PTP: monitoring the announce messages

Solved1.66K viewsPTP
3
0 Comments

It seems that from time to time my grandmaster device is sending the announce message with incorrect information: like the "-3" seconds in the correction time field instead of "+3" seconds as in the other 7 messages.

I can detect the issue if I'm looking into Wireshark capture (analyzing the packets), but it's not very convenient because of the number of packets that I need to check.

The question is: has anyone ever come across a similar "PTP" issue? And is it possible to detect such kinds of issues with the help of DataMiner? I have an idea that a new driver can be developed which will check announce messages by inspecting the Wireshark captures, but I'm not sure that it is a good way (especially if we have 8 announce messages per second, and we need to check them all and detect the issue close to real time)...

Marieke Goethals [SLC] [DevOps Catalyst] Selected answer as best 10th July 2023

1 Answer

2
1.21K 1 Comment

Hi Alex. We had users reporting similar issues, i.e. problems with individual PTP messages. Typically PTP boundary clocks or slave devices are not able to report such problem (however one bad PTP message could have a very bad impact on any PTP-aware device). DataMiner can interface with specialized PTP or network analyzers, those can be either online or offline analyzers. DataMiner can indeed trigger a wireshark capture to only record PTP messages. A .pcap file can then be analyzed offline directly in wireshark or for example in the EBU LIST tool. If you need something close to real-time, we can reach out to our tech partners, e.g. PacketStorm, Tektronix or Bridge Technologies if they can detect a single bad PTP message in real-time.

Marieke Goethals [SLC] [DevOps Catalyst] Selected answer as best 10th July 2023
Mahito Murakami [DevOps Advocate] commented

Hi Thomas,

Can you do the similar approach like FlowCorrector specifically to capture the PTP messages behavior directly by Dataminer, rather than relying on 3rd party analizer ?

One of our customer is facing PTP related issues, and struggling in finding the better way to diagnose them. Typically those issues might occur around BMCA failover period, but you can not capture such event by using IP analyzer kind of device, because you may not know at which end point device the issue occurs.

So I thought you might want to capture those PTP messages from all the network switches in the infrastructure directly to Dataminer.

I don’t know if those switch fabric is capable to transmit such “PTP” flow information to Dataminer, though.