And the webserver responds appropriately with the ECN flag set in the SYN-ACK packet. I think we should account for this, and accommodate it.
I tried simply removing those flags from the checks on lines 289 and 305 but that doesn't do the trick.
@NullHypothesis Would you be able to take a look into this? π
For the runs already captured, where this could have been an issue, we have the pcaps, plus since I logged the FourTuple, we know exactly which sets of packets SYN-ACK and ACK packets to extract this info from the pcap.
Found a π with the function that checks for SYN, SYN/ACK packets. (First reported in #4(comment))
In tcp_handshake.go, L289:L305 in isSynSegment() and isSynAckSegment(): I found the following legitimate case where the ECN and CWR flags are being set by the VPN, presumably to indicate the capability for handling Explicit Congestion Notifications:
And the webserver responds appropriately with the ECN flag set in the SYN-ACK packet. I think we should account for this, and accommodate it.
I tried simply removing those flags from the checks on lines 289 and 305 but that doesn't do the trick. @NullHypothesis Would you be able to take a look into this? π
For the runs already captured, where this could have been an issue, we have the pcaps, plus since I logged the FourTuple, we know exactly which sets of packets SYN-ACK and ACK packets to extract this info from the pcap.