You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I think this issue is related to the ordering of operations and the test's simulated UDP and TUN devices using channels.
I can repro this deadlock with the following:
for i in {0..1000}
do
go test -v -tags=e2e_testing -count=1 ./e2e -run Test_Case1_Stage1Race -timeout 2s
if [ $? -ne 0 ]; then
break
fi
It'll reproduce within the 1000 runs, and get a stack trace similar to the above.
If I edit the udp_tester.go's channels to be unbuffered (line 51 and 52 of udp/udp_tester.go), I can repro the same stack trace immediately.
If I edit the udp_tester.go's channels to have a buffer size of 10, I can't repro the issue (or at least as easily.)
Reported stack trace on test run https://github.com/DefinedNet/nebula/runs/5554733633?check_suite_focus=true#step:8:133
The text was updated successfully, but these errors were encountered: