Using 2-step synchronization, an upstream master clock will send periodic Sync messages and corresponding FollowUp messages to the downstream slave. Sync and FollowUp message pairs have the same sequence ID value.
Sync/FollowUp message pairs should be received by the PTP slave node in sequence:
- Receive Sync (seq N)
- Receive FollowUp (seq N)
- Receive Sync (seq N+1)
- Receive FollowUp (seq N+1)
Under extreme congestion conditions, FollowUp messages might be delayed, lost or dropped in the Network.
The above message is generated by sfptpd
if it has received a
FollowUp (seq N) after it has already received a Sync with (seq N+1) and so it expecting
the corresponding FollowUp (seq N+1).
These rare events are indicative of extreme network congestion – which is delaying messages, the effects of which will probably be evident in other network applications.
Use tcpdump to capture a small pcap file which includes one or more messages which are reported out of sequence.
Send the following to support-nic@amd.com:
- The pcap file.
- The corresponding
sfptpd
stats_log file. - All files from the /var/lib/sfptpd directory.