[Bug] test_connection_cleanups is flaky
vicsn opened this issue · comments
🐛 Bug Report
It appears that sometimes, when a validator is picked in the test, it fails:
thread 'test_connection_cleanups' panicked at node/router/tests/cleanups.rs:74:9:
the deadline has elapsed for condition: tcp0.num_connected() == 1 && tcp1.num_connected() == 1
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace
A quick print statement indicated: Unable to connect to '127.0.0.1:63168' - '127.0.0.1:63168' disconnected before sending "Message::ChallengeResponse"
Steps to Reproduce
cd node/router && cargo test test_connection_cleanups
Due to random sampling of the node type, the test only sometimes fails.
Expected Behavior
The test should pass.
Your Environment
Ran it on mainnet-staging 491f3078a
The fix is in #3164. Increasing the timeout a little ensures the test passes every time.
I may have spoken too soon, I'll investigate some more. It's less flaky with a longer timeout but it occasionally still fails 🤔