astf mode on linux interfaces not working

[dietmargoldbeck]

Hi trex Team,

for some measurements i would like to run linux on a notebook
The astf mode is showing very high packet loss and gives me asymmetric tcpdumps even when both ports
are directly connected via a copper cable.
I wrote this small shell script to compare astf mode to stf and i get 2 tcpdumps with the same number of dns packets.
In the astf case the tcpdumps are very different and approx half of the packets http responses are not in the trex port 0 tcpdump.

` #!/bin/bash

mkdir -p /root/pcap
timeout 60 tcpdump -s1500 -w /root/pcap/trex-tcp-0.cap -i enp0s31f6 &
timeout 60 tcpdump -s1500 -w /root/pcap/trex-tcp-1.cap -i enx00e04c2182e0 &
cd /usr/local/src/trex/v3.04
./t-rex-64 -d 45 --astf -f astf/http_simple.py -m 1 2>&1 | tee trex-http.log

sleep 20
timeout 60 tcpdump -s1500 -w /root/pcap/trex-udp-0.cap -i enp0s31f6 &
timeout 60 tcpdump -s1500 -w /root/pcap/trex-udp-1.cap -i enx00e04c2182e0 &
cd /usr/local/src/trex/v3.04
./t-rex-64 -d 45 -f cap2/dns.yaml -m 100 2>&1 | tee trex-dns.log
sleep 20
trex astf output shows packet loss Total-pkt-drop : 2091 pkts
Total-tx-bytes : 5218577 bytes
Total-tx-sw-bytes : 0 bytes
Total-rx-bytes : 4961907 byte

Total-tx-pkt : 5323 pkts
Total-rx-pkt : 3232 pkts
Total-sw-tx-pkt : 0 pkts
Total-sw-err : 0 pkts
Total ARP sent : 4 pkts
Total ARP received : 4 pkts
[...]
tcps_rexmttimeo | 0 | 63 | *retransmit timeouts
tcps_sndrexmitpack | 0 | 611 | *data packets retransmitted
tcps_sndrexmitbyte | 0 | 797132 | *data bytes retransmitted
tcps_rcvpartduppack | 41 | 0 | *packets with some duplicate data
tcps_rcvpartdupbyte | 64668 | 0 | *dup. bytes in part-dup. packets
tcps_rcvoopack | 365 | 0 | *out-of-order packets received
tcps_rcvoobyte | 756348 | 0 | *out-of-order bytes received
tcps_rcvdupack | 0 | 87 | *rcvd duplicate acks
tcps_predack | 0 | 664 | times hdr predict ok for acks
tcps_reasalloc | 56 | 0 | *allocate tcp reasembly ctx
tcps_reasfree | 56 | 0 | *free tcp reasembly ctx
tcps_reas_hist_4 | 56 | 0 | *count of max queue <= 4
tcps_sack_recovery_episode | 0 | 12 | *SACK recovery episodes
tcps_sack_rexmits | 0 | 85 | *SACK rexmit segments
tcps_sack_rexmit_bytes | 0 | 123080 | *SACK rexmit bytes
tcps_sack_rcv_blocks | 0 | 350 | *SACK blocks (options) received
tcps_sack_send_blocks | 350 | 0 | *SACK blocks (options) sent
- | --- | --- |
UDP | --- | --- |
- | --- | --- |
- | --- | --- |
Application | --- | --- |
- | --- | --- |
- | --- | --- |
Flow Table | --- | --- |
- | --- | --- |
err_len_err | 75 | 0 | *pkt with length error

`

any ideas?