vyzo

I updated a small thing on the base branch (the pr is on top of another pr), that necessitated the rebase. It's a small thing, but it potentially saves allocations,...

Excellent!

this is weird. the only possible explanation is that the connection manager gets stuck, which is an issue @Stebalien has identified.

We are working on a fix with https://github.com/libp2p/go-libp2p-circuit/pull/76

see https://github.com/ipfs/go-ipfs/issues/6237 -- can you try building with go-ipfs master? It has the relevant patches applied.

This has to be accidental, the autonat service uses a background host object for dialbacks. Maybe it just happened for the connection manager to drop the connection?

It's possible that the extra connections to the autonat service cause the connection manager to go over its limits. @raulk any ideas?

I should point out that epidemic broadcast trees behave very poorly with multiple/random sources.

You should also check out our progress with gossipsub: https://github.com/libp2p/go-floodsub/pull/67 We want to have a mode for single/fixed source epidemic optimizations in the next iteration.

The intention is to build a mesh of a bounded degree; basically mold floodsub to something that doesn't blow up from amplification as you build larger overlays. Wrt performance, there...