Marten Seemann

Another timeout, this time during `go mod download`: https://github.com/libp2p/go-libp2p/runs/7835153567?check_suite_focus=true

And another failure just now. This is starting to impact development velocity. I've disabled the workflow manually. We'll reenable once changes have been made that will fix / reduce flakiness.

Apparently this only applies to this branch: https://github.com/libp2p/go-libp2p/pull/1615

This test is so flaky that fixing it will need to be a P0.

Log output of a failed run: ``` === RUN TestMaxAge 2022-08-10T09:40:48.841+0200 DEBUG autorelay autorelay/relay_finder.go:653 starting relay finder 2022-08-10T09:40:48.841+0200 DEBUG autorelay autorelay/relay_finder.go:246 found node {"id": "12D3KooWAGEPewgrHmWnvmbt1GArsWGBXBiQo1mEu23tb2iJdZRd"} 2022-08-10T09:40:48.841+0200 DEBUG autorelay autorelay/relay_finder.go:246 found...

Funny, this PR is driving autorelay crazy. One more thing blocked on the refactor (#1587).

Fixed the flaky test by only enabling the TCP transport in tests. This is not very nice, but I don't any way to prevent us from establishing multiple connections.

We should fix https://github.com/libp2p/go-libp2p-peerstore/issues/203 and https://github.com/libp2p/go-libp2p-peerstore/issues/204 first. Tests are flaky.

https://github.com/libp2p/go-libp2p/runs/7745508053?check_suite_focus=true ``` === RUN TestHandshakeConnectionCancelations transport_test.go:47: using a ECDSA key: QmViCTRxC6Hr755XJf9AVoZ7aHrmLE9SYsRijJhqQScXth transport_test.go:47: using a Ed25519 key: 12D3KooWNGdBE6geq4W7E9VA3EtnmNZUFzASHVFQdfjcH1EpTcCn === RUN TestHandshakeConnectionCancelations/cancel_outgoing_connection transport_test.go:211: Error Trace: /Users/runner/work/go-libp2p/go-libp2p/p2p/security/tls/transport_test.go:211 Error: Target error should be in...

We probably need to improve static relay handling (also in #1587). Given that this PR is already reviewed and approved and just waiting for the v0.21 to ship before getting...