Hi,
The 'X' in Pb is only a hint.
It will be put in the summary whenever there is anyone of the following
issues detected on the stream: Out of order sequence numbers (including
missing ones) and payload type changes. (Are there any other? I don't
remember). It doesn't even check about Max delta and jitter.
Concerning your sample:
Lost packets - Few - not a problem.
Max Delta - High! - May be a problem (indicates voice gaps) - unless you are
using voice activity detection /silence frames.
Max Jitter - Quite high! (as compared to Mean jitter) - May be a network
issue.
I suggest you drill down to in-depth "RTP-Analysis" for any streams sticking
out by any of the above values and then go through the packet list to see
any incoherencies.
Sorting the RTP-Analysis columns by "Delta" or "Jitter" (descending) may
give you a quick view at the max values.
Try to look for the packets with high "Delta" and explore the immediate time
neighbourhood of these packets.
You might see a gap followed by a sudden burst of packets (near zero Delta).
Then you have a network problem (packets are buffered somewhere).
Regards,
Lars
________________________________
From: wireshark-users-bounces@xxxxxxxxxxxxx
[mailto:wireshark-users-bounces@xxxxxxxxxxxxx] On Behalf Of gdonts
Sent: mardi 26 août 2008 15:35
To: wireshark-users@xxxxxxxxxxxxx
Subject: [Wireshark-users] Pb?, Max Delta, Max Jitter,Mean Jitter
colums in wireshark's RTP stream analysis
Hello,
I have a conference bridge (Avaya S6200 on SCO Unix) which I'm
getting intermittent reports of 'garbled' audio. The conf bridge uses G.711
u- or a-law, and incoming audio is provided by an ITSP (who have media
gateways (I believe Cisco) installed in a variety of locations that connect
to PSTN).
I've also got a PSTN connection into the bridge, where there are no
reports of poor audio, so I suspect something on the network (ie this is
only effecting calls that come in via SIP/VoIP).
I have the following connectivity:
Unix conference bridge <-> Cisco 2960 switch <-> Juniper SSG5
firewall <-> datacenter's BGP <-> public internet <-> ITSP's own POP <->
ITSP's own network <-> local media gateway in local carrier datacenter
I can't install Wireshark on the SCO Unix bridge (old 7.1.1 version
of OS), so I've got Wireshark running on a Windows server connected to a
Port mirror (or port span) so it's taking a copy of all the traffic going to
the Unix conference bridge. I also know that Wireshark can't know the state
of the various points on the network, so I'm only posting info on the
end-point (ie at the conf bridge)
I'm also open to any general suggestions that anyone has, but I know
this is a Wireshark maillist, so what I'm specifically trying to find out
about Wireshark (bless it!) is:
I've used the Statistics -> RTP -> Show all streams to get a view of
the various RTP streams of reported bad audio. I've a vague idea of the
principles behind calculating jitter, but when I see an 'X' in the 'Pb?'
column is this definitely a problem or just a suggestion that it might be an
issue? The following are the values for one sample call with reported bad
audio:
Column - forward value [reverse value]
Lost - 0 (0.0%) [47 (0.0%)]
Max Delta (ms) - 4119.77 [320.14]
Max Jitter (ms) - 1.80 [70.12]
Mean Jitter (ms) - 0.02 [0.30]
Pb? - blank [X]
The above is from a sample of about 60 minutes, so is a good
representation.
So which of the above values (if any) should I be worried about (ie
would affect audio quality)?
Any information/suggestions/comments appreciated.
Regards,
gdo