Comment 8 for bug 1880389

thanks i will try and review the logs short.

in the interim i have set up a two node linuxbridge environment and tried to repoduce this with master.

using master so far i have not been able to messure more then a 4 second drop in connectivity

[1590502142.222506] 64 bytes from 172.24.4.63: icmp_seq=280 ttl=63 time=0.479 ms
[1590502142.430451] 64 bytes from 172.24.4.63: icmp_seq=281 ttl=63 time=0.423 ms
[1590502142.638626] 64 bytes from 172.24.4.63: icmp_seq=282 ttl=63 time=0.493 ms
[1590502142.846505] 64 bytes from 172.24.4.63: icmp_seq=283 ttl=63 time=0.447 ms
[1590502146.945127] 64 bytes from 172.24.4.63: icmp_seq=303 ttl=63 time=1.75 ms
[1590502147.148957] 64 bytes from 172.24.4.63: icmp_seq=304 ttl=63 time=0.813 ms
[1590502147.350779] 64 bytes from 172.24.4.63: icmp_seq=305 ttl=63 time=0.682 ms

while i understand this is a race at least on lightly loaded host this does not appear to happen on master.

its possibel i am just not hitting the race which is why race conditions are so annoying to repoduce. at least in my local testing the RARP packets appear to be recived but i have not done a trafic capture yet to see if all of them are recived or if some of them are dropped.