Benoit Raymond

TelSip Communication



Dan Lukes

"CANCEL before final reply" processing had known to be buggy. But it has been corrected, if I remember correctly. It has been mentioned in release notes, but I don't remember the exact version. All at all, you should consider to upgrade to more recent version as there are other issues with 1.3.3 as well.


Rate useful response. It will help others to found solutions.

Thanks Dan for the prompt response.

I did check the release note when 1.3.5 was released but it does not mention this "Cancel before final reply" issue.

But that does not surprise me as I believe Cisco is sharing code among multiple devices so it could have been reported against another device.

I have upgraded couple of ATA onto 1.3.5 and I have a watcher monitoring the MemFree values. One good thing is that the units report it to syslog so quite easy to monitor and alarm when the come to a critical level.

Well, I'm not sure - may be I read it in release notes of SPA50x line.

If you consider the upgrade solved the issue, mark the apropriate answer as "correct". It will help others to found solutions.