SUGGESTIONS Wireshark Graph - any clues?

[papachumba]

So, our PBX causes some issues with calls where randomly about once in the conversation the line will cut out for a few seconds. I managed to capture this in wireshark, but have no clue what to do next. According to this the packet was dropped due to jitter buffer.
Is this indicative of network issues or something else?
Also note the Max Delta and Max Jitter in the 2nd screenshot below - first few lines represent the call from screenshot 1... Also Pb? column contains x for most streams, I read this is also indicative of a problem.

 

[atsak]

What kind of firewall are you using?

 

[papachumba]

What kind of firewall are you using?

Cisco asa

 

[papachumba]

This is the typical packet stream from the PBX to the VOIP supplier, note the jitter values going crazy:

 

[atsak]

It could be an issue at the ISP or it could be the firewall being stuck for a moment (ie busy with something else). Which ISP?

 

[papachumba]

Backbone Connect theyre called. 15mb up/down
However I have been scouring the net and keep coming across articles talking about asterisk timing specified in the kernel. My one is currently set to tsc. Some post I came across mentions trying to switch to hpet instead http://pbxinaflash.com/community/in...fers-attended-works-blind-just-hangs-up.9087/. Do you think this may improve things or am I wasting my time?

 ntpq -p
    remote          refid      st t when poll reach  delay  offset  jitter
==============================================================================
 
 
-212.70.148.19  94.26.7.48      3 u  308  512  377  62.638  -3.996  0.184
+96-7.cpe.smnt.p 213.199.225.30  3 u  510  512  377  35.903  -3.416  0.604
electra.pinklem 140.203.204.77  2 u  19d  64    0    1.712  -0.265  0.000
*spinnaker.darkt 193.67.79.202    2 u  34  512  377  11.932  -1.100  0.252
gatekeeper.csr. .INIT.          16 u    - 1024    0    0.000    0.000  0.000
chronos.csr.net .INIT.          16 u    - 1024    0    0.000    0.000  0.000
+mail.my-inbox.c 82.219.4.30      3 u  491  512  377  33.019  -3.593  1.448

 

[wardmundy]

papachumba FYI: I've replaced some of your images with Twitter copies. Image files consume enormous space. Please post links from other sites such as Twitter using Skitch or a similar product to conserve what little space we have. Thanks.

 

[papachumba]

Hi warmundy! I did use postimage.org for all images, this shouldnt have wasted any of your server space or bandwidth?

 

[wardmundy]

OK. Sorry. Wasn't paying close enough attention.

 

[atsak]

I've never had to adjust the timing in the current version of Asterisk - the older ones were another matter altogether. I think that's a dead end. It looks like you're losing a bunch of UDP packets all at once. This usually happens with congestion. Are you using that connection for anything else?